You are on page 1of 82

Before the


In re: Food Labeling; Health Claims and )
Label Statements; Request for 1 Docket No. 91N-0098
Scientific Data and Information 1 (Fiber and Colorectal Cancer)


Julian M. Whitaker, M.D.; Pure Encapsulations, Inc; XCEL Medical Pharmacy,

Ltd.; the American Preventive Medical Association; and Durk Pearson and Sandy Shaw

(collectively the “Joint Commenters”), by counsel, hereby submit this supplement to their

comments filed on November 22, 1999.


Results of recently published scientific literature reviews confirm earlier results of

trials conducted by Macrae (1999)‘; Jansen (1999); Negri (1998); Caygill (1998); Le

Marchand (1997); Hill (1995, 1997), and the conclusions of the European Organization

for Cancer Prevention (ECP) (1998) which found that dietary fiber has a chemoprotective

effect on colorectal tissue, independent of the other components of plants. The American

Gastroenterological Association (AGA) recently published its findings and

recommendations based upon of an exhaustive literature review of hundreds of studies

involving thousands of subjects (Kim, 2000). That review examined experimental,
animal, observational, epidemiological, correlation, case-controlled, and intervention

studies. The AGA concluded that there is strong evidence supporting the protective

effects of dietary fiber among correlation and case-control studies conducted in

populations with different patterns of diet and colorectal cancer. The AGA noted that

three meta-analyses of case-control studies provide strong support for the dose-dependent

protective effects of dietary fiber or fiber-rich foods against colorectal carcinogenesis.

AGA found that those studies indicate that on average the subjects with the highest intake

of dietary fiber have a 50% lower risk of developing colorectal cancer than those with the

lowest intake. After weighing the evidence, including seemingly contradictory results,

the AGA concludes that increasing total fiber intake to more than 30g/day from the

standard 1Og/day North American diet can protect against colorectal cancer (Kim, 2000).

Hill (1999) reviewed scientific literature that examined the role of specific diet

components and colon carcinogenesis. The review concluded that there is consistent

evidence for the claimed association and a plausible mechanism for that association,

Giacosa and Hill (1999) published an explanation and restatement of the European

Organization for Cancer Prevention (ECP) consensus. The ECP found that based upon a

review of 58 studies of diet and colon cancer, 19 of which measured cereal fiber, a diet

high in cereal fiber is associated with a reduced risk of colorectal cancer. Reddy (1999)

found in his review that several lines of evidence are supportive of dietary fiber’s

chemopreventive properties. He states “animal model studies clearly suggest that wheat

bran consistently inhibits colon carcinogenesis. Case-control studies show reasonably

strong evidence that dietary fiber reduces the risk of colon cancer in humans. Dietary

’ Abbreviated references in this comment correspond with those presented in full in the bibliographical
listing. See Attachment 1.

intervention studies provide evidence that wheat bran supplementation decreases the

levels of several putative tumor promoters in the colon.”

Two recent studies having negative results have appeared in the scientific

literature. Both studies suffer from flawed controls and both dealt with aging populations

with pre-existing colon tumors. There is no plausible biological mechanism by which

fiber can be expected to correct the pre-existing genetic damage in people with pre-

existing colon tumors. The studies rely upon unscientific self-reporting as the measure

for compliance, a methodology generally criticized as yielding unreliable results

(Simone, 2000, Muller, 2000).

In the Polyp Prevention Trial (Schatzkin, 2000) adults over the age of 35 who had

one or more histologically confirmed colorectal adenomas were studied to determine if a

high fiber, low-fat diet would influence the rate of adenoma recurrence. 1,037 people

received 50 hours of nutritional counseling over four years and were instructed to

consume a daily diet consisting of 18 grams of fiber/day and 20 percent fat. The control

group of 1,042 were to make no changes to their diet. The subjects completed a four day

food record; the intervention group completed one every six months and all subjects

completed one each year. The data from the four day record was incorporated in the

study analyses as representative of the entire preceding interval (six months or one year

respectively). At the end of 4 years, the rate of adenoma recurrence did not differ

between the two groups. The authors concluded that “adopting a diet that is low in fat

and high in fiber, fruits, and vegetables does not influence the risk of recurrence of

colorectal adenomas.” Remarkably, the low fat, high fiber diet had no effect on weight or

cholesterol levels in either the intervention group or control group. A diet that is fat
restricted and has a high fiber content should have lowered weight and cholesterol and

yet in this study there was no difference in either weight or cholesterol levels between the

intervention group and control.

The most logical explanation is that the subjects were not actually eating the diet

that they reported. The alternative conclusion, that low-fat, high-fiber diets do not lower

weight or cholesterol, is unacceptable in light of many studies to the contrary.

The second trial studied 1303 people who had colorectal adenomas removed

within three months of entrance into the study (Alberts, 2000). Subjects were asked to

consume either a high-fiber supplement (13.5 grams/day) or a low-fiber supplement (2

grams/day). The subjects were studied over a 34-month period. At the end of that time,

there was no difference in the rate of adenoma recurrence between the two groups. The

authors concluded that a dietary supplement of wheat-bran fiber does not protect against

recurrent colorectal adenomas. Assurance of compliance was even more questionable in

this study than in the Schatzkin study. The investigators stated that “compliance with the

protocol was evaluated primarily by counts of returned cereal boxes and fiber bars at each

visit and secondarily through a specialized intake calendar.” The authors did not report

subject weight or cholesterol levels so objective corroboration of diet compliance is not


Dr. Charles B. Simone is a medical oncologist, radiation oncologist, and

immunologist who has studied the association between fiber and colorectal cancer for 22

years. Dr. Simone has reviewed the Schatzkin and Alberts studies (Attachment 3). He

concludes that neither study is valid because each lacks objective evidence confirming
compliance with the protocols. Other scientific critics have also found lack of proof of
compliance to invalidate the studies and, thus, make reliance on them improper. (Ornish,

2000; Davis, 2000; Gerber, 2000). Furthermore, scientists, including the study authors,

found that the “findings cannot be interpreted as evidence that a high-fiber cereal

supplement or a low-fat high-fiber diet is not effective in protecting against the later

stages of development of colorectal cancer” (Byers, 2000) and cannot be applied to

primary prevention risk reduction (Muller, 2000).

When the two aforementioned methodologically flawed studies on persons with

pre-exisiting cancer tumors are removed, the entire body of well-designed studies

published in the peer-reviewed literature overwhelmingly supports the conclusion that

“consumption of fiber may reduce the risk of colorectal cancer.” Accordingly FDA must

authorize the claim.

The Joint Commenters believe that the scientific evidence overwhelmingly

supports the association between fiber and colorectal cancer risk reduction and satisfies

the congressionally intended definition of significant scientific agreement. In addition,

and consistent with Pearson and the First Amendment, if the agency finds the proposed

claim supported by some evidence but not enough to satisfy a defined “significant

scientific agreement” standard, the claim must nevertheless be authorized with such

disclaimer or disclaimers as the agency reasonably deems necessary to avoid a potentially

misleading connotation. Only approval with appropriate disclaimer can ensure

compliance with the First Amendment.
Respectfully submitted,


By Counsel:

Emord & Associates, P.C.
1050 Seventeenth St., N.W., Suite 600
Washington, D.C. 20036
Phone: (202) 466-6937
Fax: (202) 466-4638
Date: September 25, 2000


1. Ir\Jo authors listed] Consensus statement on cereals, fibre and colorectal and breast
cancers. Proceedings of the European Cancer Prevention consensus meeting. Santa
Margheritia, Italy, 2-5 October 1997. Eur J Cancer Prev. 1998 May;7 Suppl2:Sl-83.

2. [No authors listed] Consensus meeting on cereals, fibre and colorectal and breast
cancers. ECP consensus panel on cereals and cancer. Eur J Cancer Prev. 1997
Dec;6(6):5 12-4.
3. [No authors listed] Global Review of Diet and Cancer Links Available (Review of
Food, Nutrition, and the Prevention of Cancer: A Global Perspective. American
Institute for Cancer Research, Washington, DC 1998. 670 pages) JAMA. 1997 Nov
26; 278(20) 1650.
4. [No authors listed] Position of the American Dietetic Association: Health
implications of dietary fiber. J Am Diet Assoc. 1997 Ott; 97( 10)1157-l 159.
5. [No authors listed] Primary Prevention of Colorectal Cancer and Polyps: Does Fiber
have a Role? Proceedings of a symposium. New York City, New York, USA.
December 2, 1997. Am J Med. 1999 Jan 25;106(1A):lS-51s.

6. Alberts DS, Martinez ME, Roe DJ, et al, and the Phoenix Colon Cancer Prevention
Physicians’ Network. Lack of effect of a high-fiber cereal supplement on the
recurrence of colorectal adenomas. N Eng JMed. 2000; 342: 1156-62.
7. Byers T. Diet, colorectal adenomas, and colorectal cancer. N Eng L Med. 2000 Apr
20; 342 (16); 1206-7.
8. Caygill CP, et al. Relationship between the intake of high-fibre foods and energy and
the risk of cancer of the large bowel and breast. Eur J Cancer Prev. 1998 May;7 Suppl
2:Sl l-7.
9. Chaplin MF. Bile acids, fibre and colon cancer: the story unfolds. J R Sot Health.
1998 Feb;l18(1):53-61.

10. Davis BM. High Fiber Diet and Colorectal Adenomas. NEJM, 2000; 343: 736.

11. Earnest DL, et al. Progress Report: The Arizona Phase III Study of the Effect of
Wheat Bran Fiber on Recurrence of Adenomatous Colon Polyps. Am J Med. 1999
Jan 25; 106(1A): 43S-45s.

12. Faivre J, et al. Primary prevention of colorectal cancer through fibre
supplementation. Eur J Cancer Prev. 1998 May;7 Suppl 2:S29-32.
13. Faivre J, et al. Chemoprevention of colorectal cancer. Recent Results Cancer Res.
1999;151:122-33. Review.
14. Franceschi S, et al. Italian study on colorectal cancer with emphasis on influence of

cereals. Eur J Cancer Prev. 1998 May;7 Suppl2: S 19-23.
15. Freeman HJ. Role of high fibre foods in the prevention of colorectal neoplasia. Can J
Gastroenterol. 1999 Jun;l3(5):379-80. Review.

16. Garay CA, et al. Chemoprevention of colorectal cancer: dietary and pharmacologic
approaches. Oncology (Huntingt). 1999 Jan; 13( 1):89-97; discussion 97- 100, 105.
17. Gerber M. High Fiber Diet and Colorectal Adenomas., NEJM, 2000; 343: 736.
18. Giacosa A, Hill MJ. Cereals, fibers, and cancer prevention. ECP Consensus Panel.
Adv Exp Med Biol. 1999; 472:269-72.
19. Hill MJ. Cereals, cereal fibre and colorectal cancer risk: a review of the
epidemiological literature. Eur J Cancer Prev. 1998 May;7 Suppl2:S5-10.

20. Hill MJ. Cereals, cereal Iibre and colorectal cancer risk: a review of the
epidemiological literature. Eur J Cancer Prev. 1997 Jun;6(3):2 19-25.
21. Hill MJ. Mechanisms of diet and colon carcinogenesis. Eur J Cancer Prevent. 1999
Dee; 9 Suppl 1: S95-8.
22. Jacobs LR. Fiber and colon cancer. Gastroenterol Clin North Am 1988
23. Jacobs LR. Effect of dietary fiber on colonic cell proliferation and its relationship to
colon carcinogenesis. Prev Med. 1987; 16:566-571.

24. Jacobs LR. Relationship between dietary fiber and cancer: metabolic, physiologic,
and cellular mechanisms. Proc Sot Exp Biol Med 1986 Dec;183(3):299-3 10.

25. Jansen MC, et al. Dietary fiber and plant foods in relation to colorectal cancer
mortality: the Seven Countries Study. Int J Cancer. 1999 Apr 12;81(2):174-9.
26. Kim YI. AGA technical review: impact of dietary fiber on colon cancer occurrence.
Gastroenterology. 2000 Jun; 118(6): 1235-57.
27. Kritchevsky D. Dietary fibre and cancer. Eur J Cancer Prev. 1997 Oct;6(5):435-41.

28. Kritchevsky D. Cereal fibres and colorectal cancer: a search for mechanisms. Eur J
Cancer Prev. 1998 May;7 Suppl2:S33-9.

29. Le Marchand L, et al. Dietary fiber and colorectal cancer risk. Epidemiology. 1997

30. Macrae F. Wheat bran fiber and development of adenomatous polyps: evidence from
randomized, controlled clinical trials. Am J Med. 1999 Jan 25;106( lA):38S-42s.

3 1. MullerRJ. High Fiber Diet and Colorectal Adenomas., NEJM, 2000; 343: 737.

32. Negri E, et al. Fiber intake and risk of colorectal cancer. Cancer Epidemiol
Biomarkers Prev. 1998 Aug;7(8):667-71,

33. Nelson RL, et al. Determination of factors responsible for the declining incidence of
colorectal cancer. Dis Colon Rectum. 1999 Jun;42(6):741-52.

34. Ornish D. High Fiber Diet and Colorectal Adenomas., NEJM, 2000; 343: 736.

35. Potter JD. Colon cancer--do the nutritional epidemiology, the gut physiology and the
molecular biology tell the same story ? JNutr 1993 Feb;123(2 Suppl):418-23.

36. Potter JD , McMichael AJ. Diet and cancer of the colon and rectum: a case-control
study. J Nat1 Cancer Inst 1986 Apr;76(4):557-69.
37. Reddy BS. Preventionof colon Carcinogenesis by Components of Dietary Fiber,
Anticancer res. 1999; 19: 3681-3.
38. Santini DL, Rigas B, Shiff SJ. Role of diet and NSAIDs in the chemoprevention of
colorectal cancer. J Assoc Acad Minor Phys. 1999; lO(3): 68-76.
39. Schatzkin A, Lanza E, Corle D, et al, and the Polyp Prevention Trial Study Group.
Lack of effect of a low-fat, high-fiber diet on the recurrence of colorectal adenomas.
N Eng JMed. 2000; 342: 1149-55.

40. Scheppach W, et al. WHO consensus statement on the role of nutrition in colorectal
cancer. Eur J Cancer Prev. 1999 Feb;8( 1):57-62. Review.

4 1. Shike M. Diet and lifestyle in the prevention of colorectal cancer: an overview. Am J
Med. 1999 Jan 25;1’06(1A):l lS-15s; discussion 5OS-51s.

42. Simone CB, Simone NL, CB Simone II. Consumption of fiber reduces the risk of
colorectal cancer: A Review. International Jhtegrative Med. July-August 2000.

43. Slavin JL. Implementation of Dietary Modifications. Am J Med. 1999 Jan 25;
106( 1A): 46S-49s.




Attilio Giacosa’ and Michael J. Hill’
on behalf of the ECP Consensus Panel* (see end of chapter)

‘National Cancer Institute, 16132 Genova, Italy
* ECP (UK) Headquarters, Wexham Park Hospital, Slough,
Berks SL2 4HL, United Kingdom

All plant foods contain plant cell walls which contain dietary fibre and a range of
other anticarcinogenic agents including vitamins, antioxidants, tannins, polyphenolics,
and flavonoids. In general, vegetables contain relatively modest amounts of dietary
fibre but are rich in a wide array of anticarcinogens, the amounts and classes of which
vary between vegetable type. Cereals are relatively rich in dietary fibre and also contain
phytate and a range of anticarcinogens. However, these latter are partly removed with
the husk during milling. Fruit contains the least dietary fibre but contains an array of
anticarcinogens which differ from those in cereals and vegetables.
Current hypotheses suggest that fruit and vegetables provide protection against
cancer mainly through the action of their anticarcinogens. In contrast,cereals have been
assumed in the past to act mainly through the action of dietary fibre.
In this Consensus Statement “cereals fibre” will imply unrefined or high-
extraction cereal, with its husk (and the accompanying anticarcinogens) largely intact.
In Europe, cereals may be consumed as breakfast cereals which are often rich in dietary
fibre and also rich in B vitamins and anticarcinogens. At other time of day, cereals are
usually eaten as breads, pasta, rice, pastries, etc. These are usually made from low-
extraction cereals which contain lower levels of dietary fibre and anticarcinogens;
wholemeal breads and products are richer in both, however.
Different cereals contain different amounts of dietary fibre and anticarcinogens
(rice has least and wheat and rye have most of both). Further, rice is almost always
eaten in polished and refined form and so contains even less dietary fibre and anticar-
cinogen than usual. The cereals which are most often consumed in unrefined and high-
extraction form are wheat and rye.
The postulated mechanisms of action indicate that the protective action will be
greater in the unrefined cereal than in that in which the husk has been removed by
Ari~wzces in Nurrifion nruf Cancer 2, edited by Zappia er al.
Klu\ver Academic / Plenum Publishers, New York, 1999. 169
Ccrcsk, Flhcr, nnd Canrcr Preventinn 271 . I

milling. In mosl epidemiological sluclics Ihc ccrcals are primarily low-estfaclion prod- RREAST CANCER
ucts and so arc low in dietary fihrc and anticarconngens. A major recommendation was
that in future, questionnnircs should he framed to distinguish between low-cxtrnction There is suggestive evidence that cereal libre provides protection against breast
and high-extraction cereals. cancer. Although many epidemiolo~icnl studies have shown that cereal Bbre has a pro-
On 1997 the European Organization for Cancer Prevention (ECP) held in S. tective effect, others have shown no effect and there is insufficient evidence to reach a
Mnrgherita Ligurc (Italy) a Consensus Meeting on the role ofcercals, fihre, and cancer delinitive conclusion. In Stuttgart, the WI-IO Consensus Group on Breast Cancer con-
prevention. The ECP Panel (reported at the cncl of this paper) nchievcd a consensus cluded that the epidemiological evidence was suggestive of a protective effect (as did
statement that is reported in the following pages.’ we) and recommended that cereal fibre consumption should be increased.
It is generally accepted that high levels of circulating oestrogens and insulin
growth factors represent major risks for the development of breast cancer. Diets low
in fat and rich in cereal tibre reduce levels of plasma oestrogens. in particular by inter-
COLORECTAL CANCER fering with their enterohepatic circulation and so increasing the rate of faecal excre-
tion. Such diets also contain phytoestrogens. which have been proposed to he
A diet rich in high-fihrc ccrcal is associnfctl with a rcduccd risk of colorcctnl protcctivc. Rose et al.’ and Woods et al.’ have shown that diets low in fat and high in
cancer. In support of this wc cite the review of 5X previous studies ol diet and colon wheat bran fihre significantly reduce plasma levels ofoestrndiol and oestradiol suphnte.
cancer. in only I9 of which cercnl fibrc was mensurcd. Of these, I6 reported an inverse Ebrc intakes have also been shown to be inversely related to total, subcutaneous and
association bctwccn ccrcnl fihrc and colon cnnccr risk and the other three showed no extra-abdominal fat and to lower insulin level.s.‘Ihcse findings reflect the influence 0r
relationship 2, 3 (l-till, 1997. 1998). In addition, the review of Food ilrltl Ag-iculturc libre in controlling aspects of the insulin-resistance syndrome.
Organization data by Caygill et al. ’ showed that there is an inverse relationship
between the risk of colorcctnl and of hrcnst cancer and cereal and vegctahle disap-
pcsrance. no relationship with fruit and starchy root intake and a positive correlation OTHER SITES
with total cncrgy inlake.
Thcsc data are consistrnt with thnsc from the Italian study: in the context nf lhc There is good reason to examine seriously the relationship between cereal fihre
Italian diet, high consumption of rctined ccrcnl was shown to be a major contributor intake and cancer at other sites. The preliminary analyses reported by La Vecchin and
to high total cncrgy intake and was a risk factor for cancers of the colon and breast. Chatenoud”’ suggested that people who reported consuming whole grain cereals had
This suggested that the rcnl association was with total cncrgy intake. a lower risk of cancer at a range of other sites in addition to the large bowel and breast.
T?~is consensus rcn!Y%n~s and extends the consensus reached by the Colon Group There were many potential confounding factors in these Italian data, and they need to
at the World Health Orgnni7ation (WI-IO) Consensus Conference in 1996,” and with be confirmed. However, there are good theoretical reasons for suspecting a general pro-
the Committcce on Medical hspcctq of Food Policy (COMA) recommendaticlns in the tective effect. If the mechanisms proposed to explain the protective effects against
United Kingdom. hrcast cancer are true. !hen WC would cspect them to apply also to other hormone-
A variety of mechanisms has been proposed for the protective d-f-cd 0r related cancer sites such as cndometrium, ovary, and prostate. Carcinogen binding in
cereal fibrc. l3urkitt’ popularisccl the idea tl~at a diet high in fibrc-rich foods could the colon lumen might also give rise to a generaliced protection, and the presence of
influence the COLIIW ofcolorectnl cnrcinogcncsis: f-lc proposed that it was fermentation anticarcinogens in the cereal husk would provide a mechanism similar to that proposed
ol the libre itself that gave the protection through (1) increased raecal weight: for vegetables and fruit, If such a generalised protection were to he confirmed if would,
(2) incrcascd frcqucncy of dcfccation: (3) dccrcascd transit time: and (4) dilution of course, strengthen the recommendation to increase intakes of high-fibre cereals.
of the colonic contents. The cvidcncc is strongest for (1) and (4) being important,
although there is cvidcnce against all four mcchnnisms. In addition hc proposed that
iibrc metabolism influenced microbial growth in the colon, an area we know very GENERAL RECOMMENDATIONS
littlc about.
More rcccntly, mcchnnisms involving the metabolic consequences of fibre mc- l Questionnaires need to hc directed in future to the study of food groups (e.g.
tnholism hnvc been proposed including (S) alterafion of energy metabolism. It is now cereals) rather than nufricnfs or anutrients (e.g. dietary fihre), since the latter
generally accepted that cncrgy restriction will inhibit cnrcinogcncsis illld a libre-rich are highly heterogeneous and not nccessnrily well quantitnted.
diet may make a contribution to ovcrnll cncrgy manngcmcnt; (6) influence bile acid l In view of Ihc clntn prcscntcd in the review by I Iill,’ mcln-analyses of the cnse-
nict;ll~c~li~rn, :I Il~c~>rv tl1:11 ;~ptx~‘:irs Ir> rcfllsc IO I:” :tw;tv; (7) protltlc*licm crf <horl-chain control and lhc cohort sludicc \houltl bc c:lrricd 001.
falty acids. which may inhibil cilrcillOg:cn~sis llirough ils Cffccls 011 colonic pl I, allcl l M:lny ol ~hc cffccts ol’ dietary fibrc that provide prcltcction against colorectal
through the supply of Iwtymtc. This hftcr hs been slmvn in vitro to promofc apop- and brcnst cancer arc conccmcd with events in the cnccum and proximal colon.
tosis, and ccl1 diffcrcnti:ltion, bolh of which arc CClllrill to the carcinogcncris pr0Ccss. WC need lo unrlcrstand much more illXJtlt Ihc ecology of Ihis imporlant but
In vivo verification of these actions is still awaited. experimentally inaccessible subsite of the large bowel.
212 A. Giacnca and M. .I. llitl
24 ’

2. Ilill M.J. (1997). Cerc:~ls. crrc:~l lihre. and colwcc~:d CI~CCI rick: a review of the rpidemiological
lirerncurc. Eur J Cnnccr I’rev 62 1%25.
3. Hill M.J. (199X). Compcrsition nnd control of ilcnl contcnfs. Eur J Cancer Prcv 7 (soppl 2)575-S7X.
4. Ceygiil C.P.J.. Chnrlctl A.. and Hill M.J. (199X). Relationship hetwcen Ihe intake of high-fihre foods CARNITINE SYSTEM AND TUMOR
and energy nnd lhe risk of cancer of the large howl and breast. Eur J Cancer Prev 7 (supp’l2):Sl I-St 7.
5. Franccschi S.. Rwero A.. Pnrpinel M.. Cincosa A..and I.3 Vecchia C. (1998). Italian study on coloreckd
cancer with emphwis on the inllucnce of cerealc Eur J Cancer Prev 7 (suppi 2):S19-S23.
h. Riesalski I1.K. ;md Fiirst I! (1997): WI0 cwwnsus confcrcncc on diet and mnccr: 2%30 Novcmhcr
1996. Eur J Cancer I’rev 631.5. Menotti Calvnni.’ Raffaela Nicolai,’ Alfonso Barhnrisi: Emilia Reda,’
7. Rurkitt IX! (1971). Some ncglcclcd knds IO cnnccr c:mmlion. J Nntl Cancer Ins1 47:913-916. Paola Bertatti.’ and Gianfranco Pclus$
R. Rose D.I?. Goldmnn M.. Connelly J.M.. and Slrong LT. (19%‘). Ifigh ftbcr diet reduces %xum estro-
gen concenfretion~ in premcnopnos:~l women. Am J Clin Nutr 54520-525.
9. Woodr M.N.. Gorhach II... Lonpzopc C.. Goldin RR.. Dwyer J.T., and Merrill-LaRrodc A. (1989).
‘Scientific Department. Sigma Tau !$A., Via Pontina Km 30.400.
Low-fnc. high-fiber dicr and strum estrcme sullnte in premenopausal women. Am J Clin Nu~r Pomezia, Rome, Italy
49:1179-llR3. ‘Institute of Clinical Surgery, Faculty of Medicine, 2 University of
10. La Vecchin C. and Chntenoud L. (1998). Flhres, whole-grain foods and breast and other cancer.% Eur Naples, Italy
J Czmccr Prev 7 (snppl?):S25X?X.
‘CNR, Via Toiano 6, Arco Felice (Naples), 2 University of Naples, Italy

Dr. Michael Hill (Chairman of ECP) Slough. IJK
Dr. AtCilio C&cow (Scientific Coonlinnfor of ECP) Genoa. Italy
Dr. David ReckIcy l’lymouth. UK
Dr. Christine PJ C’;#yp.ill Slough. UK
Dr. Paula Chaws Lisbon. Portqnl Carnitine, a name derived from the Latin cnnrk (flesh), was isolated from meat
Dr. David Evans London. UK extracts in 1905’ and early its chemical formula (C,l-I,5NOJ was proposed. Its struc-
Prof. Jean Rivre Dijon. France
ture, a trimcthylhetaine of y-amino-fi-hydroxyhutyric acid, was correctly identified and
Dr. Rhio Fwinnti radun. Ililly
Prof. Silvia Frrmceschi Aviano. Ilnly
published about twenty years later.’ Initially, some circumstances Icd to consider car-
Dr. Miqucl Gassull Rndnlonn. Spain nirinc as a vitamin. By about 194.5, all of ~hc important vitamins of the I3 group had
Dr. Marierle Gerhcr Monlpclicr. Fr:mce hccn identified, but the interest in the discovery of still missing B-vitamins. their lack
Dr. Inn T Johnwn Norwich. I JK being possibly correlated with anemia, was tremendous. In those years Fraenkel and
Dr. David Kritchevskv I’hiladclphin. 1lSA
coworkers observed that the mealworm 72~chrio rnolitnr required for normal growth
rror. Cwlo I;1 Vecchi:l Milan, Ilaly
Prof. Pnul Maingwl Rrwselc. Rel~ium
and survival, in addition to at Icnst eight ol the known B-vitamins, also folic acid and
Dr. Alnin Maskena Rrussels. Dclgium a new factor contained in brewers yeast 01’ in liver cxtract.which they tentatively named
Dr. Roherl W Own Heidclburg. Germany vitamin-n, (T for Tcnehrio).’ The unlavorahle properties of this factor (it was hygro-
Dr. Joseph Rafter Stockholm. Sweden scopic and extremely water soluble, thus, hard to crystallize) made its isolation dilficulf
Prof. Inn N Rolwlnncl Irish LJnitcd Nulr. AWW.
hut. finally. the missing vitamin-Rr was identified as carnitine.4 The widespread dislri-
Dr. David Somhp,nlc Norwich, UK
Maaslrichl,‘flre Netherlands
hution of carnitine was established in microorgmisms, lower animals, and in all organs
Prof. Reinhold Smcckhrup,gcr
of mammals, and in plants too.’
nut soon al’tcr. the finding that microorganisms as well as higher animals were
also able to synthesize carnitinc by themselves, came to light.M Hence, the assumption
upon which cnrnitine was included among vitamins failed.
The physiological role of carnitine in microorganisms has not been elucidated
’ for ;I long time. To date it is known thaw the role of cnrnitine in growth stimulation
and mctaholism in microorganisms varies depending on species and living conditions.
For cxamplc in f3cherichirr co/i, cnrnitinc and other quaternary compounds. such
Advnnrrs it1 Nutrition nnd Cancer 2. edited hy Zappin rl nl.
Kluwcr hcndcmic / Plenum Puhli&cfi. New York. 1999. 273
of Cobferfal
Cancer 123

Chemoprevention of Colorectal Cancer be a potential target for secondary prevention 3s well ns for primary preven-
tion. Several arguments lend credence to the notion that the adenoma-carci-
noma sequence is a multistep progress. Cancer can be prevented by interven-
J. Fnivre and C. Eonithon-Kopp tion either at the stage of ndenomn growth or nt that of transformation into
Registre bourguignon tics Cancers Digestifs (INSERM CR19505). Many case-control studies, 2nd some cohort studies, have provided sub
Fncult~ de Mfdecine, 7 Boulevard Jeanned’Arc, F-21033Dijon Cedex. France stnntial epidemiological evidence for the overwhelming role of diet in the OC-
currence of the disease (Potter et nl. 1993). There is fairly consistent evidence
concerning the effect of vegetables ns n protective factor nnd of caloric intake
3s n risk factor. There is some evidence relating fnt intake or protein intake
to colorectal cancer, whereas fibre intake, cnlcium intake and antioxidant vi-
tamins may be inversely related to colorectal cancer. Ilowever, analyticnl
studies have yielded equivocal findings. The dntn nvailnble sre not sufficient
Abstract to serve as a basis for firm specific dietary advice, but they provide attmc-
tive hypotheses, which in turn suggest ;I rntionnl bnsis for B preventive
Epidemiological studies hnvc cmphnsised the major role of diet in the aetiol- appronch. Faced with this situntion, it is important to test these hypotheses
ogy of large bowel cnnccr. Attempts to identify cnusntive or protective factors within the fmmework of intervention studies in order to evaluate the possibi-
in epidcmiologicnl nnd cxpcrimentnl studies hnvc Icd to some discrepzmcics. lities of primary prevention. The objective of this report is to review the de-
The time 11~scome to test the most importnnt hypotheses within the fmme- sign. along with the nvailable results, of mndomized colorectal cancer che-
work of intervention studies. Among studies spccificnlly devoted to colorectnl moprevention trials. Only studies with cnncer or precnncerous lesions (i.e.
carcinogenesis, eight have been completed nntl five nre ongoing. They evnlu- adenomas) BS the main end-points nre included here. Studies evaluating the
nte the effect of the intervention on ndenomn recurrence and, in three stud- effect of drugs are not considered.
ies, on adenomn growth. Five intervention trials considering cardiovasculnr
disenscs and different c;\nccr sites will provide dntn on the effect of the inter-
vention on colorect;tl cancer incidence. Vitamins nnd antioxidants, fibre or
cnlcium sufplementntion, aspirin therflpy nnd dietary modificntions arc evol- Vitamins and Antioxidant Trials
uated. Most of the zwnilnble dnta do not support the iden of a protective ef-
fect of vitamins nntl nntiaxitlnnts ngninst colorectnl carcinogencsis. It is to0 In recent years, much nttcntion has been paid to lhe potentinl advantnges of
early to dmw nny conclusions on the effects of fihre, calcium supplementn- antioxidant vitamins, including /I-carotene, retinoids, vitamin C, and vitamin
tion, aspirin therapy and c!iet,ry intcrvrntion. The results of ongoing studies E, and of other micoronutrients, such ns selenium, as chemopreventive
will be nvnilable within 2 ycnrs. If one of the evnlunted interventions proves agents for large bowel cancer. The main features of these trials are sum-
efficient, the benefits of n simple, snfc nnd inexpensive prophylaxis for ;1very marised in Table I. Among the 15 chemopreventive studies with colarectal
common cancer will bc clcnr. carcinogenesis as nn end-point, I I are at lenst partly concerned with the pos-
sible preventive effect of vitamins and/or nntioxidnnts. The populntion in-
volved is represented BS follows: in 6 studies subjects who had previously
hnd ndenomn and who were polyp free nt the time of recruitment; in 2 stud-
I Introduction ies, individuals with familinl adenomntous polyposis previously treated by to-
tal colectomy and ileorectnl annstomosis (I~nssey et al. 1982; De Cosse et ni.
I The most recent estim;ltes of the worldwide incidence of colorectnl cancer 1989); nnd in 3 studies, volunteers included in Inrge trials assessing the ef-
rnnk it third nmony, the most frequent cancers, with nbout 560700 new cases fects of micronutrients supplementntion on cancer sites snd cnrdiovasculnr
per yenr (Parkin ct nl. 199.1).It is n major public hcnlth problem in nil devel- tlisenses (ATUC Study Group 1934; Physicinn’s Hcnlth Study cited in Flenne-
optl corlntrirs ill Wrslc~-n I:r~rclpc, Norll\ Amrricn nnd the South Scn Islnnds. kens et nl. 1996; Ilcrcbrrg C! 611.1993). Thr rffrct of vitamin C Aone wns
Dcspitc ;ulvnnccs in tli;lgnostic trchniqurs nnci trcntment, the S-year survivnl tested in 1 study (Ibsscy et al. 1982), the effect of /i-carotene alone in 4 stud-
rntcs remain poor nnd are estimnted to be 30% in Europe (Rerrino et ~1. ies (Greenberg et al. 1994; MacLennnn et nl. 1991; ATIX Study 1994; I-lenne-
1995). Thcrc is littlc improvement with time. Strong evidence indicates that n kens et al. 1996) snd the effect of vitamin E nlone in 1 study (ATBC Study
high proportion of colorectnl cancers arise in adenomas. These lesions could 1994). Vitamin C rend vitamin E were evalunted in 2 studies (De Cosse et al.

RecentResults in Cancer Research.
Vol. 151
SznnlCostnljnr~lnn(Eds.):Chemoyrevenlionof Cnncrr
(1 Sl~rilll:“-\Crl;rl: llrrlin llcitlcll~rrp, 1999
114 1, Faivreand C. Honithon-Kopp Chemoprevenrionof ColorectalCancer 12s

1989; Mc Keown-Eyssen et al. 1988), and various combinations of vitamins Table 1. Study designs,
end-points and results of chemoprevention trials of vitamins and antioxidants
and of nntioxidnnts in 4 studies (Roncucci et al. 1993; Hofstad et al. 1995; in colorectal
Bonelli et al. 1994; llercbcrg et al. 1393). All these studies were double-blind Study Subjects Intervention No. of Duration End-point results
randomized trials except for the Modenn study (Roncucci et al. 1993), in with subjects
which the reference gronp had no treatment. All the studies except 2 had a Bussey et al. Familial Vitamin C 2 g/day 49 15-24 No significant
parallel design, meaning that the effect of one or several treatments wx 1982, London polyposis months reduction in the
compared with the effect of the placebo. A 2x2 factorial design was used in 2 number of rectal
studies (Greenberg et al. 1994; MncLennan et al. 1991). The advantage of this adenomar
design was that it allowed an estimation of the effect of the two combined DeCorseet al. Familial Vratmin C 4 Igdlday + 58 4 yeafs Noeffecton the
1989,NewYorkpofyposir vitamin E 400 mg/day number of rectal
treatments and that it gave more power to the study than a parallel scheme adenomas
with the same number of patients. McKeown- Previous Vitamin C 400 mg/day + 185 2 years No effect on adeno-
The main end-point was ndcnomn recurrence in 5 studies (McKeown-Eys- Eyssenet al. adenoma vitamin E 400 mg/day ma recufrence
sen et al. 1988; Roncucci et al. 1993; Greenberg et nl. 1994; MacLennan et al. 1988,Toronto
Roncucci et al. Previous Vitamin A 30000 W/day 255 3 years Significant reduc-
1995; nonelli et al. 1994), variation in size of ndenomas left in situ in 3 stud- 1993, Modena adenoma c vitamin E 70 mg /day tion in adenoma
ies (Bussey et al. 1982; De Cnssc ct al. 1389; Hofstad et al. 1992), and colo- recurrence
rectal cancer incidence in 3 studies (ATW Study 1994; Hennekens et al. Greenberg et Previous D-Carotene 30 mg/day + 864 4 years No effect on adeno-
1996; Hercberg et al, 1993). Most trials aimed at evaluating the effect of sup- al. 1994, USA adenoma vitamin C 1 g/day + ma recurrence
plementation on adenomn recurrence or adenoma growth were small. The vitamin E 400 mglday
Hofstad et al. Previous /I-Carotene15 mglday + 116 3 years No effect on adeno-
only large study was the one carried out within the National Polyp Study in 1995, Oslo adenoma vitaminE 75 mg/day + ma growth or ade-
the USA (Greenberg et al. 1994). Trials using adenoma recurrence or adeno- vitamin C 150 mg/day + nomarecurrence
ma growth as the primary outcome have the advantage of being relatively selenium 101 mg/day
small in size because a large number of events are expected during follow- Maclennan et Previous /&Carotene 20 mglday 378 4 years No effect on adeno-
al. 1996, adenoma ma recurrence
up. For instance, the rntc of patients with new adenomas is expected to be Australia
30% at 3 ycnrs. However, whereas a relatively small sample size is suflicient Bonelli et al. Previous /J-Carotene 15 mg/day + 279 5 years Adenoma
to give the power needed to test the effectiveness of the intervention, some 1994, Genova adenoma vitamin E 75 mg/day + recurrence
studies are obviously too small to provide any firm conclusion. In contrast, vitamin C 150 mglday +
-~,--1..- l n. --,J-..
setenam IUI mg/ady
studies with invasive cancers as the main end-point require several tens of ATBC
1994, Male smokers /!-Carotene 20 mg/day t 29 133 4-13 years No effect on tolo-
thousands of subjects. Finland SO-69years vitamin E 50 mg/day rectal cancerinci-
The duration of the studies varies according to the main end-point. Trials dence
that use ndenomn recurrence or ndenoma growth as the main end-point have Hennekens et Medical /I-Carotene 50 mg on 22000 5 years No effecton cancer
the advantage of being relatively short in duration, ranging from 2 to 5 years al. 1996,USA doctors alternate days incidence
SIJVIMAX, Volunteers ’ /I-Carotene 6OCOmg/day 15000 8 years Cancer incidence,
(Table 1). Studies with colarectal cancer as the primary outcome require a France, 1993 I ’ -r- vitamin C 120 mglday cardiovascular dir-
longer follow-up period. generally at least IO years. + vitamin E 15 mg/day f eases
The degree of complinncc with the supplements is of importance. It was selenium 101 mglday +
between 70% and 85% in most studies: 73% (Bussey et al. 1982), 79% (De zinc 20 mg/day
Cosse et al. 1989), 75% (McKcown-Eyssen et al. 1988)‘ 86% (Greenberg et al.
1994), 81% (Hofstnd et al. 1995). It was only 45% in 1 study (Roncucci et 41.
1993). Compliance with the linnl cndoscopy was 73% in the St. Mark’s Study, group (the reduction was significnnt at the 9 month follow-up, but disap-
79% in the New York Study, 78% in the Toronto Study, 87% in the National peared over the next follow-up periods). A study with a similar design was
Polyp Study. 72% in the Australian Study, 87% in the Oslo Study and only performed in New York (De Cosse et al. 1989). There was no effect of vita-
26% in the Modena Study. min !Z and vitamin C on the number of adenomas.
The first chemoprcvcntivc study conccrnine colorectnl cancer cnrcinogrn- Of the 5 published studies that have tcstrd the effect of antioxidant vitn-
csis w:\s pcrforn~cd :I( SI. M;lrk’s I lospitnl. I.ontlon, on patients treated for niins on ntlenomn recurrence, 4 ore negntive nnd I is still on-going (nonelli
polyposis coli with the rectum left in place (Bussey et al. 1982). In the trcnt- et al. 1994). A Canadian study found no effect of supplemental vitamins C
ment group, there wns n non-significant trend to ;I reduction in the number and E on the rate of recurrence of adenomas over a 2-year period
of rectal adenomas and in the ndenomn area compared with the control (McKeown-Eyssen et al. 1988). In an American strrdy there WRS no evidence
12G I. laivrcand C. Bonithon-Kopp Chemoprevention Cancer 127
of Colorecral
that /&nrotcnc or vitamin C and vitamin E rcducccl the risk of new adcno- A lot of information is nvnilnble on the effect of antioxidant vitamins on
mas (Greenberg et al. 1994). Neither diet treatment appeared to be effective colorectal cancer carcinogenesis. This information allows the conclusion that
in any of the subgroups studied delined according to sex, age, number of antioxidant vitamins and micronutrients have no effect on adenoma recur-
previous adenomas and serum level at entry or subtypes of adenoma identi- rence, adenoma growth or colorectal cancer risk.
fied at follow-up examinations (number of colorectal adenomas, size of the
largest adenoma and location of the adenomas). In the Oslo study, no effect
of a combination of p-carotene, vitamin E, vitamin C, selenium and calcium Fibre Trials
was found on the adenomn growth of an adenoma <l cm in size that had
not been extirpated (I-Iofstacl et al. 1995). Moreover, there was no effect The results of analytical studies on dietary fihre are rather contradictory. It
either from year to year or when the size of the left-in adenoma and/or the must be emphasised that dietary fibre is not a homogeneous entity and that
location of the ndcnomn, gender and cancer among first-degree relatives different components may have different physiolo@al effects. Food composi-
were taken into account. In the Australian study there was the suggestion of tion tables lack data on the different types of dietary fibre. In this context,
nn adverse effect: the recurrence rate of large ncienomns (>l cm) increased studies examining the effect of a single source of fibre on experimental carci-
(borderline significance) in the group rrcciving /?-carotene supplementation. nogenesis in rodents are of interest. Pectin, cellulose, liEnin, guqum, alfalfa,
In contrast, a trial in Modenn showed a significant reduction in the adenoma carrageen and cutin seem to have little effect (Fnivre et al. 1991). However, a
recurrence rate in patients receiving vitamins A, C and E compared with protective effect has been observed in most studies for wheat bran and muci-
non-treated patients (Roncucci et al. 1993). The numbers of patients with a laginous substances (such as ispnghuln husk), particularly during the pro-
new adenoma at colonoscopy were 4 of 49 treated and 28 of 54 untrentet! pa- moting phase. The relevance of these data to human cancer must be evnlu-
tients. The main limitations of this study were the small number of patients Met! in intervention studies.
(resulting in a lack of precision in efficacy estimates), the short follow-up Fibre supplementation is proposetl in forlr cllrmopreventive stridies (Table
period (only a quarter of the subjects had a colonoscopy after 2 years) and 2). The efrect of wheat brnn ( 22.5 g/d;~y) tol:ethcr with vitamins C and E has
the fact that a substnntial proportion of randomly assigned pnticnts did not been evaluated in patients with familial polyposis arid with the rectum left in
undergo a follow-up colonoscopy at all. Because of these limitations, the re- place (De Cosse et a!. 1989). Its effect on adenoma recurrence was studied in
sults of this study need to be regarded with caution. the Australian study, with a dose of 25 g/day (McLennnn et al. 1995), and in
Some results are also available from the large trials that have colorectnl the Arizona study, with 13.5 g/day (Vargns and Alberts 1992). A multicenter
cancer incidence as an end-point. The Alpha Tocopherol, Beta Carotene, European study performed within the European Cancer Prevention Organisa-
Lung Cancer Prevention Study in Finland was logistically a success (ATBC tion (ECP) has been assessing a mucilaginous substance in the form of ispa-
Study 1994). A total of 29 133 male smokers aged 50-69 years participatet! in
the chemoprevention trial, accumulating 169751 follow-up years. During the
course of the study, 68 incident cases of colorectnl cancer appeared in the n- Table2. Studydesigns,end-points
andresultsof chemoprevention
trialsof fibrein colorwtalcancer
tocopherol group versus 81 in the croup not receiving a-tocopherol, and 76 carcinogenesis
in the /I-carotene group versus 73 in the group and receiving p-carotene. In
the United States, 22071 mnlc physicians aged 40-84 years were randomized Study :; .Subjectswith Intervention No.of Duration End-pointresults
in a double-blind placebo-controlled trial of /I-carotene, 50 rng on alternate
days. Fewer than 1% had been lost to follow-up and compliance was 78% in De Cosre et al., Familial Wheat bran 58 4 years Nonsignificant reduction
the group that received /I-carotene. Overall, 167 colorectal cancers were ding- 1989, New York polyposis 22.5 @day + in the numberof rectal
vitamin C 4 g/ adenomas
nosed in the intervention group and 174 in the placebo group (Hennekens et ,.
day + vitamin E
at. 1996). 400mglday
The SUVIMAX study in France is still-going (Hercbcrg et al. 1993). NO Maclennan et Previous Wheat bran 378 4 years Significant reduction in
data on colorectal cancer incidence were reported from the CARET study al. 1995, adenoma 11 q/day the number of adenomas
Australia ~1 cm in the low-fat/
(Ommcn et al. 1996). A total of 18 134 subjects at high risk of lung cancer high-fibre group
(hrnvy smokers nntl nsl~cstos-cxl~osctl workers) were included to assess the Vargas and Previous Wheat bran 1400 5 years hdenoma recurrence
effect of /I-carotcnc ant! vitamin A. This study was stopped prematurely be- Alberts et al. adenoma 13.5 g/day
CZILISCthe active treatment croup was found to have a significantly higher 1992, Arizona
risk of lung cancer than the placebo group. Faivre et al. Previous lrpaghula husk 656 3 years Adenoma recurrence
1997, Europe adenoma 3.8 g/day
128 1.Faivre
andC.Bonithon-Kopp Chcmoprevention Cancer 129
of Colorectal

ghula husk, 3.8 g/day (Fnivre et al. 1997). This dose was that proposed by the and only one out of six case-control studies suggests a protective effect of .
manufacturer to obtain stool bulking. Most of the above-mentioned studies high calcium intake.
are larger than the chemopreventive trials of vitamins. Their duration varies Four intervention studies aimed at evaluating the possibility of primary
between 3 and 5 years. The compliance rate for fibre intake was 79% in the prevention of colorectai cancer with calcium supplements have been carried
New York study (De Cosse ct al. 1989) and 74% in the Australian study out or are on-going (Table 3). Ail these studies are investigating subjects
(MncLennnn et al. 1995), and is currently 77% in the ECP study (intermedi- with a previous history of colorectnl adenoma. As mentioned before, such
ate results on 564 subjects who ended the study before April 1997). Compli- trials have the advantage of being both relatively small in size and short in
ance with the final colonoscopy is of great importance for the interpretation duration. In the ECP study, it was estimated with an assumed 30% recur-
of the results. 11was 92% at 2 ycnrs and 72% at 4 years in the Australian rence rate at 3 years in the placebo croup that 210 subjects per group are
study (McLennan et al. 1995). In the 1X1’ study, intermediate results indicate needed to detect a 15% difference between the tested group and the placebo
that compliance with the 3 year colonoscopy was 89%. group ((1=0.05; power=0.90. two-tailed test). As for the polyp growth study,
The first fibre chemopreventive study was performed in patients treated it can be estimated that there is an even higher proportion of patients with
for polyposis coli who had undergone total coiectomy and ileorectal anasto- an increase in size of the unresrcted adenoma. In the ECP study, eligible pa-
mosis and who were foiiowcd up at the Sloan-Kettering Institute in New York tients had to have at least one adenoma over 5 mm in diameter or two ade-
(De Cosse et al. 1989). The ratio between the initial number of adenomns nomas. This gives more power to the study because such subjects have a
and that at the follow-up examination was the main trial outcome. The in- higher recurrence rate than subjects with only a small adenoma. All these
tent-to-treat analysis suggested a limited effect of the treatment in the group studies use adenoma recurrence as the primary nutcome. The Oslo study has
receiving wheat bran, vitamin C and vitamin E compared with the groups re- the additional feature that the effect of the intervention on the growth rate of
ceiving vitamins alone or a placebo. There were significant differences at 33 an adenoma less than 1 cm in diameter left in situ in the large bowel is to be
and 39 months only. When compliance was taken into account there was a evaluated. None of the on-going studies has coiorectnl cancer as the main
stronger benefit in the combined fibre - vitamin group, particularly at the 2- end-point.
year midpoint of the study. The calcium being tested in the four studies is in the form of calcium cnr-
In the Australian multicentre study there was no evidence that any inter- bonate or calcium giuconolactate various doses: 1.2 g/day (Baron et al. 1995),
vention reduced the recurrence rate of adenomas at 2 or 4 years (McLennan 1.5 g/day (Rooney et al. 1994), 1.6 g/day (with a mixture of antioxidants; Hof-
et al. 1995), but a significant reduction in the incidence of large adenomns stad et al. 1995) and 2 g/day (Fnivre et al. 1997). The study duration varies
(21 cm) was found in the low-fat diet group. The effect was observed when from one to another. It was 2 years in one study, 3 years in two studies and
the low-fat diet was combined with wheat bran. This study suggests that a 4 years in one study (Table 3). In the Oslo and the Nottingham studies a
low-fat diet combined with wheat bran supplementation may reduce the risk control colonoscopy was performed yearly. In the two other studies control
of adenoma growth in patients with small adcnomas. colonoscopy has been planned only for the end of the study.
The final results from the ECP study and from the Arizona study will be The degree of compliance is an important factor in the success of the
available soon. study, since the study power depends on both the sample size and the degree
In conclusion, the results available provide some evidence for an inhibi- of compliance with the intervention. The compliance rate was 88% in the
tion of adenomn growth through a hi@-fibre diet and/or a low-fat diet. The
results of ongoiny, studies arc cxprctrtl to provide further arguments tu sup-
port these conclusions. Table3. Studydesigns,end-Points
andresultsof chrmoprevention
trialsof calciumin tolorecralcar-
Study ,,.Subjectr
with Intervention Noof DurationEnd-pointresults
Calcium Trials _: subjects
Hofstadet al. . ! See.Table1
It has been hypothesised that :I high intake of calcium may decrease the risk 1995,Oslo ’
Rooney et al: Previous
adenomaCalcium1.5g/day 79 2 years Noeffecton adeno-
of colorrctnl cancer. Srtpporl for this l~yl~olhcsis was obtained from a 19-year 19g4,Nottingham ma,recurrence
prospective study in ~hc USA and from the fact that oral intake of calcium Baronet al. 1995,Previous adenomaCalcium1.2g/day 930 4 years Adenoma recurrence
may induce a more quiescent equilibrium of epitheiial cell proliferation in USA : .,s:
the colonic mucosa of subjects at high risk of colorectal cancer. However, Faivreet al. 1997,PreviousadenomaCalcium1.0g/day 656 3 years Adenoma
such results have been reported in only half of the cell proliferation studies,
130 1.Faivre
andC.Bonirhon-Kopp Chemoprevention Cancer 131
of Colorectal

Nottingham study (Rooney et nl. 1991), 81% in the Oslo study (Hofstnd et ai. Conclusion
1995) nnci 73% in the ECP study (intrrmedinte results on 564 subjects who
ended the study before April 1997). Compliance with the finnI c~lon~scopy Altogether 15 chemopreventivc studies (sometimes with scvernl arms) have
cxnminntion w;\s 88% in the Nottinghnm study, 87% in the Oslo study nnd been performed to evaluate the possibilities of primary prevention of CO~O-
89% in the ECP study (intermediate results). rectal cancer: 8 in Europe, 6 in North America nnd I in Austrniin. Study pop-
The two completed studies were smnll. In the Nottingham study no effect uintions are made up of subjects with previous ndcnomas or with remaining
of calcium wns found on ndcnoms recurrence after 2 years; the recurrence adcnomns (i.e. intermediate steps in the natural history of the disease) or of
rntc was 11% in both the calcium nnd the plnccbo groups (Rooney et nl. volunteers included in lnrgc trials on cardiovnscuinr diseases and/or other
1994). In the Oslo study no effect on polyp growth wns found, but there was cancers in which colorcctai cancer risk is one of the end-points. In addition
a possible protective role of calcium and antioxidants against new ndcnoma to chemoprcvention studies, 3 studies consider dietary interventions. Such
formation. The two on-going studies - the ECP study and the American studies are more difficult to impiemcnt nnd evniunte than are chemoprcven-
study - nre larger. They will provide compicmentnry information within 1 tive studies. The first study of this type WRS performed in Toronto
year. (McKeown-Eyssen et ni. 1994). In the intervention group, ZIlow-fat diet (20%
It is not yet possible to draw firm conclusions on the effects of calcium of energy from fat) and 3 high-fihrc diet (SOg$iay) wns advised. After 12
supplcmcntntion in colorcctnl cnrcinogencsis, pnrticuinrly on adenomn months of counseling, fret consumption wns 25% of energy in the interven-
growth or adcnomn recurrence. On-going studies are expected to provide tion group snd 33% in the control group, nnd fibre consumption WZIS35 g
further information. and 15 g, respectively. There wns 3 nonsignificnntiy reduced risk of adenomn
recnrrcnce in women 2nd nn opposite risk in men. Thus, the issue of a gcn-
dcr-related effect on ndenomn recurrencr remains R definite question to be
Aspirin Trial addressed in much larger studies. In the Australian study, ns already men-
tioned, ;! low-fat diet (<25% of calories from fat) was proposed in one arm
Scvernl lines of evidence support the notion that aspirin nnd other monostcr- of the study (MacLcnnnn et al. 1995). The National Polyp Study proposed a
oid anti-inflnmmntory drugs mny prevent large bowel cancers. Most case- low-fat diet (<20% of total cr\iorics from fat), n high-fibre diet (at least 18 gl
control and cohort studies intlicllte n 30%-50% reduction in risk of colorcc- kcal of wheat bran) and fruit and vcgetnbles (5-8 servings per dny) in the in-
tai cancer among rcgulnr users of aspirin. The results arc consistent both for tervention arm (Freedman nnd Schntzkin 1992). Overall, 2094 subjects hnve
colon cnncer nnd rectal cancer mortality or incidence and for ndenomn oc- been randomized in this study aimed at cvniunting ndenoma recurrence.
currence. The results nrc not uniform, however, nnd n few studies found no This review does not consider trials with only indirect end-points. In such
benefit with aspirin use. studies, available results arc suggestive of treatment efficncy in reducing CO~O-
Only one chemoprcvcntion study has investigated the effect of aspirin on rectal cancer risk, though not decisive. These results nre of interest within in-
occurrence of colorectal cancer (Gnnn et ni. 1993). !n this study, performed tervention studies, ns they represent n unique opportunity for better under-
in mnic physicians in the USA, one nspirin tablet (325 mg) or a placebo was standing of the pathogenesis of colorectnl carcinogcnesis. Levels of cell prolif-
taken every other d;ry. This study wns stopped after 5 years because of evi- eration in the intestinni mucosa have been evaluated in several studies (Mac-
dence of protection ;\gninst myocartii~l infarction. No protection by aspirin Lennan et al. 1991; Fnivre et al. 1397). Chnngcs in the proliferation pattern
against colorcctnl tumours was seen. The relative risk wns 1.15 for cancer have been correlated with the risk of colorcctal turnours. It is worth evaluating
and 0.86 for adenomns for subjects randomized to aspirin group. The rcla- the effect of the intervention on coionic cell proliferation. A detailed analysis 01
tiveiy short duration of treatment cnn cxplnin this result. Some dntn suggest bile acids nnd related compounds is also plrmned in some chcmoprevcntive
that regular aspirin use for 10 years or longer is required for the inverse as- studies (Hofstnd et al. 1995; Fnivrc et al. 1997). Their involvement in colorectnl
sociation to become nppnrcnt. Furthermore, cancers found soon nfter mn- carcinogenesis hns been put forward. and the objective of the intervention is to
domizntion were probably present when :\spirin therapy began and wouId decreasetheir toxic effects. In this context, it is important to document changes
most likely not have been affected by aspirin use. There is little information in their concentrations in fneccs, for better definition of their role in the initial
regarding the optimal dose of aspirin. Benefits and risks have to be better phases of colorectal cnrcinogenesis. Assessments of the underlying nutritionnl
drfincd. IJrcnusc of llic known lonicity of :lspirin thcrc is not fl sufficient stnfus before nncl after the intervention are important in the interpretation of
basis to recommend aspirin to the public for preventing colorcctal cancer. the results. Diet needs to be estimntcd, with pnrticulnr emphasis on the mnin
hypotheses concerning colorcctnl cancer carcinogenesis.
It can be concluded that most available dntn do not support ;\ protective
effect of antioxidant vitamins (vitamin C, /I-carotene, vitamin E, association
132 1. Faivrr and C. Bonithon-Kopp Chemopreventionof ColorectalCancer 133

of these vitamins) or micronutrients (selenium, zinc) on adenoma recurrence llcrcherg S. Briancon S, Pavier A (1993) Le projet SUVlMAX/lOOOOO volontaires pour la re-
and growfh and/or colorectnl cancer risk. Results from small calcium chemo- cheiche en nutrition dnns le dnmaine de la prPvenrion. Cah Nulr Diet 1:55-64
Hofstad II, Vatn M, Hoff C, Larsen S, Osnes M (1992) Growth of colorecral polyps: design
preventive studies are difficult to interpret, and the same applies to the effect of a prospective, randomized, placebo-controlled intervention study in patients with co-
of dietary fibre. Although results are conflicting, there are some arguments Jorectal polyps. Eur J Cancer Prev 1:4lS-422
in favour of a protective effect of dietary fibre and/or a low-fat diet on adeno- Hofstad n. Almenningen K, Vatn M, Norheim Andersen S. Owen RW, Larsen S. Osnes M
ma growth. The results of on-going preventive studies will provide further (1995) Effect of calcium and antioxydants on growth of colnrectal polyps (absrract). Cut
37:A 34
data on the effect of calcium nnd libre on colorectal cnrcinogenesis. They will
McKeown-Eyssen G, llolloway C. Jazmaji V, Bright-See E. Dion P, Bruce WR (19X8) A ran-
be available within 1 year. domized trial of vitamins C and E in the prevention of recurrence of colorectal polyps.
Cancer Res 48:4701-4705
Acknowledgments. This study, performed within the ECP colon group, was supported for McKeown-Eyssen GE,Bright-SeeE, RruceWR, Jazmaji V, TorontoPolyp PreventionCroup
its coordination hy the Europe Against Cancer Programme, the Association ConIre Le Can- (1994) A randomized trial of a low fat high fihre diet in rhe recurrence of colorecrnl
cer (l3russels). the Associnrion Luxrmhonrgeoise Contre Ic Cancer and the French Ministry polyps. J Clin Epidemiol 47:525-536
of Health (PHRC). Thr calcium and its placebo were provided by the Sandoz France Com- MacLennan R, Rain C, Macrae E Gratten H, f!attis~utta D, nokey EL, Chapuis P, Goulston K.
pany. The fihre and its placcho treatment were provided by Reckitt and Colman (UK). Lamhert J, Wahlquist H, Ward M (1991) Design and implementation of the Australian
polyp prevention project. Front Castrnintest Res l&60-73
Maclannan R, Macrae PA, Rain C, Ilattistutta D. Chapuis P, Cratten II (1995) Randomized
trial of intake of fat, liher and beta carntene to prevent cnlorectnl adenomas. J Nat1 Can-
References cer fnst 87:1760-1766
Ommen CS, Goodman GE, Thornquist MD et al (1996) Effects of a combination of beta-
Alpha-Tocopherol. Reta Carotene Cancer Prevention Srutly Group (1994) The effect of vita- carotene and vitamin A on lung cancer and cardiovascular disease. N Engl J Met1
min E and beta carotene on the incidence of lung cancer and other cancers in male 334:1150-1155
smokers. N Engl J Med 330:1029-1035 Parkin DM, Pisani P, Ferlay J (1993) Estimates nf the worldwide incidence of eighteen ma-
Baron JA, Tosteson TD. Wargovich MJ (1995) Calcium supplementation and rectal mucosal jor cancers in 1985. Int J Cancer 54:94-606
proliferation: a randomized confrollcd trial. J Nat1 Cancer lnst R7:1303-1307 Potter JD, Slattery ML, nostick RM. Capstur SM (1992) Colon cancer: 3 review of rhe epide-
Rerrino F, Sant M. Verdecchia A. Capocaccia II. Itakulinen 1; Esreve J (1995) Survival of miology. Epidemiol Rev 15:499-545
cancer patients in Europa. The EUROCARI~ study. (IARC scientiIic publications no. 132). Roncucci L, Di Donatn P, Carati L. Ferrari A, Perini M. Ilertoni G, Redogni G, Paris D. Sv3-
IARC. Lyon noni 6 Cirola M, Ponz de tion M (1993) Antioxidant vitamines or lactulose for the pre-
Bonelli L, Conio M, Picasso M, Mnssn P, Dodero M, Ravelli P, Missale G, Uruzzi P, Aste II vention of the recurrence of colorectal adennmas. Dis Colon Rectum 36:227-234
(1994) Chcmoprevention of metachronous atlenomas of the large bowel: a dnuhle blind Rooney PS, Gifford KA, Clarke PA, llardcastle JD. Armitage NC (1994) A double-blind ran-
randomized trial of antioxidants (abstract). 3rd United European Gastroenterolngy Week, domized controlled of dietary calcium supplementation in individuals with adenomas
Oslo, abstract book, A61 (one-year results) (abstract). Dis Colon Rectum 37:1’41
Bussey HJR, De Cosse JJ, Deschner EE, Eyers AA, Lesser ML, Morson DC, Ritchie SM, Tho- Vargas PA, Alberts DS (1992) Primary prevention of colarectal cancer through dietary mod-
mas JPS, Wadsworth ]A (1982) Randomized trial of ascorbic acid in polyposis cnli can- ification. Cancer 70~1229-1235
cer. Cancer 50:1434-1439
De Cosse JJ. Miller 1111,Lesser ML (1989) Effect of wheat liber and vitamin C and E on rec-
tal polyps in patients with familial adcnomatous polyposis. J Nat1 Cancer Inst 81:1290-
Paivre J, Wilparr M, Boutron MC (1991) Primary prevention of large bowel cancer. Recent
Results Cancer Res 122:X5-99
Paivre 1, CouiIlnult C. Kronbor~ 0, Itarh U, Ciacora A, De Olivrira il. Obrador T. O’Morain
C, ECP Colon Group (1997) Chemoprevention of metachronous adenomas of the large
bowel; design and interim results of a randomized trial of calcium and Iihre. Eur J Can-
cer Prev 6:132-13&l
Freedman IS. Schatzkin A (1992) Sample size for studying intermediate end points within
intervention trials or observational studies. Am J Epidemiol 136:1148-l 159
Cann PI!, Manson JE, Clynn RJ (1993) Low-dose aspirin and incidence of colnreclal tu-
mours in a randomized trial. J Natl Cancer lnst 85:1220-1224
Greenberg ER, Baron ]A. Tosteson TD, Freeman DM. Beck CJ, Bond JH, Colacchio TA, Col-
ler ]A, Frank1 MD, Ilaile RW, Man&l 12, Nierenberg DW, Rothstein R, Snover DC, Ste-
vens MM, Summers RW, Van Stnlk RU (1994) A clinical trial of antioxidant vitamins to
prevent colorectal adennmn. N Enel J Med 331:141-147
Hennekens Cl-l. Iluring Jr, Manson Jli et al (1996) Lack of effect of long-term supplementa-
tion with beta carotene on the incidence of malignant neoplasms and cardiovascular dis-
ease. N Engl J Med 334:1145-l 149
The New England Journal of Medicine



ABSTRACT wheat-bran fiber has been shown to dilute fecal con-
Bnc,&~c~~nri The risks of colorectal cancer and centrations of bile acids43sand to bind bile acids, there-
adenoma, the precursor lesion, are believed to be in- by increasing their fecal excretion.697Although an in-
fluenced by dietary factors. Epidemiologic evidence verse correlation 1va.sobserved between mortality rates
that cereal fiber protects against colorectal cancer is from colon cancer and per capita cereal consump-
equivocal. We conducted a randomized trial to deter- tion,* the results of the feJv analytical epidemiologic
mine whether dietary supplementation with wheat- studies of associations benveen the consumption of
bran fiber reduces the rate of recurrence of colorec- whole-grain cereal and the risk of colorectal cancerg-1;
tal adenomas. or adenomalb have been equivocal. Some metabolic
iUt?ethocls We randomly assigned 1429 men and end-point studies,s~*7including our onn,4 have shown
women who were 40 to 80 years of age and who had that wheat-bran fiber decreasesfecal mutagenicity and
had one or more histologically confirmed colorectal
adenomas removed within three months before re-
reduces concentrations of fecal bile acids, although no
cruitment to a supervised program of dietary sup- effect was found on rates of proliferation of rectal mu-
plementation with either high amounts (13.5 g.per cosal TLVOstudies found that a supplement of
day) or low amounts (2 g per day) of wheat-bran fiber. wheat-bran fiber had no effect on the risk of recurrent
The primary end point was the presence or absence colorectal adenoma.19*2°
of new adenomas at the time of follow-up colonos- In 1990, we initiated a multicenter trial to deter-
copy. Subjects and physicians, including colonosco- mine whether wheat-bran fiber can prevent the re-
pists, were unaware of the group assignments. currence of colorectal adenomas.
Resr& Of the 1303 subjects who completed the
study, 719 had been randomly assigned to the high- METI-IODS
fiber group and 584 to the low-fiber group. The me-
Study Design and Subjects
dian times from randomization to the last follow-up
colonoscopy were 34 months in the high-fiber group Details of the design and methods of the study have been de-
and 36 months in the low-fiber group. By the time of scribed previously.” Briefly, subjects were recruited bcnvecn Scp-
the last follow-up colonoscopy, at least one adenoma tember 1990 and July 1995 from multiple centers in the Phocti,
Arizona, metropolic& area. The study protocol was approved by
had been identified in 338 subjects in the high-fiber the institutional review boards of the 22 .participating health care
group (47.0 percent) and in 299 subjects in the low- centers in the Phoenix area and by the human-subjects committee
fiber group (51.2 percent). The multivariate adjusted of the Univcrsiy of Arizona. All subjects provided written informed
odds ratio for recurrent adenoma in the high-fiber consent.
group, as compared with the low-fiber group, was 0.88 We idenrified men and women who were 40 to 80 years of age
(95 percent confidence interval, 0.70 to 1.11; P=O.28), f:Dm whom one or more colorectal adenomas, measuring at leas:
and the relative risk of recurrence according to the 3 mm in diameter at colonoscopy, had been removed within thy
number of adenomas, in the high-fiber group as com- three months before recruitment. To be eligible, subjects had to
pared with the low-fiber group, was 0.99 (95 percent have an adequate nutritional status and normal renal and liver fkc-
tion and to have a Southxvest Oncology Group pcrformancc sta-
confidence interval, 0.71 to 1.36; P=O.93).
tus of 0, 1, or 2.21 TVe escludrd persons xrho hadhad invasive can-
Concl&ons As used in this study, a dietary sup- cer ,\vithin the previous five years; those with a his ory of colon
plement of wheat-bran fiber does not protect against resection; those who hJd txvo or more first-deg:ec Kladves \vith
recurrent colorectal adenomas. (N Engl J Med 2000;
02000, Massachusetts Medical society.
From the Arizooz CancerCenter(D.S.A., hl.E.hI., D.J.R., J.Xf.G.-R,
J.R.hi., J.B.L., AI.E.R.), the College of Public Health (D.S.A., bl.E.hl..
D.J.R., J.R.hl., M.E.R.), and the Depaxn-cnts of Pathology (A.B.B.) and
HE risks of colorectal cancer and adenoma, bIcdicine (D.S.A., D.L.E., R.E.S.), University of Arizona, Tucson; the Gn-
the precursor lesion, are believed to be in- tu for Health Rescuch, Portland, Orcg. (CR.); the Dcparuncnt of PC&
attics, Univcrsiy ofArkJnsas for Medical Scicnccs,Licrlc Rock (PAY); ~nz
fluenced by diet.1 Burkitt’s proposal that a Veterans A&irs XIcdisal Ccnrer, Tucson, Ariz.
high-fiber diet protects against colon can- quests to Dr. Alberrr at the Arizona Cancer
cer VYIS based on the low rates of colorectal cancer in son, AZ 85726-5024, or at dalbcns~azcc.arizona.cdu.
Orhcr authors uerc Dianne Parish, B.S., Kris Kooncc, B.S., and Liannc
Africa.2 Insoluble fibers, such as wheat-bran fiber, are F&s, M.P.H., Arizona Cancer Ccntcr, University of Arizona, Tucson.
thought to protect against colon cancer by absorb- ‘The members of the phoenix Colon Cancer Prevcncion Physicians’ h’cr-
ing carcinogens in the gastrointestinal tract.3 Indeed, work arc listed in the Appcndir.

1156 . April 20, 2000

colorectal cancer, severe metabolic disorders, or other severe ill- to detect a 25 percent reduction in the recurrence of adenomas
nesses; and those with an intake of more than 30 g of dietary fiber and a power of 0.94 to detect a 30 percent reduction.
per day on the basis of their responses to the Arizona Food-Fre- An interim analysis conducted in the latter part of the study
quency suggested a difference between groups in the proportion of subjects
Subjects who successfully completed a six-week run-in period who stopped taking the assigned supplement: 12.7 percent stopped
-z, M.S.,
by consuming at least 75 percent of the amount of a lou-fiber sup- in the low-fiber group, and 23.3 percent stopped in the high-fiber
‘H.D., plement supplied (2 g per day) were randomly assigned to receive group. Therefore, for the remainder of the accrual period, the orig
1.D., a high-fiber supplement (13.5 g per day) or a lou-fiber supplement inal 1:l schedule of randomization was changed to 4:1, with four
(2 g per day) ofuheat-bran cereal. Mth the esception ofthe cereal- subjects assigned to the high-fiber group for every one assigned to
fiber intervention, no other dietary changes were required. The the low-fiber group.
treatment assignments were not revealed to the subjects, their phy- We counted all adenomas, ahether detected during the first co-
sicians, or members of the study staff. The fiber supplements were lonoscopy (at y-ear 1) or subsequent colonoscopic examinations.
ziI con- provided by Kellogg (Battle Creek, Mich.) and \vere available in sev- Subjects in whom an adenoma was found during the one-year co-
-s, there- eral forms: unsweetened loop-shaped cereal and stveetened and un- lonoscopy were not withdrawn from the study. live separate analy-
h an in- sweetened shredded cereal. Analysis of the fiber content per serv- ses were performed. The first included all subjects who underwent
ity rates ing showed the following: high-fiber loops, 13 g; low-fiber loops, colonoscopy one or more times after randomization, with recur-
nsump- 2 g; high-fiber unsweetened shredded cereal, 13 g; low-fiber un- rence defined as the identification of one or more adenomas after
sweetened shredded cereal, 4 g; high-fiber sweetened shredded ce- randomization. The second set of analyses included only subjects
liologic real, 10 g; and low-fiber sweetened shredded cereal, 3 g. Cereal who underwent colonoscopy at one year and one or more times
,tion of boxes were color coded into six groups to help maintain the study thereafter. Recurrence was defined for these analyses as the idcn-
mcer9.15 blinding. Midway through the study, high-fiber wheat-bran-fiber tification of any adenoma after the onc-year colonoscopy. Diffcr-
xabolic bars (containing 10 g of fiber) and low-fiber bars (4 g of fiber) ences between the high-fiber group and the low-fiber group in the
were developed by Kellogg. Subjects who had completed two years rates of colonoscopy at one year and during follo\v-up were ana-
‘: shown
of the study were allowed to elect to consume up to 25 percent lyzed with the USCof &i-square tests, and the difference between
:ity and of their daily fiber supplement in the form of a fiber bar. the groups in the length of time from randomization to the last
ugh no Compliance with the protocol was evaluated primarily by counts colonoscopy was assessed with a log-rank test. DifIerences in char-
-tal mu- of returned cereal boxes and fiber bars at each visit and second- acteristics and in the incidence of adverse events among patients
nent of arily through a specialized intake calendar. Each index was used to with recurrent adenomas in the twn groups were tested with chi-
generate an overall compliance score; subjects who consumed more square tests.
current than 75 percent of the cereal supplement were classified as com- Multivariate adjustment to test for an effect of wheat-bran fiber
plying with the protocol. On the basis of these data, members of was initially performed with the use of logistic regression (presence
1deter- the clinic research staff initiated individualized measures, as nec- vs. absence of an adenomaj. We used generalized estimating cqua-
the re- essq, to increase compliance. tions with a Poisson link Function to model the number of recur-
rent adcnomas at each colonoscopy, adjusting for the timing of
Colonoscopy colonoscopy and assuming an exchangeable correlation structure
The study protocol specified that follow-up colonoscopy be per- among the repeated procedures .a? Generalized estimating equa-
formed twice afier the initial qualifying colonoscopy. The first co- tions were used to estimate the adjusted relative risk of the recur-
lonoscopy was to take place one year a&r randomization (to iden- rence of adenomas for the high-fiber group as compared with the
tify and remo-;e adenomas missed at the qualifying colonoscopy), lotv-fiber group, whereas logistic regression was used to estimate
5ccn de- the adjusted odds ratio (as an estimate of the adjusted relative risk).
:en scp- and the second two years thereafter. However, the national rec-
oinmendations regarding the frequency of colonoscopic surveil- Initial models fitted to test the effect of group assignment were
‘hoenix, adjusted only for the randomization period. Subsequent statistical
-oved by lance of patients with a history of colorectal adenomas changed dur-
ing the study from one and three years after the initial resection modeling also adjusted for sex and the number of adenomas at
Ah care the base-line colonoscopy (both of which are strong predictors of
mmirrce to three years after resection.*+-26 Thus, there was a decrease in the
rate of colonoscopy at one year among subjects enrolled in the the risk of recurrence) and factors that were found to be signifi-
lformcd cantly different beoveen groups at base line. The significance of
latter part of the trial.
the treatment effect was assessed with the Wald statistic.
3 of age
Data Collection
; at least RESULTS
thin the Results of endoscopy and pathological analysis were collected for
s had to each colonoscopy reported during the study. Using standardized Enrollment and Randomization
er hnc- guidelines, we abstracted data on the completeness of the cxamina-
nce sta- tion and on the location, size, and histologic features of any resect- We identified 4705 potentiaLly eligible subjects. Of
ivc can- ed adenomas. these, 2058 declined to participate, 1006 were found
)f colon Complete blood counts and blood chemical analyses were per- to be ineligible, and 102 dropped out before the run-
:es with formed during screening and during the run-in phase of the stud)
and annually thereafter. Diet was assessed according to the same
in phase. The remaining 1509 subjects entered the
schedule with use of the Arizona Food-Frequency Questionnaire, Sk-week run-in phase, nhich consisted of the daily
which has been evaluated with respect to reliability and validity in intake of a supplement low in wheat-bran fiber (2 g
this populationrr Information on adverse events was obtained ev- per day). Of the 3699 eligible subjects, 1429 (38.6
eq three months at the time the dietary supplement was dispensed. percent) successhtlly completed the run-in period and
1.B.) and underwent randomization, 627 to the loin-fiber group
rhc Gn- Statistical Analysis
of PC&- The original trial design and approach to analysis were de-
and SO2 to the high-fiber group.
A!); and scribed in detail by Emerson et aLa8The target sample size of 1400
print rc- subjects was based on a three-year rate of recurrence of adenomas Base-Line Characteristics of the Subjects
24, Tuc-
of 40 percent and on an estimate that 10 to 15 percent of adcno- Table 1 shows the base-line characteristics of ah
mas bvould be missed during the colonoscopy at base line. Given
a predicted dropout rate of 25 percent over a period of three years,
1429 randomized subjects and of the 1303 subjects
we estimated that 950 subjects would complete the intervention. (91.2 percent) who completed the study by under-
Given this sample size, the srudy had a statistical power of 0.82 going at least one colonoscopy after randomization.

Volume 342 Number 16 * 1157
The Nelv England Journal of -_-.


(N=1429) THE STUDY IN=13031
(s=627) (s=802) (x=584) (P719)
Age - yr 66.0Z8.8 66.8 -9.0 66.O”a.m 66.428.8
Malt sex - no. (%) 409 (65.2) 538 (67.1) 385 (65.9) 486 (67.6)
Dietary intake
Energy - kcal/day 1875%636 19412709 10742629 19392692
Total ht - g/day 71.0+32.0 75.1z35.1 70.7-31.4 74.7234.6
Dicczy fiber - g/day 18.8Z8.3 18.528.2 18.9Z8.3 18.628.1
Calcium - mg/day 852~371 8582385 8492368 856%38j
Alcohol - g/day 6.1~10.9 8.1517.9 6.4~11.1 8.4~18.0
IO-year history of regular aspirin USC 165 (26.3) 230 (28.7) 154 (26.4) 213 (29.6)
-- no. (%)
Current smoker - no. (%) 67 (10.7) 136 (17.0) 5; (9.8) 121 (16.8)
History of adcnomasbefore bare-line 210/544 (38.6) 272/722 (37.7) 199/504 (39.5) 253/64S (39.0)
colonoscopy - no./total no. (X)
History of colorcctal cancer in 1 parent 99 (15.8) 141 (17.6) 91 (15.6) 129 (17.9)
oc sibling -no. (%)
Six of largest adcnoma - mm 9.727.1 10.127.6 9.5~6.8 10.127.7
No. of adcnomns 1.8Zl.5 1.8~1.2 1.8~1.5 1.8Z1.2
Location in proximal co103alone 165/624 (26.4) 220/799 (27.5) 155/581 (26.7) 95/716 (27.2)
- no./rotil no. (%)
Villous histologic ficdings 95/625 (15.2) 119/SOl (14.9) 91/582 (15.6) 07/718 (14.9)
- no./total no. (%)t

‘Plus-minus ~alucs arc means zSD.
IThis category included tubulovillous and villous adrnomxs.



number of SubjectShed number (percent)

Low-fiber 584 548/554 (93.8) 536/534 (91.8) 472/544 (86.8) 459/543 (84.5) 425,408 (83.7) 399/504 (79.2)t
High-fiber 719 626,‘719 (87.1) 597,‘719 (83.0) 468/601 (77.9)t 442/598 (73.9)t 409/552 (74.l)t 376/544 (69.l)t

‘Complizncc was defined as consumption of more than 75 percent of the assigned dietary supplcmcnts. Numbers of subjects do not total 1303 because
of dropouts, deaths, or missing data. Compliance was asscsscdby a counr of the boxes of cereal and fiber bars rcturncd at each plmncd clinic visit and by
an assessmentof required calendar notxions made by subjects concerning the number of ccrcal boxes or fiber bars consumed each day.
t1’<0.05 for the comparison with the low-fiber group.

Of these 1303 subjects, 138 undenvent only the one- ception of the first year of the study, there was a sig-
year colonoscopy. The results for all randomized sub- nificant difference in compliance between the nvo
jects who underwent colonoscopy after randomiza- groups (Table 2): the proportion of subjects who con-
tion were included in an intention-to-treat analysis. sumed more than 75 percent of the cereal supple-
ment was Ion-er in the high-fiber group than in the
Compliance low-fiber group (P<O.O5). Counts of rerurned bos-
We assessed compliance with the dietary-supple- es indicated that compliance declined with each year
ment regimen by two methods: a count of cereal box- of the study, so that by the third year, 84 percent of
es returned to the study sites and a calendar record the low-fiber group and 74 percent of the high-fiber
of consumption kept by each subject. With the ex- group were consuming more than 75 percent of the

1158 - April 20, 2000

supplement. On the basis of responses to the Arizona cording to either the initial 1:l scheme or the 4:l
Food-Frequency Questionnaire, which includes in- subsequent scheme; however, subjects who undenvent
take of fiber from the wheat-bran-fiber supplement randomization during the later period undenvent sig-
and other sources, the mean total intake of fiber was nificantly fewer colonoscopic examinations during year
27.5 g per day in the high-fiber group and 18.1 g per 1 than those nho undenvent randomization during
day in the low-fiber group. the initial period. This difference clearly resulted from
the change in clinical screening practice.
Recurrence of Adenomas Table 4 shows the rates of recurrent adenomas
As noted in the hiethods section, we changed the among the 1303 subjects Lvho completed the study.
randomization scheme during the latter part of the The median observation period was 34 months in he
study. As a result, 276 of the 1303 subjects undenvent high-fiber group and 36 months in the low-fiber
randomization according to a 4:l ratio (high fiber to group (P=O.O06). By the time of the last follo\v-up
lo~v fiber) (Table 3). We did not detect significant dif- colonoscopy, the percentage of subjects R’ith one or
ferences benveen the high-fiber and low-fiber groups more recurrent adenomas \vas 51.2 percent in the low-
in the number of colonoscopic procedures performed fiber group and 47.0 percent in the high-fiber group
among subjects who undenvent randomization ac- (P=O.13). After adjustment for the randomization
scheme used, the odds ratio for the presence of at least
one recurrent adenoma in the high-fiber group, as
compared with the low-fiber group, was 0.88 (95
TAaLE 3. NUhrBER OF CoLo~osCorr~ EWJISATIOXS AFTER percent confidence interval, 0.70 to 1.11; P=O.28).
RX.~OMIZ.UIOS, ACCORDISG TO TBEATMEX~ GROUP When the analysis was restricted to the 889 sub-
jects who underwent both a one-year colonoscopy
and another examination two years later, the recur-
Low-hem GROUP HIGH-RIERGROUP rence rates in the high-fiber and low-fiber groups were
VARIAEXE IN=5231 IN=5041
not significantly different. With the use of general-
1:l Randomization scheme (n=1027) ized estimating equations, the relative risk in the high-
hlc;m no. of colonoscopics 2.17_~0.&?3 2.15?0.90
1 Colonoscopy - no. (%) 101 (19.3) 113 (23.4)
fiber group, as compared wiih the low-fiber group,
S2 Colonosco~ics - no. (%) 422 (80.7) 391(77.6) was 0.99 (93 percent confidence interval, 0.71 to 1.36;
P=O.93) for all 1303 subjects and 1.0s (95 percent
IN=611 (N=215) confidence interval, 0.71 to 1.64; P=O.73) for the
__-. 859 subjects who underwent colonoscopy during
4.1 Randomization schcmc (n= 276)
l Mean no. of colonoscopics 1.64zO.61t 1.66-cOSSt year 1. Additional adjustments for sex, the number of
1 Colonosiopy - no. (%) 26 (42.6) 86 (40.0) colonoscopic examinations, the number of adenomas
Z2 Colonoscopics - no. (%) 35 (57.4) 129 (60.0) found at the base-line colonoscopy, and base-line var-
‘Plus-minus v&x arc means ZSD. iables that differed significantly between treatment
DM tPCO.05 for the comparison with the corresponding group in the 1:l groups did not change the results. Separate analyses
‘RD randomization scheme. revealed no significant differences in the rates of re-





no. with Z-1 recurrent mean no. of
adenomasltotal no. I%1 recurrent adenomas
a sig-
: two After random- 1303 338/719 299/S% 0.85 0.61 0.57
ization (47.0) (51.2) (0.70-1.11) (i7;?.36)
) con- After colonos- 859 16X/468 153/421 1.04 0.60 0.53 1.0s
apple- copy ac 1 year (35.9) (36.3) (0.79-1.35) (0.71-1.64)
n the
box- l CI dcnotcs confidcncc inccrval.
1year tThc odds ratio for the prcscnce ofat lcast one rccurrenr adcnoma in the high-fiber group as com-
pared with the low:fibcr group, adjusted for the randomization scheme,is shown.
‘nt of $Thc r&tivc risk of rccurrcnt adcnomas in the high-fiber group as compared with the low-fiber
-fiber group, adjusted for the randomizxion schcmcand riming ofcolonoscopics with the USCof generalized
)f the estimating equations (Poisson link function), is shown.

Volume 342 Number 16 . 1159
The New England Journal of Medicine


Ctaucrmsnc IN=2991 IN=3381 VALUE’ VARIAFU IN = 5841 IN=7191
no. of subjects (%I no. of subjects(%I
Size of largest adenoma 0.71 Death 10 (1.7) 13 (1.8)
<l cm 205 (68.6) 227 (67.2) Disease*
21 cm 94 in4j 111 (32.8j All cancers 38 (6.5) 51 (7.1)
No. of adcnomas 0.03 cadi0va5cu1ar discasc 11 (1.9) 6 (0.8)
1 145 (48.5) 144 (42.6) Stroke 5 (0.9) 6 (0.8)
2 66 (22.1) 61(18.0) Gastrointestinal effects*
23 SS(29.4) 133(39.3) Nausea 21 (3.6) 44 (6.1X
Location of adcnoma 0.004 Abdominal pain 69 (11.8) 136 (18.9)t
Distal colon and rectum 87 (29.1) 77 (22.8) Diarrhea 65 (11.1) 145 (20.2)t
Proximal colon 144(48.2) 140 (41.4) Constipa:ion 78 (13.4) 91 (12.7)
Both 60 (20.i) 110 (32.5j Intestinal gas 135 (23.1) 243 (33.8)t
l+x specified 8 (2.7) 11 (3.3) Abdominal bloating 59 (10.1) 121 (16.8)t
Histologic findings 0.51
Tubular adcnoma 197 (65.9) 224 (66.3) *Some subjects had more than one adverse cvcm.
Tubulovillous or villous adcnoma 25 (8.4) 28 (8.3) tPCO.01 for the comparison with cbc low-fiber group.
Both 10 (3.3) 19 (5.6)
Not specified 67(22.4) 67(19.8)

*The chi-square test was used.

currence between women in the low-fiber group and occurrence of extracolonic cancer (P=O.58), cardio-
women in the high-fiber group (40.7 percent vs. 40.8 vascular disease (P=O.37), or stroke (P=O.69). The
percent, P=O.99). Among the men, there were fewer number of subjects who reported gastrointestinal
recurrent adenomas in the high-fiber group than in effects was significantly higher in the high-fiber group
the low-fiber group (50.0 percent vs. 56.6 percent); than in the low-fiber group for all effects except con-
this difference was of borderline statistical significance stipation (Table 6). Most of these adverse effects
(P=O.OS). There was no evidence of an effect ofsup- were mild.
plementation with wheat-bran fiber among male sub-
jects who underwent colonoscopy during the first year. DISCUSSION
When we assessed the characteristics of the recur- In this double-blind trial, we found that a dietary
rent adenomas (Table 5), there was no significant dif- supplement of wheat-bran fiber had no statistically
ference between the two groups regarding the size of significant protective effect against recurrent colorec-
the adenomas (I’= 0.71) or their histologic appearance tal adenomas. This finding was unchanged whether
(P = 0.5 1). However, there was a significantly higher the analysis was based on all colonoscopic proce-
proportion of subjects with three or more recurrent dures performed after randomization or only those
adenomas in the high-fiber group than in the low- performed after one year in the study. This method
fiber group (P=O.O3). When subjects were classified of analysis has been used in other intervention stud-
according to the sites of the recurrent adenomas (prox- ies of recurrent colorectal adenoma.30 Moreover, the
imal colon or distal colon and rectum or both), the high-dose supplement of wheat-bran fiber had no ef-
high-fiber and low-fiber groups were significantly dif- fect on the number of recurrent adenomas in subjects
ferent (P=O.O04); this result was largely due to the who had a recurrence. Our results are consistent with
higher number of subjects in the high-fiber group those of the Toronto Polyp Prevention Trial19 and the
\lrho had recurrent adenomas in both the proximal Australian Polyp Prevention Project.20 Although our
colon and distal colon and rectum. secondary analyses suggested an effect of the high-
fiber supplement among men, this result probably rep-
Adverse Events resents a chance finding; in the Toronto Polyp Preven-
During the course of the study, nine cases of colo- tion Trial the effect of a low-fat, high-fiber diet was
rectal cancer \vere reported, two in the low-fiber greater among women than men.19 Furthermore, con-
group and seven in the high-fiber group (P=O.20). trary to the findings of the Australian trial, we did not
As shown in Table 6, among the 1303 subjects tvho see any evidence that the high-fiber supplement we
completed the study, there were 23 deaths: 10 in the used reduced the rate of recurrence of large adenomas.
low-fiber group and 13 in the high-fiber group. There We observed no protective effect of the high-fiber
were no significant differences between groups in the supplement on the number, location, or histologic fea-

1160 - April 20, 2000

tures of the recurrent adenomas. The combination tion with wheat-bran fiber may reflect inadequate
of these observations argues against the idea that di- follow-up: three years may be too short. It could be
etary supplementation with uheat-bran fiber can pro- argued that the total amount of dietary and cereal fi-
tect against recurrent colorectal adenomas. As report- ber consumed by the subjects in the high-fiber group
lBER ed in this issue of the Jozw~~111, Schatzkin et al. found was insufficient to protect against recurrent ade-
191 that a low-fat, high-fiber diet also failed to louver the nomas. It is also possible that a high-fiber diet may
risk of recurrence of colorectal adenomas.3’ be beneficial only in persons with lower base-line in-
We observed a relatively high rate of recurrent ad- takes of total fiber than those in our study. Alterna-
enemas in the proximal colon in both the low-fiber tively, the use of wheat-bran-fiber supplements may
group and the high-fiber group (45.2 percent and only protect against the progression of large ade-
41.4 percent, as compared urith respective rates of nomas to carcinomas. However, both the Nurses
26.4 percent and 27.5 percent at base line). When Health Study15 and the Health Professionals’ Fol-
rates of recurrent adenomas in the proximal colon are low-up Study16 failed to find that cereal fiber pre-
added to the rates of recurrence occurring in both vents colon cancer. Since cereal fiber has potentially
the proximal colon and the distal colon or rectum, healthful effects in the prevention of coronary heart
68.2 percent of the subjects in the low-fiber group disease,32*33public health recommendations3%3j that
and 74.0 percent of those in the high-fiber group emphasize increased consumption of complex car-
had recurrences in the prolcimal colon. The high rates bohydrates, whole-grain foods, and cereal products
of recurrent adenomas in the proximal colon strong- may nevertheless be appropriate.
ly suggest that colonoscopy, rather than sigmoidos-
copy, is the preferred method of surveillance, espe- Suppoxcd in pxr by Public Health Scrvicc grants (CA-41108 and CA-
cially in patients wirh a history of adenoma in the 23074) from the Xxional Catlccr Institute and by the KcUogg Company.
Dr. hludncz was supported by a Career Dcvclopmcnr Award (KOl
proximal colon. CA79069-10) from the National Cancer Institute. The contents of this ar-
In large, randomized clinical trials, randomization ticlc arc solely the responsibility of the authors and do not necessarily rcp-
is expected to result in a relatively equal distribution rcscnt the official views of the National Cancer Instinxc.
Prcscnted in abstract form ac the 91st annual meeting of the American
cardio- of subjects with respect to risk factors of interest. In Association for Cancer Research,San Francisco, April 1-5, 2000.
9). The our trial, there lvas a balanced distribution with re-
testinal spect to base-line age and sex, but imbalances in terms We are indr5ted to Dr. Victor Fulgoni Cfonnerly of Kclfop$ for
r group of exposure to tobacco, alcohol consumption, and Fis constant support; to Dr. Lee Scchest, Lisa Hesr, Ellen Graver,
and Kathleen IVoolffor their espert ndvice concerriirafl conrp!innce
‘pt con- total intake of fat. Nevertheless, the multivariate logis- atrd dietary arsessment; to Dr. Wilhant Stini for measrtrenrent of
effects tic-regression analysis, after adjustment for random- bone minrralizntion; to Drs. Mikel Aickin and Daniel McGee for
ization period, sex, smoking status, alcohol consump- their advice concerni,rg biostatistical de@ and analysis; and to
tion, and enerw intake, did not show a significant Kancy Mason-LiddiI, Evelyn Anthony, Anne DeJouJ-Rwhnau, and
Roberta Graham for their tireless efforts in the day-to-day perfw-
effect of supplementation with \f,heat-bran fiber on anre of the trinl.
dietary the recurrence of colorectal adenomas. Thus, our re-
istically sults do not appear to be related to an imbalance in APPENDIX
olorec- the base-line characteristics of the subjects or to the The mcmbcrs of the Phoenix Colon Cancer Prevention Physicians’ Nct-
rhether change in the randomization scheme from a 1:l ra- work were a.sfollows: hf Cohen, I?G.Foutch, R.T. McDermott, Jr., R. Saw-
proce- tio to a 4:l ratio in favor of the high-fiber group. yer, Jr., A. Toraya, B. htcCoUum, C. Stein, D. Mcline, D. Wadas,D. Douglas,
D.H. Wiiston, D. Johnson, D. Larson, D.-S. Cho, E.I. Lcff, E. Cooper,
i those Our experience underscores the difficulty of per- E.I. Alpcr, F. Ram&, F. Lcukouitz, F.J.Kogan, G. Scvcriio, G. Burdick,
nethod forming large-scale nutritional intervention trials, in J. Parcl, J. Burgcss, J. Kirkpauick, J. Shaver, J. Singer, J. hlcllcn, J. Bickcl,
n stud- terms of both recruiunent and compliance with the J.E. Phclps, J. Hanigsbcrg, J. Harlan, J. Mucllcr, J. Murphy, J. Leighton, K.S.
venkatcsh, K. I’arcnt, L. Pass,L.A. Bcttingcr, L. Rigbcrg. L. Shields, ME.
:er, the protocol. Of 3699 eligible subjects, 1303 (35.2 per- Harrison, bl. Goldblat, M. Hocfcr, M. Shaukat, M. Al-man, hf. Rock,
1no ef- cent) successfillly completed the trial. In addition, by hl. Schwimmcr, M. Shapiro, hi Sanish, AI. Anderson, P.S.Ramanujam,
ubjects P.J.Bcrggrccn, I? Eumar, R. Kcatc, R. Shah,R. Brooks, R. Jonas,R. Manch,
the third year of the study, only 74 percent of the sub- R.J. Spcnccr, R. Leon, R. .Sano\vski,R. Heigh, S. Bcllapravalu, S. Brown,
nt with jects in the high-fiber group, as compared with 84 S. Glouberman, S. Winograd, S. Kanner, v. Sartor, E Tay!or (dcccascd);
md the percent of those in the low-fiber group, consumed Data and Safety hfonitoring Comminee - E.R. Greenberg (Norris Cotron
Cancer Center, Hlurover, S.H.). R. Prentice (Fred Hutchinson Cancer Cen-
gh our more than 75 percent of their supplemental cereal on ter, Scattlc), E. Gritz (hI.D. Anderson Cznccr Center, Houston), R. Ha&
z high- a daily basis (a level \ve defined as indicative of com- (h’orris Comprehensive Cancer Center, Los Angeles).
dy rep- pliance). Despite these difficulties, the mean intake of
‘reven- total fiber was 27.5 g per day in the high-fiber group REFERENCES
iet was and 18.1 g per day in the low-fiber group. It can be 1. Porter JD. Food, nutrition and the prcvcntion of cancer: a global pcr-
e, con- argued that this level of intake over a period of three socctive. W&irxton, D.C.: World Cancer Research Fund/;\meric.%? Insti-
lid not t&c for Cancer Research, 1997.
years is inadequate to prevent recurrent adenomas; 2. Burkitt DI? Eoidcmiolom of cancer of chc colon and rectum. Cancer
:nt we however, OUTcompliance data indicate that higher dai- 1971,28:3-13. ’ L’
1oma.s. ly consumption of nvheat-bran fiber for longer periods 3. Eastwood hM. The physiological effect of dietary fiber: an update.
Annu Rev Nutr 1992;12:19-35.
h-fiber is not practical in adults older than 65 years of age. 4. Alberts DS, Ritcnbaugh C, Stop JA, ct al. Randomized, double-blinded,
sic fea- The lack of effect of three years of supplementa- placebo-controlled study of cffccr of wheat bran fiber and calcium on fecal

Volume 342 Number 16 - 1161
The New England Journal of Medicine

bit acids in patients with rcscctcd adcnomacouscolon polyps, J Natl Car- 20. Maclcnnan R, Macrac F, Bain C, et al. Randomized trial of int& of
ccr Inst 1996;88:81-92. fat, fiber, and beta carotene to prcvcnt colorcctal adcnomas: the Australian
5. Lampc JW, Slavin JL, Mclchcr EA. Potter JD. Effects ofccrcal and vcg- Polyp Prcvcntion Project. J Nat1 Cancer Inst 1995;87:1760-6.
ctablc fiber feeding on porcntial risk factors for colon cancer. Cancer Epi- 21. Martinez ME, Reid ME, GuillCn-Rodriguez J, et al. Design and base.
dcmiol Biomarkcrs Prcv 1992;1:207-11. line characteristics ofrrudy participants in the Wheat Bran Fiber trial. Can
6. Rcddy BS, Watanabe K, Wcisburgcr JH, Wyndcr EL. Promoting cffccr ccr Epidcmiol Biomarkcrs Prcv 199&X113-6.
of bile acids in colon carcinogcncsis in germ-free and conventional F344 22. Crecn S, W&s GR. Sourhwcsr Oncology Group standard response
rats. Cancer Rcs 1977;37:3238-42. criteria, endpoint definitions and toxicity criteria. Invest New Drugs 1992.
7. Hill MJ. Bile acids and colorcctal cancer: hypothesis. Ew J Cancer Prcv 10:239-53.
1991;l:Suppl 2x59-74. 23. Ritcnbaugh C, Aickin M, Tarcn D, et al. Use of a food frcqucnc)
8. Armstrong B, DoU R. En~ironmcntal factors and cancer incidence and qucsrionnairc to screen for dietary cligibiliy in a randomized cancer pm.
mortality in different countries, with special rcfcmncc to dietary practices. vention phase III tria!. Cancer Epidcmiol Biomarkcrs Prcv 1997;6:347-54.
Inc J Cancer 1975;15:617-31. 24. Prcvcncivc Scrviccs Task Force. Guide to clinical preventive scniccs:
9. La Vccchia C, Ncgri E, Dccarli A, et 21.A case-control study of diet report of the U.S. Prcvcntive Scrviccs Task Force. 2nd cd. Balrimorc:
and cola-rectal cancer in northern Italy. Int J Cancer 198&41:492-S. Wiiiams i?cWiins, 1996.
10. Tuyns AJ, Kaaks R, Hacltcrman M. Colorectal cancer and the con- 25. Winawer SJ,Flctchcr RH, Miller L, et al. Colorcctal cancer screening:
sumption of foods: a casecontrol study in Bc1gium.Nutr Cancer 1988;ll: clinical guidelines and rationale. Gasrroentcrology 1997;112:594-642. [Er.
189.204. rata, Gastrocntcrology 1997;112:fO60, 1998;114:625.]
11. Pctcrs RR, Garabrandt DH, Yu MC, Mack IX. A case-control study 26. Bycrs T, Lcvin B, Rothcnbcrgcr D, Dodd GD, Smith R4. American
of occupational and dietary factors in colorccrll cancer in young men by Cancer Society guidclincs for screening and surwillancc for early dctcction
subsitc. Cancer Rcs 1989;49:5459-68. of colorcctal polyps and cancer: update 1997. CA Cancer J Clin 1997,47
12. Bidoli E, Franccschi S, Tdaminl R, Barra S, La Vccchia C. Food con- 154-60.
sumotion and cancer of the colon and rectum in north-castcm It&,.I Int -J 27. hladncz ME, Marshall JR, Graver E, et al. Rcliabiiry and validity of a
cancer 1992;50:223-9. self-adrninistcrcd food frcqucncy qucstionnairc in a chcmoprcvcntion trial
13. Thun MJ, Callc EE, Namboodiri MM, et a[. Risk factors for fatal co- of adcnoma rccurrcncc. Cancer Epidcmiol Biomarkcrs Prcv 1999;8:941-6.
lon cancer in a large prospcctivc study. J Natl Cancer Inst 1992;84:1491- 28. Emerson SS, McGee DL, Fcnncrty B, Hion L, Garcwal H, Albcrts
500. D. Design and analysis of studies to rcducc the incidence of colon polyps.
14. Giovannucci E. Rimm EB, Stampfcr MJ, Coldits GA, Aschcrio A, Stat Mcd 1993;12:339-51.
Willctt WC. Intake offat, meat, and fiber in relation to risk ofcolon cancer 29. Digglc PJ, Lang K-Y, Zcgcr SL. Analysis oflongirudinal data. Oxford,
in men. Cancer Rcs 1994;54.2390-7. England: Clarcndon Press,1995.
15. Fuchs CS, Giovannucci EL, Colditz GA, ct al. Dietary fiber and the 30. Baron JA, Beach hi, Mandcl JS, et al. Calcium supplcmcnts for the
risk of colorcctal cancer and adcnoma in women. N Engl J Mcd 1999;340: prcvcncion of colorcctal adcnomas.N Engl J Med 1999;340:101-2
169-76. 31. Schaukin A, Lanza E, Code D, ct al. Lack of effect of a low-fat, high-
16. Plau EA, Giovannucci E, Rimm EB, ct al. Dietary fiber and distal co- fiber dicr on the recnrrcncc of colorcctal adcnomas.N Engl J hkd 2000;
lorcctal adcnoma in men. Cancer Epidcmiol Biomarkcrs l’rcv 1997;6:661- 342:1149-55.
70. 32. Rimm EB, Arch&o A, Giovannucci E, Spicgclman D, Stampfcr MJ,
17. Rcddy BS, Simi B, Engle A. Biochcmicll cpidcmiology of colon can- Willcrr WC. Vcgctablc, fruit, and cereal fiber intake and risk of coronary
cer: cffcct of types of dicrary fiber on colonic diacylglyccrols in women. heart discasc among mtn. JAM.& 1996;275:447-51.
Gastrocnterology 1994;106:883.9. 33. Walk A, Manson JE, Stampfcr MJ, ct al. Long-term inrakc of dietary
18. Albcrts DS, Einspahr J, Ritcnbaugh C, et al. The cffcct ofwhcat bran fiber and dccrcascd risk of coronary heart discascamong women. J-AMA
fiber and calcium supplementation on rc:tal mucosal proliferation rates in 1999;281:1998-2004.
patients with rcscctcd adcnomatous colorccral polyps. Cancer Epidcmiol 34. National Academy of Scicnccs.Diet, nutrition and cancer. Washing-
Biomarkcrs Prcv 1997,6:161-9. ton, D.C.: National Academy Press,1989.
19. MC&own-Eysscn GE, Bright-See E, Bruce WR, et al. A randomized 35. WHO Study Group on Diet, Nutrition and Prcvcntion of Noncom-
trial of low fat high fibrc diet in the rccurrcnce ofcolorcctal polyps. J Clin municable Discascs. Diet, nutrition, and the prcvcntion ofchronic diseases.
Epidcmiol 1994;47:525-36. [Erratum, J Clin Epidcmiol 1995;48:i.] Gcncva: World Health Organization, 1990.

1162 - April 20, 2000
The K;ew England Journal of Medicine

Editorials development of new polyps, but they are not a good
way to study the role of diet or nutrients in the later
stages of the evolution of adenomas to colorectal
cancer. In the context of the long course of this evo-
lution, the three- or four-year period assessedby &in-
DIET, COLORECTAL ADENOMAS, ical trials is very brieK6 All the studies conducted to
AND COLORECTAL CANCER date have been limited to this particular length of
time. It is therefore appropriate to question the rel-

B ECAUSE coiorectal cancers usually arise from
adenomatous polyps, it is believed that prevent-
ing the growth of adenomas in the colon and rectum
evance of these trials for the prevention of colorectal
cancer. Although adenomas are a risk factor for co-
lorectal cancer, most adenomas do not evolve into
or removing any that appear will prevent colorectal colorectal cancer, of course, and the clinical impor-
cancer.’ Many have therefore a\vaited the results of tance of small adenomas (those that are less than 1 cm
the tivo important trials published in this issue of in diameter), especially small tubular adenomas that
the Jorrrml- a study of Lvheat-bran fiber by Alberts contain neither villous histologic features nor areas
et aL2 and the Polyp Prevention TriaL3 Both trials of dysplasia, is not clear. The majority of the adeno-
are well conceived, \vell designed, well implemented, mas found during follow-up in adenoma-prevention
and clearly presented. What is disappointing, how- trials have been small, tubular adenomas without vil-
ever, is that the findings of both trials are negative. lous or dysplastic features. In the study by Alberts et
Three to four years of either taking a daily wheat-bran al., only 12.9 percent of the recurrent adenomas had
supplement or following a diet that Leaslow in fat villous features, and in the Polyp Prevention Trial,
and high in fruits and vegetables had no effect on the only 16.4 percent of the recurrent adenomas had vil-
incidence of new colorectal adenomas.2T3These find- lous features or dysplasia or were large. In a small
ings, considered alongside previous negative results Australian trial, the use of a cereal-fiber supplement
of trials that assessed the ability of other nutritional was associated with a marginally significant reduction
factors to prevent adenomas, lead to a clear conclu- in the incidence of larger adenomas,7 but neither of
sion. The relevance of these findings for the preven- the two current trials found any evidence of protec-
tion of colorectal cancer is less certain, however. tion against large or advanced adenomas. Although
Surveillance by colonoscopy afier the diagnosis of the numbers of cancers were small in the study bj
an adcnoma is a major clinical challenge.’ If an in- Alberts et al. and the Polyp Prevention Trial, it is dis-
te,rvention could be found that reduced the growth of appointing that in both trials the incidence of cancer
new adenomas, then colonoscopic surveillance could Lvas slightly higher in the intervention group than in
be less freqtient for people with a history of adeno- the control group (7 cases vs. 2 and 10 cases vs. 4, re-
ma. This approach would reduce the costs and in- spectively).
convenience of the procedure, while still lowering the The authors of both trials rightfilly concluded
risk of adenoma. Several groups have examined the that their findings cannot be interpreted as evidence
effect of various nutritional interventions on the risk that a high-fiber cereal supplement or a low-fat,
of new colorectal adenomas using the same clinical high-fiber diet is not effective in protecting against
model employed by Alberts et al. and the Polyp Pre- the later stages of development of colorectal cancer.
vention Trial.* In this model, patients xi,ho have had Short-term trials of this type are still reasonable for
an adenoma removed are randomly assigned to a nu- the assessment of treatments to prevent the growh
tritional intervention or to a control group, and the of neiv adenomas, but new designs are also needed
efficacy of the intervention is assessedat the time of to study the effects of nutritional and chemopreven-
the clinically indicated folloxv-up colonoscopy, one tive agents on the later stages of the development of
year later, three to four years later, or at both times. colorectal cancer.
Supplementation with vitamin C, vitamin E, beta car- Trials in bvhich adenomas are not removed present
otene, or cereal fiber or the adoption of a diet low in clinical and ethical problems, although the results of
fat and high in fruits and vegetables has sholvn no ef- one such study were recently reported, in which pol-
fect on the incidence of new adenomas.2-4 Calcium yps xvere left in place for three years to assessthe ef-
supplements n-ere somelrrhat effective, but they re- fects of a combination of nutrients on their gro&.’
duced the incidence of adenomas b!ponly 17 percent.5 No effects on growth were detected, though there
Although there may be other reasons to follow lots- wrere marginaUv fen-er nenp polyps among the subjects
fat, high-fiber diets or to take these supplements, it who were receiving a fixture of beta carotene, vita-
is clear that these interventions do not appreciably min C, vitamin E, selenium, and calcium supple-
reduce the rate of formation of next adenomas with- ments.* Perhaps agents like aspirin, selective inhibitors
in a period of three to four years - the standard of cyclooxygenase 2, selenium, and folate - alone
length of foIlon,-up in such studies. or in combination - will be found to be useM in
Clinical trials are a convenient ivay to study the the clinical management of adenomas.

1206 - April 20, 2000


.ot a good A different, though related, question is vvhether a REFERENCES
!-I the later low-fat, high-fiber diet will reduce the risk of colo- 1. Mrnawcr SJ, Flctchcr RH, Miller L, cc al. Colorccral cancer screening:
ColOrecta] rectal cancer. Findings from obsemadonal studies show clinical guidelines and rarionalc. Gas~ocn~crology 1997;112:594-642.
f this evo- that diets high in fruits and vegetables are associated [Errata, Gastrocnt~rology 1997,112:1060, 1993;114:625.]
2. Albcrts DS, hlartlncz hiE, Rex DJ, ct al. Lack ofcffcc: ofa hizhh-fiber
:d by clin- with lower risks of cancer at many sites, including ccrcal supplcmcnt on the recurrence oFcolorcctaJ adcnomas.KEna J Mcd
ducted to the colon and rectum.9J0 The idea that the intake of 2000;342:1156-62.
insoluble fiber alone explains population-based dif- 3. S&at&in A, Lanza E, Corlc D, ct al. Lack of cffccr of a low-fat, hi&-
length of fiber diet on the recurrence of colorccd adcnomas.S Engl J hfcd 2000;
n the rel- ferences in the risk of colorectal cancer may well 342:1149-55.
colorectal have been overly simplistic and incorrect. In fact, the 4. Bprs 7: Whar can random&d, controlled trials tell us about nutrition
and cancer p:cvcntion? CA Cancer J Clin 1999;49:353-61
or for co- amount of cereal fiber in the diet has not consistent- 5. Baron J-4, Beach hi, Handel JS, et al. Calcium supplcmcnts For the prc-
solve into ly been associated with the risk of colorectal cancer vention of colorcctal adcnomas. N Engl J hlcd 1999;340:101-7.
-al impor- in observational studies9J0 Moreover, cereal fiber tak- 6. Kinzlcr Ku; Vogelstcin 8. Lessons from hcrcdirary colorcctal cancer.
Cd 1996;87:159-70.
thanlcm en in the form and amounts used in the study by Al- 7. &facLeMm R, Macrac F, Bain C, et al. Randomized trial of intake of
3mas that berts et al. not only has no effect on the incidence far, fiber, and beta carofcnc to prevent colorcctal adcnomu. J Natl Cancer
of colorectal adenomas, but also causes gastrointes- Inst 1995;87:1760-6.
nor areas 8. Hofstad B, Almcndingcn K, Vatn hI, CCal. Growth and rccurrencc of
le adeno- tinal side effects. In observational studies, the evi- colorcctal polyps: a double-blind a-year intervention with calcium and an-
revention dence that a low-fat diet reduces the risk of colorec- tioxidants. Dig&on 1998;S9:148-56.
9. The American Cancer Society 1996 Advisory Commirrcc on Diet, Nu-
thout vil- tal cancer is also mixed, but a higher intake of fruits trition, and Cancer Prc~ntion. Guidelines on diet, nutrition, and cmccr
ilberts et and vegetables (especially vegetables) has been found prevention: reducing the risk of cancer with healthy Food choicer and phys-
omas had to be beneficial more consistently.9J0 ical activity. CA Cancer J Clin 1996;46:325-41.
10. World Cancer Rcscarch Fund. Food, nutrition and the prevention of
ion Trial, Observational studies around the world continue cancer: a global pcrspcctiw Washington, D.C.: American Institute for
.s had vil- to find that the risk of colorectal cancer is lower Cancer Rcscarch, 1997.
II a small among populations with high intakes of fruits and
pplement vegetables and that the risk changes on adoption of a 02000, hiassachusctts hlcdical Society.
.-eduction different diet, but we still do not understand why.9Jo
Ieither of It is unclear whether any single aspect of the diet -
jf protec- a particular vitamin, phytochemical, or dietary prac-
Wiough tice such as the method of cooking meats - accounts MISSED DIAGNOSES OF ACUTE
study by for this relation. Randomized, controlled trials are CORONARY SYNDROMES IN THE
, it is dis- commonly regarded as providing more definitive sup-
of cancer port for causal inferences than are observational stud- EMERGENCY ROOM - CONTINUING
pthanin ies, because they can control for confounding by the CHALLENGES
vs. 4, re- many factors related to the dietary behavior of inter-
est. Randomized, controlled trials can usually answer
only narrowly defined questions, however, and they
cannot easily assess the effects of the long-term di-
I? ATIENTS \vith suspected acute coronary syn-
dromes, or acute cardiac ischemia, account for
nearly 1.7 million hospital admissions per year in the
low-fat, etary patterns that have been shonn to be associated United States. Behveen 2 and 8 percent of patients
; against \vith a lower risk of colorectal cancer in observation- with myocardial infarction are mistakenly released
11cancer. al studies. This is clearly true of polyp-prevention from the emergency department.1-3 In addition to
Iable for trials, since such studies are particularly limited by the implications for patient care of the failure to di-
growth their short follow-up periods. agnose acute coronary syndromes, the threat of mal-
) needed Randomized, controlled trials have now sho\vn us practice suits is also of concern. An estimated 20
lpreven- that the use of some of the diets and nutritional sup- percent of the money awarded in malpractice suits
jment of plements thought to lolver the risk of colorectal can- against emergency department physicians is related to
cer has no short-term benefits with respect to prevent- the misdiagnosis and mistreatment of acute coronary
1present ing adenomas. There may be many reasons to eat a syndromes. Thus, it is not surprising that physicians
esults of diet that is lolv in fat and high in fiber, fruits, and in the emergency department tend to be cautious
uch pol- vegetables or to supplement the diet with a food high when making decisions about patients \vith chest
s the ef- in cereal fiber, but prel,enting colorectal adenomas, symptoms, admitting far more patients with possible
growth.8 at least for the first three to four years, is not one of acute coronary syndromes than are ultimately found
;h there them. With regard to questions about diet and co- to have them. Acute myocardial infarction or unsta-
subjects lorectal cancer, though, definitive answers still seem ble angina is confirmed in no more than 30 percent
ne, vita- to be beyond the reach of both observational epide- of patients xvho are admitted with suspected acute
supple- miologic studies and randomized, controlled trials. coronary syndromes. These potentially unnecessary
thibitors hospitalizations result in health care costs of more
- alone TIM BYERS, M.D., M.P.H. than $5 billion annually in the United States. There
fief%1 in Unix&y of Colorado School of Mcdicinc is considerable interest in increasing the efficiency of
Denw, CO 80262 health care delivery in order to reduce the number of

Volume 342 Number 16 * 1207
GASTROENTEROLOGY 2000;118:1235-1257

AGA Technical Review: Impact of Dietary Fiber on Colon
Cancer Occurrence

A GA Gorterning Gourd on Not ember 1.5, 1999.

olorectal cancer (CRC) is one of the most common sought, the relationship tends co be less convincing.”
C cancers in the developed world. In the United States,
CRC is the fourth most common cancer (after lung,
Therefore, these observations suggest char overall diet
and lifestyle, rather than individual factors, ply the more
prostate, and breast cancers) and the second most com- important role, thus underscoring the importance of as
mon cause of cancer death (after lung cancer).’ In 1998 yet undetermined inreractions among dietary compo-
alone, 131,600 new cases of CRC are expected to be nents and lifestyle factors in the development of CRC.
diagnosed, and an estimated 56,500 deaths will have Investigators have begun to recognize the need to
been caused by the disease.’ The deve!opment of CRC is elucidate a unifying mechanism by which these factors
thought to be the result of an intimate and still poorly modulate colorectal carcinogenesis7p8 and to examine
understood interplay between environmental and genetic combinations of dietary and lifestyle modifications in the
factors. Dietary and lifestyle factors are among the most prevention of CRC (e.g., the Polyp Prevention Trial).”
important environmental factors implicated.*g3 It has Dietary fiber is one of several factors whose role in
been estimated that 35% (lo%-70%) of all cancers are colorectal carcinogenesis has been extensively studied.
attributable to diet and that 50%--75% of CRC in the However, the precise nature and magnitude of the
United States may be preventable through dietary modi- relationship between fiber intake and CRC risk have not
fications.” U.S. CRC mortality rates among the white been clearly established. Dietary guidelines from the
population decreased by 29% from 1950 through 1990, American Cancer Society and the National Cancer Insti-
tute, which encourage healthy eating habits and lifestyle
with a more pronounced decrease in women than in
modifications, recommend that individuals eat more than
men.: This decrease is probably attributable to improved
5 servings of fresh vegetables and fruits and 20-30 g of
early detection as well as lifestyle and dietary changes.>
fiber per day. However, the validity of these recommenda-
A cause-and-effect relarionship between dietary or
tions has not been scrutinized rigorously.
other lifestyle factors and CRC is difficult to establish.
The objective of this technical review is a critical
Because of inherent limitations associated with study
analysis of currently available data from epidemiological
design, epidemiological, animal, and interventional stud-
and clinical studies in humans of dietary fiber’s effect on
ies examining this relationship have often produced
colorectal carcinogenesis. All human studies concerning
conflicting results. Therefore, the precise nature of the
CRC and its precursor, adenoma, and fiber, grains,
relationship of CRC with each nutrient or lifestyle factor
cereals, vegetables, or fruits published in the English
and the actual magnitude of the relationship are not clear. language from 1970 through 1999 were considered.
Among dietary factors implicated in colorectal carcinogen- These studies were found in the MEDLINE and CAN-
esis, consumption of red meat, animal and sacurated fat, CERLIT databases, in several extensive reviews,2*j.‘0-1’
refined carbohydrates, and alcohol, as well as total caloric and in references in the identified studies. This revien
(energy) intake, is believed to be positively related.‘,3 On emphasizes results from all the published prospective
the ocher hand, the intake of dietary fiber, vegerables, (cohort) epidemiological studies and randomized intenen-
fruits, antiosidant vitamins, calcium, and folate is nega- tion studies in humans. Seminal studies of a descriptive
tively associated with the development of CRC.*13 In
addition, obesity, increased body mass index, and a
sedentary lifestyle are associated with increased risk.‘s3 Abbreviations used in this paper: CAPP, Concerted Action Polypo-
sis Prevention trial; Cl, confidence interval; CRC, colorectal cancer;
There tends to be agreement among epidemiological FAP, familial adenomatous polyposis; HNPCC, hereditary nonpolypo-
studies regarding the risk of CRC and irs relarionship sis colorectal cancer; IGF, insulin-like growth factor; OR, odds ratio;
with overall diet and lifestyle.6 However, when many of RR, relative risk; SCFA, short-chain fatty acid.
0 2000 by the American Gastroenterological Association
the findings are examined closely and correlations be- 0016-5085/00/$10.00
tneen CRC and individual dietary and lifestyle factors are doi:10.1053/gast.2000.7927

and case-control epidemiological nature as well as previ- Table 1. Dietary Fiber and Related Compounds
ously published meta-analyses or pooled (combined) Classification
analyses of case-control studies are reviewed. Possible Nonstarch polysaccharides:
mechanisms by which dietary fiber may suppress colorec- Noncelluloses: hernicelluloses, pectins, gums, mucilages
tal carcinogenesis are also discussed. Nonpolysaccharides: lignins
Classification based on solubility
Soluble (highly fermented): pectins, gums, mucilages, some hemi-
Dietary Fiber: Definition, Sources, celluloses
Insoluble (poorly fermented): celluloses, lignins, most hemicellu-
and Consumption loses
Minor components
Because the term riietnvy fi6ev encompasses a wide
Phytates, cutins, saponins, lectins, protein, waxes, silicon
range ofcomplex materials, it is difficult to define. There Related components
is no internarionally accepted definition or method for Resistant starch and protein
determining the dietary fiber content of foods.lS Dietary Lignans

fiber was initially defined by Burkitt and Trowell’6 as the
complex carbohydrate in the diet from plant sources that be in the range of 12-15 g/day.23 Almost all of the
escapes small bowel digestion and thus reaches the colon. epidemiological studies used these assays to estimate the
The U.S. Experr Panel on Dierary Fiber defined dierary dietary fiber intake. Currently available assays rhat
fiber as the endogenous components of plant materials in account for resistant starch (in the range of 3-5 g/day in
the diet that are resistant co digestion by enzymes the North American diet24) estimate the amount of
produced by humans.” Analytically, dietary fiber is polysaccharides that reach the colon to be in the region of
composed predominantly of nonstarch polysaccharides 15-25 g/day. Even wirh the inclusion of resistant starch
and nonpolysaccharides (mainly lignins).ls Nonstarch in the assessment of dietary fiber intake, currently
polysaccharides include cellulose and noncellulosic poly- available assays account for less than one third of total
saccharides (e.g., hemicelluloses, pectins, gums, and dietary fiber char reaches the colon to sustain the known
mucilages). l8 This definirion excludes other substances in rate of colonic bacterial synrhesis.“*” Therefore, al-
the plant materials such as phycates, cucins, saponins, though the assays chat are currently available to estimate
lectins, proteins, waxes, silicon, and other organic constitu- nonstarch polysaccharides and resistant starch are very
ents.18 Dietary fiber can be further analytically classified precise, they do not accurately measure dietary intake and
as soluble (some hemicelluloses, pectins, gums, and seriously underestimate the amount of dietary fiber that
mucilages) and insoluble (most hemicelluloses, cellu- reaches the colon and is available to participate in the
loses, and lignins), depending on its solubility in water mechanisms of CRC prevention.“~” Because of these
and buffer solution.‘s W’hen the effect of dietary fiber on difficulties, epidemiologists began to study the relation-
the cclon is being considered, the classification of fiber as ship between intake of “fiber-rich” foods (e.g., cereals,
fermenrable (i.e., metabolized by colonic bacteria) and fruits, and vegetables) and the risk of CRC; most of these
nonfermentable is also useful. It became apparent in the studies suggested a strong inverse relationship.“~‘”
early l!XOs that some starch escapes small bowel diges- Dietary fiber is found mainly in vegetables, fruits,
cion and reaches the colon. Stephen et a1.19esrimated that grains, seeds, nuts, and legumes. In the Second National
5%--10% of dierary srarch reaches the colon and called Health and Nutrition Examination Survey (NHANES II,
this “resistant starch.” Dietary fiber and related com- 1976-1980), mean dietary fiber intake in the U.S. adult
pounds are summarized in Table 1. population (> 19 years old) is 11.1 g/day or 13.3 g/day,
The development of new analyrical methods’*-*’ to depending on the methodology used.2j On any given day,
esrimate the dierary fiber conrenr of foods allowed 509 of the U.S. population reports a dietary fiber intake
epidemiological studies to better define the relationship of < 10 g/day, and only - 10% consume >20 g/day.23 On
berneen the intake of dietary fiber and the risk of CRC. a per IOOO-kcal basis, women consume more dietary fiber
However, these assays still underestimated the actual (6.5 g/1000 kcal) than men (5.5 g/1000 kcal) at every
dierary fiber conrent in foods. They20-22 are based on the age. 23 Both men and women show the same pattern of
assumption char all starch is digested in the small bowel increasing dietary fiber intake with age when fiber is
and that other complex carbohydrates are completely examined in relation ro total caloric intake.23 A marked
undegraded; dietary fiber is therefore considered to racial effect is evident; blacks having lower dietary fiber
include all pianr polysaccharides except srarch and non- intake than whites in both sexesand all age groups.23 It is
polysaccharides. These assays estimated the amount of uncertain at present wherher mean dietary fiber intake in
nonstarch polysaccharides in the North American diet to the U.S. adult population has significantly increased

since the completion of the second NHANES. This issue prospective studies involving repeated assessment of diet
is being analyzed from the recently completed NHANES III among tens of thousands ofsubjects are feasible.
(198%1994), which included 40,000 noninstiturionalized
people aged 22 months and oversampled blacks, Mexican
Correlation Studies
Americans, children, and elderly people.23Table 2 lists the Correlation studies examine the relationship be-
fiber concenc of frequently consumed fruits and vegetables. tween the per capita consumption of a dietary factor and
the prevalence, incidence, or mortality of CRC in the
population. Correlation studies can examine this relation-
Epidemiological Evidence ship among populations residing in differenr countries or
The “fiber hypothesis” was first introduced when among different groups within a country either ac a given
Burkitt’“n2’ recognized the rarity of CRC in most African time or over a certain period (i.e., time-trend). They
popularions and was impressed by the high fiber and low provide provocative initial evidence chat a particular
refined carbohydrate content of the diet in Africa and dietary factor has a role in the development of CRC and
other underdeveloped areas of the world. Since then, this hence are considered worthy only of hypothesis forma-
purported inverse relationship between dietary fiber tion. Of 28 published international, within-country
incake and the risk of CRC has been investigated by 3 correlation and time-trend studies of CRC and fiber,
types of human epidemiological studies: correlation (or vegetables, grains, fruits, and cereals, 23 (82%) showed
ecological), case-control, and prospective studies. either a strong or a moderace protective effect of dietary
in nutritional epidcmiological studies, dietary intake fiber or “fiber-rich foods” or equivocal results that were
of certain nutritional Lctors is assessedby several meth- nevertheless consistent with the fiber hypochesis.2~3*‘~1’
ods. In the 24-hour recall method, the basis of most Four studies found no evidence for a protective effect of
national surveys, subjects are asked to report their food fiber, and 1 study showed a significant excess risk of CRC
intake during the previous day. This method has the associated with high intake of fiber-rich foods.2*3*‘o-14
advantage of requiring no training or literacy and The limitations of interpretation of data generated from
minimal effort by the participant. The mosr serious these correlation studies are many. The older incerna-
limitation is that dietary intake is highly variable from tional studies are based on intake of crude fiber, which
day co day. In the diet recording (food diary) method, greatly underestimates total dietary fiber levels. Further-
detailed meal-by-meal records are kept of the types and more, correlation studies often fail co correct for unmea-
quantities of food and drink consumed during a specified sured confounding factors that may be responsible for the
period, typically 3-7 days. This method places a consider- observed association. They also do not control for other
able burden on the subject, thus limiting it to literate dietary variables or for any of the other known risk factors
people who are also highly motivated. The effort involved associated with CRC. Despite these shortcomings, it is
in keeping diet records may increase consciousness of food remarkable that most of these correlation studies have
intake .and encourage alteration of diet. However, the indicated a strong inverse relationship between dietary
advantages of the diet recording method are that it does fiber intake and the risk of CRC.
not depend on memory and allows direct measurement of
portion sizes. Dietary records reduce the problem of
Case-Control Studies
day-to-day variation by taking the average of a number of Case-control studies compare prior consumption
days; in addition, weekday/weekend variability, which in of a dietary factor in subjects with CRC and matched
some societies is high, can be accounted for. Short-term control subjects without CRC. Many of the weaknesses of
recall and dietary recording methods are generally expen- correlation studies can be avoided in case-control studies.
sive, may be unrepresentative of usual intake, and are Known or suspected potential confounding filccors can be
inappropriate for assessmenr of diet history. For these controlled or eliminated in the study design or controlled
reasons, many investigators now use food frequency in the data analysis. The most serious limicacion in
questionnaires, which include a food list and a frequency retrospective studies is the accuracy with which intake of
response section for subjects to report how often each food dietary factors or supplemencarion can be established.
is eaten. Diets tend co be reasonably well correlated from Individuals may misreport their habitual past diets; if
year co year, and subjects are usually asked to describe cases and controls differ in the accuracy of their dietary
how frequently they ate each food in the preceding year. recall, the ensuing comparison will be biased. In addi-
Food frequency questionnaires are easy for literate sub- tion, some individual aspects of diet, especially nucrienc
jects to complete, often as self-administered forms. content, may not vary greatly within a population, so
Processing is readily computerized and inexpensive; even case-control studies may nor show wide ranges of cancer

Table 2 (continued). Provisional Dietary Fiber Table
Table 2. Provisional Dietary Fiber Table
Fiber Fiber
Fiber Fiber
Food (&!/~00sl” ( g/serving)
Food (li!/lOOgY (g/serving)
Fruits Baked beans, tomato 7.3 18.6/l cup
Apple (without skin) 2.1 2.9/l medium-sized apple
Apple (with skin) 2.5 3.5/l medium-sized apple
Dried peas, cooked 4.7 4.7/half cup (cooked)
Apricot (fresh) 1.7 1.8/3 apricots
Kidney beans, cooked 7.9 7.4/half cup (cooked)
Apricot (dried) 8.1 10.5/l cup
Lima beans, cooked/ 5.4 2.6/half cup (cooked)
Banana 2.1 2.5/l banana
Blueberries 2.7 3.9/l cup
Lentils, cooked 3.7 1.9/half cup (cooked)
Cantaloupe 1.0 2.7/half edible portion
Navy beans, cooked 6.3 3.l/half cup (cooked)
Cherries, sweet 1.2 1.2/15 cherries
Breads, pastas, and flours
Dates 7.6 13.5/l cup (chopped)
Bagels 1.1 l.l/half bagel
Grapefruit 1.3 l.G/half edible portion
Bran muffins 6.3 6.3/muffin
Grapes 1.3 2.6/10 grapes
Cracked wheat 4.1 4.l/slice
Oranges 2.0 2.6/l orange
Crisp bread, rye 14.9
Peach (with skin) 2.1 2.1/l peach
Crisp bread, wheat 12.9
Peach (without skin) 1.4 1.4/l peach
French bread 2.0 0.67/slice
Pear (with skin) 2.8 4.6/l pear
Italian bread 1.0 0.33/slice
Pear (without skin) 2.3 3.8/l pear
Mixed grains 3.7
Pineapple 1.4 2.2/l cup (diced)
Oatmeal 2.2 5.3/l cup
Plums, damsons 1.7 1.7/3 plums
Pita bread (5 inches) 0.9
Prunes 11.9 11.9/11 dried prunes
Pumpernickel bread 3.2 l.O/slice
Raisins 8.7 2.2/packet
Raisin bread 2.2 0.55/slice
Raspberries 5.1 6.3/l cup
White bread 2.2 0.55/sIice
Strawberries 2.0 3.0/l cup
Whole-wheat bread 5.7 1.66/slice
Watermelon 0.3 1.3/4 X 8-inch wedge
Pasta and rice-cooked
Juices Macaroni 0.8 1.0/l cup (cooked)
Apple 0.3 0.74/l cup
Rice, brown 1.2 2.4/l cup (cooked)
Grapefruit 0.4 1.0/l cup
Rice, polished 0.3 0.6/l cup (cooked)
Grape 0.5 1.3/l cup
Spaghetti (regular) 0.8 LO/l cup (cooked)
Orange 0.4 LO/l cup
Spaghetti (whole wheat) 2.8 3.0/l cup (cooked)
Papaya 0.6 1.5/l cup
Flours and grains
Bran, corn 62.2 18.7/oz
Bran, oat 27.8 8.3/oz
Asparagus, cut 1.5 1.5/7 spears
Bran, wheat 41.2 12.4/oz
Beans, string, green 2.6 3.4/l cup
Rolled oats 5.7 13.7/l cup (cooked)
Broccoli 2.8 5.0/l stalk
Rye flour (72%) 4.5 5.2/l cup
Brussels sprouts 3.0 4.6/7-8 sprouts
Rye flour (100%) 12.8 15.4/l cup
Cabbage, red 2.0 2.9/l cup (cooked)
Wheat flour
Cabbage, white 2.0 2.9/l cup (cooked)
Whole meal (100%) 8.9 10.6/l cup
Carrots 3.0 4.6/l cup
Brown (85%) 7.3 8.8/l cup
Cauliflower 1.7 2.1/l cup
White (72%) 2.9 2.9/l cup
Corn, canned 2.8 4.5/l cup
Kale leaves 2.6 2.9/l cup (cooked)
Almonds 7.2 3.6/half cup (slivered)
Parsnip 3.5 5.4/l cup (cooked)
Peanuts 8.1 11.7/l cup
Peas 4.5 7.2/l cup (cooked)
Filberts 6.0 2.8/half cup
Potato (without skin) 1.0 1.4/l boiled
Potato (with skin) 1.7 2.3/l boiled =Dietary fiber values are averages compiled from literature sources.
Spinach 2.3 4.1/l cup (raw) Adapted and reprinted with permission.23
Squash, summer 1.6 3.4/l cup (cooked, diced)
Sweet potatoes 2.4 ‘2.7/l baked (5 X 2
inches) risk nrithin that population. Another common problem is
Turnip 2.2 3.4/l cup (cooked, diced) that controls are often people with another disease,
Zucchini 2.0 4.2/l cup (cooked, diced)
because hospital patients are convenient subjects to
Bean sprout, soy 2.6 2.6/l cup study; their disease might also be diet related. In such
Celery, diced 1.5 3.7/l large stalk sicuacions, the study results could be seriously biased and
Cucumber 0.8 0.2/6-8 slices with skin
2.0/l wedge iceberg
often may not show any clear difference between casesand
Lettuce, sliced 1.5
Mushrooms, sliced 2.5 0.8/half cup (sliced) controls. For such reasons, the results of case-control
Onions, sliced 1.3 1.3/l cup studies of diet and cancer are sometimes inconsistent.
Peppers, green, sliced 1.3 LO/l pod
Another problem associated with case-control studies is
Tomato 1.5 1.8/l tomato
Spinach 4.0 8.0/l cup (chopped) selection bias-because of the absence of patients who do
not survive long enough to be enrolled in the study.,

Case-control studies may produce evidence that is signifi- ent procedures for coding and analyzing data that ma)-
cant in isolation. Such evidence is strengthened by have been used in the respective original analyses.
corroboration in additional studies conducted in a num- In this pooled analysis, the risk of CRC was shown to
ber of centers and particularly by consistent results from decrease incrementally as dietary fiber intake increased,
populations with different patterns of diet and of cancer. with ORs of 1.0, 0.79, 0.69, 0.63, and 0.53 for each
Three analyses, conducted in combined analysis or quintile of consumption from the lowest to highest (P
meta-analysis formats, have critically evaluated the bulk trend < 0.0001).‘9 Consumption of more than 31 g of
of case-control studies that address the role of dietary fiber per day was associated with a 47% reduction in the
fiber in CRC.28-30 Track et al.2s analyzed 23 case-control risk of CRC compared with diets incorporating less than
studies that examined the relationship between CRC and 10 g of fiber per day (95% CI, 0.47-0.61).z9 The strong
consumption of fiber and vegetables. Fifteen (65%) of inverse association observed with fiber intake was not
these studies demonstrated either a strong or moderate affected by adjustment for total energy intake, age, ses,
protective effect of dietary fiber and vegetables.** Six height, weight, body mass index, and other potential
studies (26%) showed equivocally prorective effects of confounding factors, including vitamin C and p-caro-
fiber that were not statistically significant, that became tene.29 When the consistency of the fiber eftect across the
nonsignificant after adjustment, or char could not be studies was examined by calculating the relative risk of
distinguished from other factors in their relation to developing CRC in subjects consuming 27 g fiber per day
risk.** Only 2 studies (9%) lacked support for a protec- compared with those consuming less than 11 g fiber per
tive effect of fiber.2s Track et al.2s performed a mera- day in individual srudies, 12 of the 13 studies showed
analysis on 16 case-control studies that provided enough inverse associations with dietary fiber.‘9 In S of these 12
data in the published articles. YVith fiber-rich diets (i.e., studies, the decrease in risk was statistically significant.‘9
combinarion of fiber and vegetables), a 43% reduction in \Vhen all of the studies were combined and adjusted for
CRC risk was observed (odds ratio [OR], 0.57; 95% total energy intake, age, and sex, individuals who
confidence interval ICI}, 0.50-0.64) when the highest consumed 27 g fiber per day had a 50% reduction in the
and lowest quartiles of intake were compared.2s The risk of developing CRC compared with those who
extent of risk reduction based on begecable consumption consumed less than 11 g fiber per day (relative risk {RR},
was 52% (OR, 0.48; 95% CI, 0.41~0.57), whereas one 0.51; 9S% CI, 0.44-0.59).‘9 Estimates of RR per 27 g
based on an estimate of fiber intake was 42% (OR, 0.5s; fiber per day--estimated separately for cases of left-sided
95%, 0.5 l-0.66).*” The data did not permit discrimina- (RR, 0.52; 95% CI, 0.42-0.65) and right-sided (RR,
tion between effects ascribable to the fiber and the 0.45; 95% CI, 0.33-0.61) colon cancer and for rectal
nonfiber components of vegetables.2s cancer (RR, 0.43; 95% CI, 0.34-0.56), for women (RR,
Howe et a1.29 performed a combined (or pooled) 0.60; 95% CI, 0.4S-0.7s) and men (RR, 0.44; 95% CI
analysis of data from 13 case-control studies previously 0.37-0.53), and for 2 age groups (<50 years [RR, 0.66;
conducted in populations from North America, Europe, 95% CI, 0.43-0.991 and 250 years {RR, 0.49; 95% CI,
Asia, Australia, and South America with differing CRC 0.42-0.57}&--were consistent for all subgroups.29
rates and dietary practices. The individual data records In the original pooled analysis by Howe et a1.,29it was
for 5287 case subjects and 10,470 control subjects were assumed that a pooled estimate could be made of the
pooled for a common analysis. This approach thus differs heterogeneous results for dietary fiber and CRC risk. This
from the usual meta-analysis, in which estimates of risk heterogeneity was not examined, nor was the influence of
are pooled from published summary results. Pooled study quality considered. Therefore, Friedenreich et aL3’
analyses provide several benefits over meta-analyses of examined the study design features and dara collection
published results or narrative reviews of the literature. methods from the 13 case-control studies char had been
YVhen the actual individual subject level data from included in the original pooled analysis*” to determine
several studies are combined, detailed analyses are pos- whether they influenced the results obrained from a
sible. Because the pooled data sets constitute a large body pooled analysis.30 Friedenreich et a1.30assessed methods
of data, rare exposures can .be studied. Furthermore, the used in each study, estimated a quality score, and used a
consistency of the association across studies can be different model (a random-effects model rather than a
examined, confounding and interaction of several puta- fixed-effects model) to re-estimate the pooled OR for the
tive risk factors can be assessed, and previously unrecog- association between dietary fiber and CRC for these
nized or poorly established associations may be revealed. data.jO Key features of the methods used in each study
Using the individual data records has the advantage of and the quality score were examined in a random-effects
eliminating artifactual differences attributable to differ- model to determine whether the heterogeneity found

between study-specific risk estimates could be explained analysis of case-control studies show a significant inverse
by these variables.3” The OR for dietary fiber and CRC dose-dependent relationship berween dietary fiber intake
was 0.46 (95% CI, 0.34-0.64) for the 13 case-control and the risk of CRC and colorecral adenomas.29*36*37
studies as estimated using a random-effects model,6“ However, several shortcomings associated with case-
which was slightly lower than the OR estimated with the control studies limit interpretation of the results of these
fixed-effects model in the original pooled analysis (0.5 1).30 studies. Some of the most serious problems are that a
Two factors, whether the diet questionnaire had been large proportion of the published case-controls used
validated before use in the case-control study and nherher dietary tools, including questionnaires, that had nor been
qualitative data on dietary habits and cooking methods validated before use and that these studies did not
had been incorporated into the nutrient estimation, incorporate qualitative data on dietary habits and cook-
explained some of the heterogeneity in study resu1ts.j’ ing methods into the nutrient estimation. Furthermore,
Risk esrimares for dietary fiber and CRC were closer to because of the limitations associated with analytical
the null (i.e., 1.0) for studies with these 2 characteris- methods of determining fiber concenr in diet, as previ-
tics. so These invescigarors performed another pooled ously described, the accuracy of estimates of dietary fiber
analysis of the 13 case-control studies included in the in these studies is in question. Another problem is that
original pooled analysis and 4 additional case-control potential confounders were nor adequately controlled or
studies either excluded from or published after the corrected in some of these studies. Finally, ir is difficult to
original analysis. 3o Subjects consuming >27 g fiber per delineate the effect associated with dietary fiber from
day had a 50% reduction in the risk of developing CRC ocher potential ancicarcinogens present in fiber-rich foods
compared with those taking < 11 g fiber per day (OR, such as vegetables, fruits, cereals, and grains in case-
0.49; 95% CI, 0.37-0.65).j” control studies.
Colonic adenomas are well-established precursors of
Prospective Studies
adenocarcinoma.3’ Several case-control studies have also
found an inverse relationship between dietary fiber or Prospective (or cohort) studies assessthe diets of a
fiber-rich foods and the risk of colonic adenomas, thereby large group of healthy individuals and include follow-up
supporting the association observed with CRC.31-37 The over time, during which a number of cohort members
magnitude of the reduction in the risk ranged from 10% will develop CRC. The relationship of CRC to specific
to 60%, in these studies.3’-37 Some of these srudies also characteristics of individual diets is then analyzed. Pro-
showed a dose-dependent inverse association between spective studies avoid most of the methodological prob-
colorectal adenoma risk and dietary incake of fiber.36*37In lems of other epidemiological srudies and can control and
some studies, the protective effect associated with dietary correct confounding factors more adequately than correla-
fiber was evident only in womenii*3G and for large (>l tion and case-control studies. They also provide the
cm) aden0mas.j’ However, these studies are limited by opporrunicy to obtain repeated assessments of diet at
smail sample size. regular intervals, thus improving the validity of indi-
In summary, most of the published case-control studies vidual dietary assessment. Because of the prospective
show either a strong or a moderate protective effect of design, in which diet is assessedbefore the occurrence of
dietary fiber or fiber-rich foods or equivocal results rhac CRC, there is lirtle likelihood of selection or recording
are nevertheless consistent n-ich the fiber hypothesis. bias in cohort studies.
Three analyses of case-control studies, conducted in Earlier prospective studies investigated the relation-
combined analysis or meca-analysis formats, also provide ship becn*een dietary fiber intake and CRC mortalicy. A
strong support for rhe proceccive effect of dietary fiber or large cohort study from Japan was designed to investigate
fiber-rich foods on coloreccal carcinogenesis.2”-30 The the relationship between diet and ocher lifestyle variables
scrongesc argument for the fiber hypothesis that can be and major causes of deaths in 265,118 subjects, aged
made from case-control studies is rhe remarkable consis- 240 years, followed up for 13 years. During the 13-year
tency of rhe protective effect of dietary fiber among follow-up period, 39,127 people died.js Standardized
studies conducted iri populations with different patterns mortality rates for each cause of death were calculated
of diet and CRC. The combined analysis and meta- according ro the lifestyle variables that were studied
analyses of case-control studies suggest, on average, a when the subjects were still healthy at the time of the
50% reduction in the risk of developing CRC in initial incerview.3’ This study obsemed that CRC morcal-
individuals with the highest dietary fiber intake com- icy rate decreased as rice and wheat consumption increased
pared with those with the lowest fiber intake.2s30 Most with an RR of 0.6 in chose with the highest intake (>720
of the positive case-control studies and one combined cm3 of rice and wheat per day) compared with those with

the lowest intake (<lSO cm3iday).3s A Dutch study differentiated from chose chat affect incidence. In addi-
involving S71 middle-aged men followed up for 10 years tion, the study participants were, on average, more
showed a 3-fold reduction in cancer mortality in men in educated and affluent than rhe U.S. population as a
the highest quincile of dietary fiber intake compared with whole. Greater access to medical care and screening
men in the lowest quintile. 39 The 44 men who died of might have contributed to their lower mortality rates
cancer during 10 years of follow-up ate less dietary fiber from colon cancer. Therefore, generalization of the find-
(30.9 i 9.7 vs. 27.0 + 6.9 g/day; P = 0.001) and ings of this study to groups of lower socioeconomic status
polysaccharides (206.3 + 66.1 vs. 183.2 -+ 51.5 g/day; is questionable.
P = 0.006) than survivors.39 However, the number of Recently, several well-designed and -conducred prospec-
men who died of colon cancer in this study during 10 tive studies have examined the relationship between
years of follow-up was too small to allow statistical dietary fiber intake and the risk of CRC and adenomas,
analysis.39 Anocher study involving 25,493 white Califor- but the findings of these studies are not consistent (Table
nia Seventh-Day Adventists followed up for 21 years 3).ljls In the Nurses’ Health Scudy,43 121,700 female
showed no protective effect of cereal or green salad intake registered nurses between 34 and 59 years of age in the
on CRC mortality. 4o However, in the Seventh-Day Adven- United States completed a mailed quesrionnaire on
tist population, distribution of dietary fiber may be known and suspected risk factors for cancer and cardiovas-
narrow; therefore, protective effects of fiber may not be cular disease in 1976. Every 2 years thereafter, follow-up
observed. Anocher prospective study examined the risk of questionnaires were sent to identify new cases of cancer
developing colonic adenomas in 163 Hawaii Japanese and cardiovascular disease. In 1950, the questionnaire
autopsy subjects. 41 They constituted a subset of SO06 was expanded to include an assessment of diet, the
men originally examined between 1965 and 196s and Willett semiquantitative food ‘frequency quesrionnaire49;
those who died between 1969 and 19S4.4’ No significant 88,751 women were available for analysis in 1986,
differences were observed between subjects with and representing 6 years of follow-up. Among these women,
without adenomas in intake of dietary carbohydraces.41 150 cases of CRC were identified and confirmed. Energy-
Data from Cancer Prevention Study II, an ongoing adjusted inrakes ofcrude and total dietary fiber were both
prospective mortality srudy, support the protective role of inversely associated with the risk of colon cancer, but
dietary fiber in colorectal carcinogenesis4’; l,lS5,125 these trends were not statistically significant.43 When
men and women (average age, 57 years) in the United intake of fiber from fruit, vegetables, and cereals were
States completed a 4-page questionnaire in 1952 on their analyzed individually, only fiber from fruit was associated
diet, smoking history, alcohol intake, physical activity, with any appreciable reduction in risk, and the overall
height, weight, medication use, medical illnesses, farnil) trend was nor statistically significant.43 Some limitations
history of cancer, and other characteristics. The partici- of this well-designed prospective study are relatively
pants’ vital status was determined at 2-year intervals short follow-up (6 years) and uncontrolled confounding
through personal inquiries by the volunteers. Mortality factors (e.g., family history of CRC, physical activity) that
follow-up was assessed through 19SS. By this time, might affect the development of CRC.
79,S20 participants (6.7%) had died (1150 of CRC); The observations from this study extended for 16 years
94.2% of participants’ causes of death were determined of follow-up (19SO-1996) have recenrly been pub-
unequivocally. Dietary questions asked about the average lished.4’ During the 16 years of follow-up, 7S7 new cases
consumption of 32 food items and 10 beverages. Ivlulrivar- of CRC were identified and confirmed among the 88,757
iate analyses showed chat risk of fatal colon cancer eligible women. Afrer adjusrmenc for age, coral energy
decreased with more frequent consumption of vegetables intake, and most of the established risk factors for CRC in
and high-fiber grains (P trend = 0.031 in men and a multivariace model, total dierary fiber intake was not
0.0012 in women) after adjustment of confounding significantly associared with the incidence of CRC; the
facrors.4’ The RR for the highest vs. lowest quinrile of relative risk for the quincile group with the highest
vegetable and high-fiber grains was 0.76 in men (95% (median 24.9 g/day) compared with the lowest (median
CI, 0.57-1.02) and 0.62 in women (95% CI, 0.45- 9.8 g/day) coral dietary fiber intake was 0.95 (95% CI,
0.S6).4’ The strengths of this study are its size and 0.73-1.25), and no dose-dependent inverse association
prospective design. Its limitations include its dependence was observed (P trend = 0.59).41 No protective effect of
on a single, brief, self-administered questionnaire, lack of total fiber incake was obsemed when events during the
information on colon subsire, and relatively short fol- first 6 years of follow-up were excluded to examine the
low-up (6 years). Because of the reliance on mortality, possibility that total dietary fiber influences the risk of
facrors that influence survival could nor be clearly CRC only after several years or when the analysis was

Table 3. >2%1 65~ i’.tzj& 2nd Risk of CRC and Adenoma: Summary Of PlOSpeCtiVe Studies
Duration highest
Case/ of Highest Lowest vs. Pfor
-‘,**‘!r,r Case control follow-up intake intake lowest 95% inverse
stuo, dlagr,osis I-IO. ( g/day) (g/day) intake Cl association Comments
_-. /G (Yr)
‘drses 1-g Cr,lO, Female 15O/ 6 ~21.3 (total CU.6 0.90 0.54-1.49 0.70 No effect from
Hea!rr. : Ij99, cancer 88601 fiber) crude, fruit.
Stud/‘: vegetable,
and cereal
‘.clrses lL->c CRC Female 787/ 16 24.9 median 9.8 0.95 0.73-1.25 0.59 Fro effect from
Healtr, ‘2 593, 87970 (total fiber) median cereal and
Stud,+ fruit fibers:
risk with veg-
etable fiber
(P= 0.004)
Left colon 14 24.9 median 9.8 0.91 0.71-1.16 0.36 No effect from
Female 1012/
(total fiber) median cereal, fruit,
and rectal 26518
adenoma and veg-
etable fibers
Health PK.&, Uin Co!on 6 32.8 median 14.2 1.08 0.68-1.70 0.12 No effect from
Male 205/
sionals 119941 cancer 47744 (total fiber) median crude, fruit,
Follow-ug vegetable,
Studf: and cereal
Health Prof+$. USA Left colon 2 228.3 (total (16.6 0.36 0.22-0.60 <0.0001 Total fiber
Male 170/
sionals 119921 fiber) 0.27 0.16-0.45 <0.0001 Crude fiber
and rectal 7114
Follow-up adenoma 0.53 0.28-1.02 0.02 Fruit fiber
Study@ 0.53 0.30-1.01 0.003 Vegetable fiber
0.44 0.26-0.76 <O.OOl Grains fiber
Health Profr:s- USA Left colon Male 690/ 8 32.3 median 11.6 0.88 0.59-1.31 0.10 No effect from
sionals (1997) (total fiber) median vegetable,
and rectal 15758
Followup adenoma 8.4 (fruit fiber) 1.3 0.81 0.59-1.11, 0.03 cereal,
Study-‘7 3.4 0.69 0.46-1.03 0.007 wheat, crucif-
erous veg-
etable, and
legume fiber
9.4 (soluble No effect from
fiber) insoluble
Iowa USA Colon Female 212/ 4 >24.7 (total <14.5 0.80 0.49-1.31 NS
Women’5 (1994) cancer 35004 fiber)
US, not significant.

limited to wCJmCl1 who maintained a consistent level of rectum was also investigated among 27,530 women who
dietary fil)ur irrt;rkc during the study period.4i Cereal and reported undergoing colonoscopy or sigmoidoscopy be-
fruit fihcr wits not associated wirh any appreciable tween 19SO and 199-k. ’ 4-1There was no reduction in the
reductinn in (:I~<: risk, whereas greater consumprion of risk of coloreccal adenomas with increasing dietary intake
vegerabk I‘i!)cr MS associated with a significant increase of total, cereal, fruit, or vegetable fiber.4’
in the ri\k of‘(:RC (RR, I .35 in the highest [median 10.0 The same group of investigators also examined the
g/day} c(lmp:rrcd with the lowest [median 2.7 g/day) relationship between dierary fiber and the risk of colon
quintilr; 95% (I, 1.05-1.72; P trend for a dose-dependent cancer in men. ” The Health Professional Follow-UP
relationsllip = 0.004). M However, this adverse effect was Study is a prospective study of heart disease and cancer
no &or oOscrvcd when the analysis excluded women among 51,529 U.S. male health professionals between
who altered their fiber intake during rhe follow-up period the ages of 40 and 75 years who completed the original
(RR, 1.22; 95% Cl, 0.50-2.98; P trend for a relation- questionnaire in 1986. Again, dietary intake was assessed
ship = ().39).‘-’ The relationship berween fiber inrake and using the Willett semiquantitative food frequency ques-
he risk d ;t~Ic’no~~~;~s in the left side of the colon and the cionnaire.49 Among 47,949 men who were free of

diagnosed cancer in 1786, 205 new cases of colon cancer of fruit fiber (P trend = 0.03) and the association with
were diagnosed and confirmed between 19S6 and 1772. coca1 fiber intake became nonsignificant (P trend =
Analyses were performed in a similar fashion as in the 0.10). “‘I A strongly decreasing risk of distal colon
Nurses’ Health Study. 43~41Age, family history of CRC, adenomas was observed for soluble fiber (P trend =
obesiry, physical activity, cigarette use, alcohol consump- 0.0003) but not for insoluble fiber (P trend = 0.34) in
tion, and other confounding factors were adjusted for basic models.4’ In the full mulcivariace model, the strong
analysis. No clear association between total dietary fiber inverse association between intake of soluble fiber and
intake and risk of colon cancer was observed; the RR for distal colon adenomas persisted (P trend = 0.007).4’ No
the highest (median 32.S g/day) compared with the consistent relationship between fiber and rectal adenomas
lowest (median 14.2 g/day) quintile group with respect was observed in this study.4’
to total dietary fiber intake was 1.0s (95% CI, O.GS- In the earlier report of this study, increased total
l.i’O), and no dose-dependent inverse association was dietary fiber was strongly associated with a decreased risk
observed (P trend = 0.12).45 No significant protective of total colorecral adenomas (P < O.OOOl), and fiber from
effect was obsenred for total crude, fruit, vegetable, or vegetables, fruits, and grains was beneficial.“” However,
cereal fiber.45 in the updated report, coca1 dietary fiber was only
The same group of investigators also examined the modestly inversely associated with risk of total colorectal
relationship between dietary intake of fiber and the risk adenomas, and only fiber from fruits (not from vegetables
of colorectal adenoma in the same inale cohort of the or cereal) appeared to be protective.4’ The major differ-
Health Professional Follow-up Study.” The analysis was ence between the earbert and updatedd’ reports of this
done on 7284 individuals who had undergone either prospective study is that the earlier analysis assessedfewer
colonoscopy or flesible sigmoidoscopy. There were 170 potential confounders, which might have led to an
cases of endoscopically diagnosed adenomas of the left overestimation of the relationship between total fiber or
colon or rectum between 1756 and 1YSS.46Again, most source of fiber and adenoma risk. Another problem in the
potential confounding factors were adjusted for analysis. earlier report is the small number of cases arising during
Dietary fiber was inversely associated with risk of ad-
2 years of follow-up, compared with S years in the later
enoma (P for trend < 0.0001); RR for men in the highest
(>28.3 g/day) vs. the lowest (c16.6 g/day) quintile was
The Iowa Women’s Health Study4S included 98,030
0.36 (75% CI, 0.22-0.60). 4L6All sources of fiber (crude,
postmenopausal women aged 55-69 years who were
vegetable, fruit, and grain) were associated with de-
asked to complete a self-administered questionnaire
creased risk (P < 0.02).4’ The inverse relationship with
dealing with various health issues and diet. Nearly half of
fiber persisted after adjustment for other nutrients com-
the women (41,S37) recurned the questionnaire. This
monly found in fruits and vegetables (p-carotene, potas-
cohort was then followed up for 4 years. Dietary intake of
sium, magnesium, and vitamins C and E).46
various factors was assessedusing the Willett semiquanti-
This study46 was further analyzed with a longer
tative food frequency quescionnaire.49 The occurrence of
follow-up (19SG1994), a larger cohort (16,448 men),
CRC was documented, and the diagnosis was verified.
and newly available data on dietary composition for
After specific exclusion criteria were applied, 212 cases
specific fiber components and fiber warer solubility.47
and 35,004 noncases remained for analysis. hlean dietary
Among 16,448 men who underwent endoscopy between
1786 and 1994, 670 cases of adenoma of the distal colon intake was divided into quartiles of incremental increase,
(n = 531) and rectum (n = 159) were identified. In the and the relative risk for development of CRC was
basic model, the risk of distal colon adenoma decreased calculated for each quartile compared with the quartile
with increasing intake of total dietary fiber (P trend = with the lowest intake. A weak and statistically nonsig-
0.01) and fruit fiber (P trend = 0.001) but not with fiber nificant inverse association was observed between dietary
from cereals, wheat, vegetables, or cruciferous vegetables fiber intake and the risk of colon cancer, pnrcicularly of
(basic model).“’ The RRs comparing the highest (median the distal colon.45 Furthermore, increased total intake of
32.3 g/day for rota1 fiber and 8.4 g/day for fruit fiber) both vegetables and fruits did not reduce the relative risk
with the lowest (median 11.6 g/day for total fiber and 1.3 of CRC; similar results were obtained when each veg-
g/day for fruit fiber) quintile were 0.65 (75% CI, etable or fruit item was independently analyzed except fo:
0.46-0.91) for total fiber and 0.67 (95% CI, 0.50-0.90) garlic.4s
for fruit fiber.47 In the full mulcivariare model that In summary, published large prospective studies have
controlled for all potential confounding factors, the risk produced equivocal findings. Although the data from
ofdiscal colon adenomas decreased with increasing intake earlier prospective studies that examined the relationship

between dietary fiber incake and CRC mortality were during the first 6 years of follow-up.4’ Except for this one
inconsiscenc,3x@~“’ the most recent large prospective study with a follow-up of 16 years,4’ these studies are
study (Cancer Prevention Study II),4’ involving more limited by the relatively shorr follow-up (2-S years)?5.45-48
than 1 million subjects, showed a significant inverse This issue is important because of the uncertainty
relationship with a 30% reduction of CRC mortality in regarding the biologically relevant period of exposure
subjects consuming the highest amount of dietary fiber before the development of colorectal adenomas or CRC.
compared with those consuming the lowest amount. This Another potential shortcoming that limits the interpreca-
study also showed that the risk of fata! colon cancer tion of results is imprecise estimation of dietary fiber
decreased with more frequent consumption of vegetable intake. Although the Willerc semiquantitative food
and high-fiber grains (P trend = 0.031 in men and frequency questionnaire has been shown to be reproduc-
0.0012 in women). ” More recently p ublished prospec- ible and valid in these cohorts,s0-5’ the estimates of
tive studies of the relationship between dietary fiber dietary fiber intake were dependent on a self-adminis-
inrake and the risk of CRC or adenomas have demon- tered questionnaire. As previously mentioned, analytical
strated a protective effect of dietary fiber against discal tools used to determine the fiber content of foods also are
colon and rectal adenomas in men4G*4’ but not in women relatively imprecise and underestimate amounts of di-
(Table 3).4d When all potential confounding factors were etary fiber. Therefore, fiber values assigned to each
corrected for, however, an inverse dose-responsive associa- reported food consumed have errors. These prospective
tion was observed only for fruit and soluble fiber.47 In studies also lack data on food preparation methods,
these 2 studies, 46*47there was a 35%63% reduction in cooking, and chewing, which can alter the physiological
the risk of developing distal colon and rectal adenomas in properties of fiber. s5The 2 cohorts studied in the Nurses’
men with the highest dietary fiber intake compared with Health Study and Health Professionals Follow-up
chose with the lowest fiber intake. These studies also Study 43-47are highly educated and affluent professionals
showed a significant inverse dose-responsive relationship with relatively homogeneous lifestyles and dietary habits
(P trend < 0.001). 46*47However, it appears that diecary and thus may not be representative of the general
fiber has no significant effect on CRC incidence in men”> population. Therefore, the applicability of observations
or women (Table 3). 43*44,4sIt is possible that, at least in made in these cohorts to the general population is in
men, dietary fiber influences the early stages of coloreccal question. One solution to this difficult issue is corrobora-
carcinogenesis and not the lace stages. This hypothesis is tive evidence from international and cross-cultural pro-
further supported by the observation that dietary fiber spective studies. The other potential problem is that in
has a protective effect against small (<l cm) and not the Nurses’ Health Study and Health Professionals
large (> 1 cm) adenomas.@ Follow-up Study cohorcs,43-47 the range of dietary fiber
The strengths of recent prospective studies4’-48 are consumed might have been narrow, and thus protective
numerous: the studies were conducted prospectively and effects of fiber might have not been observed. Therefore,
involved a large number of subjects for adequate statisti- potential protective effects of extremely high intake of
cal power; most controlled for potential confound- dietary fiber (>35-50 g/day) cannot be ruled our. Some
ers41,45,47,1’;the largest study followed up study subjects studies examined the incidence of colonic adenomas only
for 14 and 16 years for colorectal adenomas and CRC, in the distal colon, and results cannoc be extrapolated to
respectively44; and the studies used the Willetc semiquan- the proximal colon.4’,46v47
titative food frequency questionnaire49 to accuracel!
estimate dietary fiber intake. One of the major weak-
nesses of these studies is that the investigators attempted
Human Intervention Studies
to correlate dietary consumption of dietary fiber ar In theory, randomized intervention studies in
baseline with subsequent incidence of CRC or adeno- humans should provide definitive support for the pur-
mas.4~~45--iyIn orher words, the dietary intake at baseline ported cause-and-effect relationship between a dietary
was assumed to reflect past and subsequent consumption. factor and CRC. HoweLrer, intervention studies are often
Whether the subjects in these studies changed their diets difficult to carry out because of the slowly progressive
during the follow-up period and how this might have nature of neoplascic transformation and the large number
affected the study outcome cannot be deduced. The of subjects necessary to achieve an adequate statistical
exception is the largest published study, with a 16-year power. However, several strategies have been developed
follow-up, which showed no protective effect of dietary to circumvent these problems. One is to study the
fiber intake even when the analysis included only those modulatory effects of nutritional factors on colorectal
who maintained a consistent level of dietary fiber intake carcinogenesis in individuals at high risk of developing

CRC. The second strategy is co use so-called intermediate rence) are associated with increased risk of developing
biomarkers of CRC rather than occurrence or recurrence adenocarcinoma compared with adenomas without these
of CRC as the endpoint. 5’~5sThese biomarkers include characteristics.j’
adenoma, proliferation markers, mitotic index, DNA Several randomized or single-arm intervention studies
aneuploidy aberrant crypts, mucins, and more recently using a high-fiber diet as a component of chemopreven-
alterations of several molecular biological markers.5”*55 tive strategies against the development of CRC have been
However, all intermediate biomarkers have limitations, conducted or are underway (Table 4). The firs: such study
and most have nor been validated conclusively in clinical was conducted on 58 subjects with familial adenomatous
studies.“*” Furthermore, except for coloreccal adeno- polyposis (FAP) who had undergone total colectomy and
mas, 5~~ changes in any of these intermediate biomarkers ileorectal anastomosis at least 1 year before encry into the
have not yet been proven co lead to a reduction in CRC tria1.58 These subjects were randomized to receive either a
occurrence and mortality. 51~55Even with adenomas, ic is low-fiber supplement (2.2 g/day) plus placebo (control
known that few adenomacous polyps progress to cancer; group), a low-fiber supplement (2.2 g/day) plus ascorbic
rhe rate is estimated at approximately 2.5 polyps per acid (4 g/day) and cx-tocopherol (400 mp!day), or a
1000 per year.31 Ir has also been well established that only high-fiber supplement (22.5 g/day) plus ascorbic acid (4
adenomatous polyps with certain characteristics (>l cm, g/day) and a-tocopherol (400 mg/day).s” The fiber
tubulovillous or villous histology, and multiple occur- supplement was from a grain source. Over the course of 4

Table 4. Summary of Intervention Studies Using High Fiber as a Chemopreventive Strategy for CRC
Total fiber
Location Case Sample Type of intaks Primary
study (Yr) diagnosis size study Intervention (g/day) Duration aIdpoint Outcome Comments

De Cosse USA Familial adwomatous 58 RCT Lowfibersupp:ement 11.3 4r Adenoma High-fiber protective Grain/cereal fiber
et aI.- (1989) polyposis. total (2.2WW regression/ cnly if z-11 g/day supplement: small
colectomy. and +Vitamin C (4 occurrence rumber: poorcorn
ileorectal WW pliance: substantial
anastomosis +Vitamin E (400 degree of intrapb
m2/W ttent and intervisit
VS. variability in fiber
High-fiber supplement 22.4 Mtamins C. E; trend intake
(22.5 p/day) tward protection
+Vitamin C (4
+Vitamin E (400
Placebo 12.2
Alberts us4 Previous 17 Single arm. Fiber suoolement 30.9 6 v.k Proliferation Overall 22% Uriontrolled: small
et at.59 (1990) CRC ““CO”- (&a: bran. Labeling index decrease rumber; limitations
trolled 13.5 g/day) [‘H]Thymidine compared with labeling index
Hith baseline
Alberts USA Previous 100 RCT 2 X 2 factorial 14.4-17.5 9mo Proliferation No effect Lin,:ations with
et al.we: (1997) colorectal fiber (wheat bran) (low-fiber Labeling index labeling index and
adenomas High (13.5 g/day) 2Kw) [‘Hjihymidine 52% decrease fecal bile acids as
Low (2.0 g/day) 25.7-28.7 Total fecal bile with high fiber b omarkers; short
Calcium (high-fiber acids (P= 0.011 duration: small
High (1500 group) Fecal 36% decrease rumber
mWW deoxycholic with high fiber
Low (250 mg,‘day) bile acids (P = 0.003)
Toronto Polyp Canada Previous 201 RCT Dietary counseling to 35 2 yr Adenoma Intention-t*treat, P;cr compliance: high
Pwention (1994) cOlorectal achieve 20% fat rewrrence no effect cvpout rate: low
GrourP adenomas calories and 50 Subanalysis in fallowup colonos-
g fiberjday those with sub copy rate
vs. stantial dietary
16 counseling.
50% reduction in
women am 90%
increase in men
with high fiber
Australian Australia Previous 424 ACT 2 x 2 x 2 factorial NA 4 yr Adenoma Low fat. high fiber S-M no. of subjects
Polyp (1995) cOlorectal C25% fat calories recurrence decreased in each of the 8
Prevention adenomas 25 g wheat recurrence of armsof2x2x2
Proje@ bran/day >10mm adenamas design: small no.
pcarotene (20 of subjects with
mWday1 210 mm aden*
mas: some differ-
ences at baseline
among groups

RCT. randomized controlled trial: NA. not available.

years, each participant underwenr procrosigmoidoscopy to fecal bile acid concentrations and excretion rates,
every 3 months, for a cotal of 1S examinations. Overall high-dose fiber supplemencacion for 9 months caused a
consumption of fiber from supplements and dietary reduction in fecal concentrations of total bile acids (52%
sources averaged 12.2 g/day in the placebo group, 11.3 reducrion; P = 0.001) and deoxycholic acid (4S%
g/day in the vitamin group, and 22.4 g/day in the reduction; P = 0.003) compared with baseline concentra-
high-fiber group. x When results were analyzed on an tions.6’ High-dose calcium supplementation also had a
intention-to-treat basis, only a weak protective effect of significant but smaller effect on the mean toral bile acid
fiber against polyp occurrence was obsened.‘” However, (35% reduction; P = 0.044) and deosycholic fecal bile
Tvhen only those with good compliance were analyzed, acid (36% reduction; P = 0.52) concentrations at 9
those who had consumed 11 g of supplemental fiber in months compared with baseline.G’ Presently, the same
addirion to their usual dietary fiber intake had a signifi- investigators have included more than 1400 parients in a
cant reduction in polyp occurrence in the rectal stump, randomized phase III trial of high-dose (13.5 g/day) vs.
and polyp number decreased incrementally as the amount low-dose (2 g/day) wheat bran fiber in parienrs wirh
of ingested, prescribed fiber increased.SS The effects of resecred colorectal polyps. ” Polyp recurrence after 3 years
vitamins C and E were not significant, although there was of daily fiber intake serves as the primary endpoint for
a trend toward protection. ss A significant fault of this this dietary intervention trial.”
study is poor compIiance with intervention modalities In the trial reported by the Toronto Polyp Prevention
over the 4 years of the study. Compliance decreased by Group from Canada, 201 subjects with adenomacous
more than 50% over 4 years in some of the group~.~~ colorectal polyps were randomized after polypectomy to
Other legitimate criticisms are uncertainty about whether receive intense counseling on a diet low in fat (<50 g/day
the 3 groups were similar with respect co dietary incake of or 20% of energy) and high in fiber (50 g/day), mainly
components other than fiber, vitamins C and E, and fat from wheat bran, or to follow a normal western diet, high
and whether any of the groups changed their diecar) in fat and low in fiber.63 After 12 months of counseling,
patterns during the study period. fat consumption was approximately 25% of energy in the
One study from the Arizona Cancer Center was a low-fat/high-fiber group and 33% in the wesrern diet
single-arm study that investigated the effect ofsupplemen- group; fiber consumption was 35 g and I6 g respec-
tal wheat bran fiber on a proliferation marker (13H)thymi- tively.‘j After an average of 2 years of follow-up with
dine labeling index) in patients who had undergone colonoscopy, an intention-to-treat analysis showed no
resection for colon or rectal cancer.‘” In this study, 13.5 g significant difference between dietary groups with regard
of supplemenral wheat bran per day significantly reduced to the recurrence of adenomatous polyps.” However,
colorectal epithelial proliferation during the S weeks of when only those subjects who had received substantial
follow-up.59 However, this was not a randomized placebo- dietary counseling were reanalyzed, it was found that
controlled study and involved only I7 subjects for the women who ate the low-fat and high-fiber diet showed a
analysis. Furthermore, the [jH)thymidine labeling index nonsignificant 50% reduction in polyp recurrence (RR,
is not uniformly accepted as an accurate means of 0.5; 95% CI, 0.2-1.9) associated with a reduced concen-
determining the proliferation index of the colonic epirhe- tration of fecal bile acids.63 Among men, the polyp
Iium.5’*s’ Finally, changes in this index have not been recurrence rate was increased by approximately 90% in
proven to decrease the incidence of CRCS’rSS the low-fat/high-fiber diet group compared with the
The same investigators have recently completed a controls (RR, 1.9; 95% CI, O.S-4.4).G3 This also fell short
double-blind, randomized phase II study using a 2 X 2 of statistical significance but was associated with an
factorial design to determine the effects of wheat bran increased concentration of fecal bile acids in these
(2.0 or 13.5 g/day) and calcium carbonate (250 or 1500 subjects.“j The main problems with this stud> were (1) a
mglday) supplementation on CjHJthymidine labeling high dropout rate (only S2% of 201 subjects received
index in rectal mucosal biopsies and fecal bile acid colonoscopic follow-up), (2) noncompliance with the
concenrracions ac 3 months and 9 monchs.G0*6’Total fiber low-far/high-fiber diet, (3) small sample size, and (4)
intake ranged from 14.4 to 17.5 g/day and 25.7 to 28.7 short duration of follow-up. However, this study points
g/day in the low- and high-fiber groups, respectively.“J’ out that physiological differences in fecal bile acids may
The results of this study, which included 100 patients exist between men and women and that these may
who had undergone complete colonoscopy with colonic account for differences in the rates of the polyp recurrence
polyp removal within 24 months of study entry, showed on the low-fat/high-fiber diet.
that neither wheat bran fiber nor calcium treatment In a recently reported study from Australia, 424
significantly decreased the labeling index.GO With respect subjects with adenomas and a “clean” colon were random-


ized to diets containing <25% of energy as fat, 25 g Several randomized intervention studies using a high-
wheat bran supplement, and/or 20 mg P-carotene per day fiber component for the nutritional chemoprevention of
in a 2 X 2 X 2-factorial prospective, randomized, CRC are currently ongoing in the United States and
controlled trial.” Endpoints were adenomas and CRCs Europe (Table 5). The Polyp Prevention Trial is a
identified by colonoscopies performed at 2 and 4 years.6“ mulri-institutional intervention study recently com-
This trial showed that neither low-fat intervention nor pleted in United States .9 The primary goal of the trial is
wheat bran supplementation alone had a significant effect to test the ability of a low-far (20% fat calories),
on adenoma recurrence. 6rt However, low-fat intervention high-fiber (18 g/1000 kcal daily) diet enriched with
combined with wheat bran supplementation significantI> vegerables and fruits (5-S servings daily) to decrease the
reduced the occurrence of large adenomas (> 10 mm) at 2 recurrence race of adenomatous polyps in patients previ-
and 4 years of follow-up (P < O.O?S).“’ In this trial, ously treated for colon adenomas.g To date, 2079 patients
p-carotene, either alone or in combination with the have been randomized co the intervention or control
low-fat or high-fiber intervention, had no effect on arm.9 This trial provides 90% power to detect a reduction
adenoma recurrence.6f Problems with this trial were (1) of 24% in the annual adenoma recurrence rare.9 The final
small number of subjects in each of the 8 arms of the 2 X colonoscopic examinations at 4 years of follow-up were
2 X 2-factorial design; (2) small number of subjects with completed in early 199S.g The European Cancer Preven-
large adenoma (> 1 cm), which was used as the secondary tion Organization study is an ongoing study to compare 3
endpoint of the trial, thereby increasing uncertainty of groups, one given ispaghula husk (a mucilaginous sub-
the results; and (3) differences among groups at baseline stance), 3.S g/day for 3 years; one given calcium, 2 g/day;
with respect to prevalence of multiple (22) and large and one given placebo. 65All subjects in this study have at
(> I cm) adenomas. least 2 adenomas or 1 adenoma that is >5 mm in

Table 5. Summary of Ongoing Randomized, Double-Blind, Placebo-Controlled Intervention Studies Using High-Fiber Diets
Case Sample Duration Primary Current
Study Location diagnosis size (n) Intervention (Yr) endpoint status
Phase Ill Arizona USA Previous colorectal 1400 High-fiber supple- Adenoma recur- Completed
Cancer Center adenomas ment (13.5 rence
Polyp Prevention wheat bran/day)
Polyp Prevention USA Previous colorectal 2079 20% fat calo- Adenoma recur- Completed
Trial9 adenomas ries/day rence
18 g fiber/1000
5-9 servings of
vegetables and
fruits/day vs.
typical North
American diet
European Cancer Europe Previous colorectal 656 3.8 g ispaghula Adenoma recur- Ongoing
Prevention Orga- adenomas (2 husk/day vs. 2 rence (prolifera-
nization Studf5 adenomas or 1 g/day calcium tion labeling
adenoma >5 vs. placebo index and fecal
mm) bile acids as
secondary end-
Concerted Action Europe (14 FAP gene carriers 468 2 X 2 factorial Incidence or pro- 150 recruited
Polyposis Pre- countries) 600 mg aspirin gression of
vention 156 30 g corn starch colonic
(13.2 resistant adenomas
Concerted Action Europe (14 HNPCC gene car- 1200 2 X 2 factorial Incidence of Ongoing
Polyposis Pre- countries) riers 600 mg aspirin colorectal
vention 2 30 g corn starch adenomas
(13.2 resistant (extracolonic
starch) malignancy, pro-
liferation, apop
tosis, genotype
as secondary

diameter (i.e., subjects with a high adenoma recurrence who therefore may be ac a lace, although preclinical, stage
rate).6r The primary endpoint is recurrence of adenoma, of colorectal carcinogenesis or have precancerous lesions.
and secondary endpoints are mucosnl cell proliferation Ocher limitations are associared with incervenrion trials
race and fecal bile acids. 65 To date, 656 subjects have been in humans. Blind or double-blind trials are usually
randomized to -3 arms of the srudy.“5 TWO randomized impossible tvith foods or dietary macronutriencs, which
trials (Concerted Action Polyposis Prevenrion [CAPP} 1 are recognizable. In nonblind studies of foods, subjects in
and 2, respectively) designed to test effects of 600 mg the control group may adopt the dierary behavior of the
aspirin and/or 30 g corn starch (equivalent co 13.2 g of treatment group if they think the treatment diet is
resisrant starch) in a factorial design in FAP and heredi- beneficial. Such trends may obscure a real benefit of
tary nonpolyposis CRC (HNPCC) gene carriers are rreatmenr. In addition, the rime between a change in the
ongoing in Europe (14 countries). The primary endpoint level of a dietary factor and any expected change in the
of CAPP 1 is incidence or progression of colonic adeno- incidence of cancer is usually uncertain. Trials should
mas. To date, CAPP 1 has recruited 150 gene carriers therefore be of a long duration. Finally, people who agree
(carget n = 468) from FAP registries in 14 European to participate in trials tend to be relatively health
counrrjes. 6G The primary endpoint of CAPP 2 is rhe conscious and highly motivared; people who are at high
incidence of colorectal adenoma. The secondary end- potential risk on the basis of dietary intake, and thus
points include the incidence of excracolonic malignan- susceptible to intervention, are likely to be underrepre-
cies, crypt cell proliferation, apoptosis, and genotype. sented. Hence, the validity of generalizing the results is
CAPP 2 has jusr begun recruitment (rargec n = 1200). limired. Therefore, results of intervenrion studies should
In summary, 6 intervention studies in humans have be interpreted with caution. They are not an epidemiologi-
been completed and published (Table 4).5s-6’~65~6’Most of cal “gold standard.” Controlled trials in which interven-
these studies included a small number of subjects (range tion shows beneficial effects are good evidence that the
17-424) and a follow-up period of 8 weeks to 4 agents used are prorective. However, studies in which
years. 5%61,63,6-1one sru,-Jy57 was uncontrolled, and 5 were intervention shows no effect, or even 3 detrimental effect,
randomized and placebo controlled.Jg~6n~61~63~6~ Except for do not show that the agents used are irrelevant or harmful
1 study x cllac recruited patients with FAP, the parcici- in the context of whole diets or among normal, healthy
pants of most studies were individuals with sporadic populations. The results of intervention studies should
colon adenomas.59-6’*65*64Three scudiesSS,63~6’used ad- not be treated as a refutation of evidence from other types
enoma recurrence or regression as the endpoint of the of epidemiological study, especially when such other
trial, and the other 3 used less well-established incermedi- evidence is backed by data from animal studies and
ace biomarkers of CRC (labeling index59qGaand fecal bile identification ofplausible biological pathways.
acids”). All studies5S~Go~6’~G3.6” except one59 used dietary
fiber supplements in conjunction with other dietary
Resistant Starch and Short-Chain
factors (vitamins, calcium, low fat). Four studies showed a
Fatty Acids
moderate protective effect of dietary fiber supplements:
decreased labeling index in 1 study,s9 decreased fecal bile Resistant starch is defined as that portion of
acids in 1 study,“’ a.nd decreased adenoma recurrence in 2 ingested starch that escapes digestion in the small
scudies.5s@ The other 2 showed no effect on labeling intestine. 19*** hiore recently, it has been suggested that
index” or adenoma recurrence.63 The srrongest evidence resistant starch be defined 3s “the sum of srarch and
to dare to support the fiber hypothesis is the Australian starch-degradation products that, on average, reach the
Polyp Prevention Project, which showed that a diet high human large intestine.“6’ Similar co nonsrarch polysaccha-
in fiber and low in fat prevents recurrence of large rides, resistant starch has been shown to increase stool
adenomas (> 10 mm).” bulk, decrease fecal pH, alter the colonic microflora,
The major weaknesses of these incervencion srudies are decrease secondary bile acid concentrations and cytocoxic-
short follow-up, small numbers of subjects, poor compli- icy of fecal water, decrease colonic mucosal proliferation,
ance with dietary interventions, high dropout rates, and increase colonic fermentation, and contribute to short-
use of less well-established intermediate biomarkers with chain fatty acid (SCFA) synthesis, especially buty-
uncerrain functional ramifications in some studies. An- rare. “-” A recently published international correlation
ocher problem is that these studies attempted to inter- study supports the protective role of resistant starch in
vene in incompletely undersrood biological pathways in the development of CRC. ” In rhis study, intakes of
special populations of adults at high risk of developing starch, nonstarch polysaccharides, protein, and fat were
CRC (e.g., those with FAP or previous colonic adenomas) compared with CRC incidence in 12 populations world-

wide. After fat and protein intakes were controlled for, Table 6. Possible Mechanisms of Action of Dietary Fiber
there was a strong inverse association between starch Increased stool bulk
consumption and CRC (correlation coefficient, Y = Dilution of potential carcinogens
Decrease in transit time (less contact time for carcinogens)
-0.70); no significant association with nonsrarch polysac- Binding with potential carcinogens
charides was observed (r = -0.29).” ‘When nonstarch Binding with bile acids
Dolysaccharides were combined with resistant starch to Decrease in fecal bile acid concentrations
Prevention of conversion of primary to secondary bile acids
give an estimate of fermentable carbohydrate, the inverse
Lowers fecal pH
association became significant: with Y = -0.52.” Resis- Reduced solubility of free bile acids
tant starch was observed to either be protective,” have no Inhibition of 7adehydroxylase, which converts primary to sec-
effect,‘3 or enhance tumorigenesis’” in chemical rodent ondary bile acids
Inhibition of bacterial degradation of normal fecal constituents to
models of CRC. In another study using a knockout potential carcinogens
murine model of the adenomatous polyposis coli gene Alters colonic microflora
(A~L-‘“~~~),‘~ resistant starch was shown to significant11 Inhibition of microbial enzymes involved in carcinogen activation
Changes in bacterial species
increase small bowel tumors.76 Two randomized, double- Stimulation of bacterial growth, which increases fecal bulk
blind, placebo-controlled intervention studies designed Fermentation by fecal flora to SCFAs
to test the effect of resistant starch on CRC in both FAP Inhibition of growth of tumor cell lines
Induction of differentiation
and HNPCC gene carriers are ongoing in Europe (CAPP
Induction of apoptosis
1 and 2; Table 5). Modulation of gene expression
Fermentation of dietary fiber and resistant srarch by Prevention of insulin resistance and hyperinsulinemia
colonic bacteria generates SCFAs. The principal SCFAs
are acetate, propionate, and buryrate, which account for be prevented. Th e b ound bile”acids or bile salts may pass
90%959% of SCFAs in the colon. SCFAs are an impor- out of the alimentary tract in the feces. The possible
cant energy source for the colonocyces. Butyrate is the putative interaction of secondary bile acids and colonic
preferred SCFA to meet colonic energy requirements. mucosal cells will thus be decreased.
SCFAs, especially butyrate, have been shown to have Dietary fiber decreases fecal pH, resulting in reduced
anticarcinogenic properties, as discussed in the next solubility of free bile acids; theoretically, this should
section.“*” Butyrate has been shown to either sup- decrease the potential tumor promoter activity of second-
press79-81or have no effect ons2 the development of CRC ary bile acids. go Furthermore, the activity of the colonic
in animal modeis. There is a paucity of data from human bacterial enzyme 7or-dehydroxylase, which converts pri-
epidemiological and intervention studies concerning the mary bile acids to secondary bile acids, is inhibited at a
effects of SCFA on coloreccal carcinogenesis. pH of <6-6.5. g1sg’Acidification of colonic contents also
increases the availability of calcium for binding to free
bile and fatty acids, thereby inhibiting their effects on the
Biological Plausibility: Potential
colonic mucosa.93 A number of epidemiological studies
Mechanisms of Action
have shown that human populations with lower fecal pH
Several potential mechanisms by which dietary have lower rates of colon cancer.90*g4 However, direct
fiber can protect against the development of CRC have experimental acidification of the colon contents in animal
been proposed and investigated (Table 6).L”*s3~84 Burkitr’s models have nor always led to a reduction in tumorigen-
initial hypothesis was that dietary fiber increases stool esis.9s Theorecicaliy, fecal acidification can also inhibit
bulk, thus diluting potential carcinogens and decreasing bacterial degradation of normal fecal consticuenrs to
transir time, which would permit less contact time potential
between potential carcinogens in the lumen and the gut Another potential mechanism of dietary fiber relates to
mucosa. 26 Additional mechanisms also have been pro- alterations in colonic microflora, which can exert marked
posed (Table 6>.‘s,sj-s4 effects on the colonic environment. These may be charac-
Ir has been demonstrated that carcinogens can bind to terized by changes in bacterial species, functional changes,
dietary fiber, but the extent of binding depends on the or production of microbial enzymes considered to be
carcinogen and dietary fiber.*>-” Dietary fiber has also important in carcinogen activation (e.g., cr-glucuroni-
been shown to bind with bile acids, thus reducing fecal dase, a-glucosidase, azoreductase, and nitroreductase).g6
bile acid concenrration.8S~s7 If the dietary fibers to which Although dietary fibers clearly modulate colonic bacterial
the bile acids or bile salts are bound are undegraded in the enzyme activity, l3 the relationship between colonic bacte-
colon, deconjugation of bile salts and conversion of rial enzyme activity and development of human CRC has
primary to secondary bile acids by bacterial enzymes may not been elucidated clearly. Dietary fibers that are

extensively degraded in the colon have been shown to compared with nondiabetic control subjeccs.‘1”*‘15 In a
increase fecal bulk by a stimulation of bJcreria1 growch.9’ population-based cohort study from Sweden (n =
Bacteria, rather than undegraded dietary fiber, are the 153$52), subjects with diabetes mellitus were found to
major water-holding component of feces.9‘ Increased fecal have on average a 40% greater risk of developing colon
bulk and reduced transit time resulting from increased cancer and a 60% greater risk of dying of colon cancer
bacterial growth would reduce the possibiliry of effective than the general population.“4 The first Cancer Preven-
interactions of carcinogens with the colonic mucosa. tion Study of the American Cancer Society with more
Dietary fiber also can decrease numbers of anaerobes, than 1 million participants showed that diabetic men had
resulting in a decrease in secondary bile acids.‘j a statistically significant 30% increase in risk ofdevelop-
SCFAs, especially butyrate, produced by fermentation ing CRC compared with nondiaberic men during 13
of dietary fiber and resistant starch by colonic bacteria years of follow-up. “’ Two recenrly published animal
appear to be an imporcanr factor in colorectal carcinogen- studies have demonstrated that exogenously injected
esis.7’,78 Although butyrace serves as the primary energS insulin promores the development of coloreccal tumocsllG
source for normal colonic epithelium and stimulates and the growth of aberranr crypt foci,“’ a putative
growth of colonic mucosa, in colonic tumor cell lines it precursor of colon cancer, thereby providing support for
inhibits growth98v99 and induces differentiation’00 and the causal hypothesis linking insulin resistance and CRC.
apoptosis. lo’ At the molecular level, buryrate has been Because dietary fiber, especially soluble fiber, affects
shown to inhibit hiscone deacetylase, resulting in hyper- glycemia and insulinemia,“* rhe insulin hypothesis
acetylacion of histones and increased accessibility of DNA could be a mechanism by which dietary fiber can
to factors controlling gene expression.*0’*‘03 Butyrace also modulate colorectal carcinogenesis. As such, this hypoth-
has been shown to alter the binding of regulatory esis merits further consideration.
transacting proteins to specific DNA sequences thar Epidemiological and experimental evidence indicating
control the expression of the gene.‘04 a causal association between dietary fiber and CRC is
A unifying hypothesis that may explain how diet and strengthened when a biological pathway or mechanism
lifestyle factors modulate coloreccal carcinogenesis has by which colorectal carcinogenesis may be modified is
recently been put forward by McKeown-Eyssen’ and identified and when this mechanism is biologically
Giovannucci.* This hypothesis suggests that the putative plausible. However, it can be argued that epidemiological
dietary and lifestyle factors associated with CRC risk data, strong and consistent, are an inadequate basis for
cause insulin resisrance and hyperinsulinemia and that any definite judgment of causality. unless supported by
hyperinsulinemia may in turn stimulate the growth of mechanistic evidence.‘19 Although investigations to elu-
colorectal tumors.‘~* Although ir remains unproven cidace potential anticarcinogenic mechanisms of dietary
whether insulin stimulates the growth of colon rumors in fiber have focused on physical properries of dietary fiber,
humans, several lines of evidence support its role. Insulin more recent work has expanded into physiological func-
is an important growth factor for colonic mucosal cells tions and molecular mechanisms of dietary fiber. A better
and is a mitogen of colonic carcinoma cells in .itro.10s,106 mechanistic understanding of how dietary fiber can
Colonic cancer tissue has both insulin and insulin-like modulate colorectal carcinogenesis can lead to a more
growth factor (IGF) 1 receprors’07,108; insulin has been rational strategy using dietary fiber supplemen;arion to
shown to exert its mitogenic effect partly through IGF-1 prevent CRC in humans.
receptors.“” Insulin receptors can be bound by IGF-1,“O
and a binding protein from IGF-1 inhibits the growth of Conclusion
colon cancer cells in vitro. “’ Another indirect line of
Summary of Causal Inference
evidence comes from the obsenacion that subjects with
acromegaly, characterized by chronic growth hormone Although valuable information can be obtained
and ZGF-1 hypersecretion, have an increased risk of from nutritional epidemiological studies examining rhe
developing CRC. I” It has been proposed that stimula- effect of diet on cancer, several shortcomings limit
tion of IGF-1 receptors by IGF-I or IGF-2 promotes interpretation of the results of these studies (Table 7).12’
colorectal carcinogenesis in subjects nich acromegaly.‘” The strongest evidence that supports the fiber hypoth-
Although epidemiological studies thar hare examined esis is the remarkable consistency of the protective effect
the relationship between diabetes mellirus and CRC risk of dietary fiber among correlation and case-control scud-
have not consistently supported this hypothesis,“j 2 ies conducted in populations with different patterns of
recently published large prospective studies indicate a diet and CRC. Three combined analyses or meta-analyses
modcsr increase in CRC risk in subjects with diabetes of case-control studies also provide srrong support for the

Table 7. Summary of Causal Inference
Criteria Supportive Equivocal Lack Comments
Consistency X Supportive evidence from most correlation ar,d case-control
studies conducted in populations with different patterns of
diet and CRC and meta-analyses and combined analyses of
case-control studies; data from prospective studies
equivocal (only supportive for distal colon and rectal
adenomas in men)
Strength of association Average 50% reduction in CRC and adenoma risk
Dose response Significant dose-dependent inverse association in most corre-
lation and casecontrol studies as well as their meta-
analyses and combined analyses: positive prospective
studies also demonstrate a dose-responsive association
Experimentation: from human X Generally supportive of findings published intervention stud-
intervention studies ies; studies limited by small numbers of pa*ticipants, short
durations of follow-up, use of intermediate markers, and
poor compliance; 5 large, well-designed studies ongoing at
Specificity X Difficult to delineate effects associated with dietary fiber from
other potential anticarcinogens present in fiber-rich foods
Epidemiological coherence X Fiber hypothesis consistent with epidemiological observations
that suggest significantly lower CRC prevalence, incidence,
and mortality in countries or populations wirh high intake of
fiber-rich foods
Analogy X Protective effects of fiber against breast, endometrial,
ovarian, and prostate cancers
Biological plausibi!ity x Several potential physiological and molecular biological
mechanisms for fiber

dose-dependent protective effect of dietary fiber or fiber- ent in fiber-rich foods and fiber are responsible for the
rich foods against colorectal carcinogenesis.2E-30 These observed protective effect of dietary fiber on the develop-
studies suggest on average a 50% reduction in the risk of ment of CRC.
developing CRC in subjects with the highest dierary fiber Of the published randomized, double-blind, placebo-
incake compared with those with the lowest intake.2s-30 controlled studies in humans that have used adenoma
However, large prospective studies conducted in specific recurrence or regression as the endpoint of tht trial,‘*Jj3y6’
popula+ons in the United States do not support the probably the best intermediate biomarker of CRC avail-
protective effect of dietary fiber on the development of able to date,3’ 2 have shown significant protective effects
CRC.43A5y4SOn the other hand, these prospective studies of wheat fiber supplementation5S~G’; the other showed no
suggest a modest dose-dependent protective effect of effect.63 In the largest intervention trial published to date
dietary fiber on distal colonic and rectal adenomas in men (the Australian Polyp Prevention Project, n = 424):’ a
only. 46,47 Although these prospective studies provided diet high in fiber and low in fat was shown to prevent
the least biased approach, the findings need to be recurrence of large adenomas (> 10 mm), probably a more
corroborated by evidence from similar international and relevant biomarker than smaller adenomas (<lo mm).
cross-cultural prospective studies. Five ongoing large, randomized, double-b!ind, placebo-
Ic is difficult to delineate the effect associated with controlled studies in the United States and Europe will
dietary fiber from ocher potential anricarcinogens present certainly provide more insight into the effects of dietaq
in the fiber-rich foods such as vegetables, fruits, cereals, fiber on colorectal carcinogenesis (Table 5).
and grains in epidemiological studies. However, mosr of The fiber hypothesis is consistent with epidemiologi-
the recently published prospective studies have adjusted cal observations that suggest significantly lower CRC
for potential confounding factors, including intake of prevalence, incidence, and mortality in countries or
vegetables, fruits, cereals, and grains as well as antioxi- populations with high intake of fiber-rich foods, includ-
dant vitamins and folate.4tis Some human intervention ing vegetables, fruits, cereals, and grains.’ 3,“,12*121The
studies have attempted to test the effect of dietary fiber protective effect of dietary fiber on the development of
supplementation on colorectal carcinogenesis while keep- CRC is also analogous to similar observations in breast
ing the intake of vegetables, fruits, cereals, and grains cancer,“,“’ endometrial cancer,123,‘2a ovarian cancer,“’
conscant during the study period.ssa It is possible that and prostate cancer, lZG albeit to a lesser degree. More
undetermined interactions among anticarcinogens pres- importantly, several biologically plausible mechanisms

exist for dietary fiber that corroborate epidemiological reduction in CRC risk , were B24.9 and <9.S g/day and
and other evidence (Table 6). However, in vivo verifica- >24.7 and e14.5 g/da); respectively (Table 3). The
tion of some of these mechanisms is still needed. Healrh Professionals Follow-up Study, which demon-
srrated a significanrly reduced risk of disral colon adeno-
Magnitude of CRC Risk Reduction ma@*.” but not CRC,4S generally compared those with
The extent to which CRC mortality races in the dierary intake of fiber of 2S.3-32.S g/day with those with
I_Jnited States might be reduced by practicable dietary dietary intake of fibec of 11.6-16.6 g.‘day. Most of the
means has been estimated at 50%-75%.4 More recently, posirive case-control and prospecrive srudies also showed
the World Cancer Research Fund panel has judged that significanr dose-dependent inverse associations berween
diets high in vegetables, and therefore high in fiber, and dierary inrake of fiber and CRC or adenoma risk.29~jo~“6~47
low in meat; avoidance of alcohol; and regular physical Amounts of fiber supplement or total fiber intake
activity may reduce the incidence of CRC by 66%- chosen for inrervention srudies vary. Of the 2 published
75%.119 Wirh respect to the extenr ofCRC risk reduction intervention studies thar used adenoma recurrence as the
associated wirh dietary fiber or fiber-rich foods, 3 com- endpoint of trial in subjects with sporadic colon adeno-
bined analyses or meta-analyses of case-control studies mas, the Australian Polyp Prevention Project used 25 g
suggest a 50% reduction in the risk of developing CRC wheat bran supplement daily in addition to usual dietary
in subjects with the highest dietary fiber intake compared inrake of fiber (the total intake of fiber was not: stated in
with those with the lowest intake.2s-30 A large, ongoing the report). 61 In contrast, the Toronto Polyp Prevention
prospecrive mortality study (Cancer Prevention Study II Study used 50 g of total fiber intake daily in the
of rhe American Cancer Sociery) with more than 1 high-fiber group compared with the low fiber group, but
million participants suggests a 30% reduction in CRC total fiber incake was 35 g/day in the high-fiber group
morraliry among individuals consuming the highest and 16 g/day in ihe low-fiber group.63 The Phase III
amount of vegetables and high-fiber grains compared Arizona Cancer Cenrer Polyp Prevention Study will
with those consuming the lowest amounr.4’ Two large determine rhe rate of adenoma recurrence in subjects
prospecrive srudies suggesr a 35%-63% reduction in the receiving 13.5 g wheat bran supplement daily in addition
risk of developing discal colon and recral adenomas in to their usual daily inrake of dietary fiber compared with
men consuming the highest amount of dietary fiber those receiving 2.5 g wheat bran supplement daily.” The
compared with chose consuming the lowest amount.46~47 Polyp Prevention Trial will determine the rate of ad-
Although it is difficult to estimate accurately the magni- enoma recurrence in subjects consuming IS g fiber/l000
rude of CRC risk reduction attributable solely to dietary kcal daily compared with those consuming usual amounts
fiber or fiber-rich foods, there appears to be a significanr of dietary fiber.9
degree of reduction. It appears that most case-conrrol, prospecrive, and
inrervenrion studies have assessed the effect of tocal fiber
Dose of Dietary Fiber Associated With intake 3-3.5 times the mean dietary fiber intake in the
Decreased CRC Risk U.S. adult population (11.1 g/day).23 The Toronto Polyp
The threshold level above which dierary intake of Prevenrion Srudy, which acrempred to derermine the
fiber is associated with a significant degree of CRC risk effecr of SO g t-oral fiber intake daily, showed only a
reduction is not well established in epidemiological and nonsignificant SOYEreducrion in adenoma recurrence in
inrervencion studies. Case-control and prospective studies n.omen.63 However, individuals assigned to the high-
have arbitrarily defined increasing quartiles or quintiles fiber incake in this rrial consumed, on average, only 35
of dietary fiber intake, which are different from study co g/day of dietary fiber instead of 50 g/day.
study and from popularion to population. In some
Duration of Intervention Associated With
popularions, rhe difference between exrreme quartiles or
Decreased CRC Risk
quinriles is quite small. In some studies, the amount of
dierary intake of fiber associated with each quartile or There is often a latency period berwern exposure
quincile is nor stated. Two combined analyses of case- to a factor that modifies cancer risk and induction of the
control studies showed a 50% reduction in CRC risk in rumor itself. A further delay occurs before development
individuals consuming 27 g/day compared with those of the tumor reaches the srage at which it can be
consuming less than 11 g/day of fiber.29v3” diagnosed; this delay varies with different factors and
The extreme quartiles or quintiles of dietary fiber different sites. Migrant studies suggest a delay between
intake in the Nurses Hcalch Study4’ and Iowa \Vomen exposure of migrants to urban-industrial diets and emer-
Healrh Study,@ which did not show any significant gence of CRC of lo-20 years.‘19 It follows that appropri-

ace diets may have their full impact in preventing cancer Target Group(s) for Fiber Intervention
only decades after they are widely adopted. These delays Studies that address target groups for intervention
must be considered in setting realistic targets for CRC are lacking in the literature. Intervention studies have
prevention with dietary fiber. Therefore, because CRC is focused on individuals at high risk of developing CRC or
strongly age related” and its incidence races increase adenomas, including those with previous adenomas,
markedly with age beginning around the sixth decade of CRC, and FAP and gene carriers of FAP or HNPCC.
life,“” fiber intervention should begin at least lo-20 Whether increasing dietary incake of fiber will reduce the
years before the peak age for CRC incidence. Prospective CRC risk in the genera! population must be deduced
and intervention studies, except one,“’ have not had a from epidemiological and intervention studies using
long enough follow-up to observe any beneficial effects high-risk individuals and intermediate biomarkers be-
associated with fiber intervention. cause of the cost and duration of the studies. AC present, ic
appears that individuals at high risk of developing CRC
and adenomas will benefit the most from fiber inrerven-
Types of Fiber or Specific Related tion. As previously discussed, the NHANES II study
Components Associated With Decreased identified a marked racial effect, with blacks of both sexes
CRC Risk and in all age groups having lower dietary fiber intake
than whites.23 Unlike the white population in the United
With respect to the exact types and sources of fiber
Scares, blacks have not had substantial improvement in
associated with c-hedecreased risk of CRC, animal studies
CRC incidence and mortality.’
suggest that insoluble and less fermentable fibers and
wheat bran are most effective.‘3 Information on this issue
is lacking in epidemiological and intervention studies in
humans. Although an early analysis from the Health Given a lack of complete scientific evidence, it is
Professionals Follow-up Study suggesred chat all sources difficult to advise patients with absolute confidence.
of fiber (crude, vegetables, fruits, and grains) were Nevertheless, the guidelines in this review- represenr
associated with decreased risk of adenoma in men,4G a reasonable conclusions based on currently available data.
more recent analysis of this cohort suggests that only Therefore, ic is reasonable co recommend tocal fiber
total dietary fiber, fruit fiber, and soluble fiber are intake of at least 30-35 g/day. Dietary fiber should be
significantly associated with decreased risk of colonic from all sources, including 5-7 servings of vegetables and
adenomas.4’ hfost intervention studies have used either fruits daily and generous portions of whole-grain cereals
wheat bran fiber supplement58-Gi or all sources of fiber.9 as recommended by the World Health Organization and
the National Cancer Institute. Because of uncertainty
Two published intervention studies have used adenoma
about the types and sources of fibgr that are most effective
recurrence as the endpoint of the trial and wheat bran
in the prevention of CRC and as yet undetermined
supplement; results of the Australian Polyp Prevention
potential interactions between fiber and other rnticarcino-
Project’” were positive, and results of the Toronto Polyp
gens present in fiber-rich foods, it is prudent to recom-
Prevention Study”j were negative.
mend a high intake of dietary fiber from all sources,
Although the role of resistant starch in colorectal
including vegetables, fruits, cereals, grains, and legumes.
carcinogenesis has recently received much attention, It is clear that as yet undetermined interactions among
convincing epidemiological evidence is lacking except for dietary components and other lifestyle factors play a more
one internarional correlation study that showed a strong imporcanc role in coloreccal carcinogenesis than indi-
inverse association between starch and resistant starch vidual dietary and lifestyle facrors. The dietary guidelines
consumption and CRC risk.” Similarly, 4 published from the American Cancer Society and the National
animal studies to date71-7G17Ghave produced conflicting Cancer Institute encourage healchp earing habits and
results, with 2 studies”“’ showing enhanced tumorigen- lifestyle modifications. All of the factors in these guide-
esis associated with resistant starch. In contrast to lines have been considered to play an important role in
resistant starch, most of the published animal studies colorectal carcinogenesis as well.‘p3 The guidelines can be
using bucyrate demonstrated protective effects of this used in conjunction with the dietary fiber recommenda-
SCFA on colorectal carcinogenesis.79*s! Because several tions. The guidelines are (1) eat each of the 5 food groups
biologically plausible mechanisms exist for buryrate, this daily (meat, dairy products, grains, fruits and vegetables);
SCFA warrants further consideration in intervention (2) reduce total fat intake to less than 25%-30% of tocal
trials. calories and saturated fat to less than 10% of total

calories; (3) eat 5 or more servings of fresh vegetables and of high-fibre foods and energy and the risk of cancer of the large
bowel and breast. Eu: J Cancer Prev 1998;7(suppl 2):Sll-S17.
fruits daily (raw better than cooked; include deep yellow 13. Jacobs LR. Fiber and colon cancer. Gastroenterol Clin North Am
vegetables and dark green cruciferous vegetables); (4) eat 1988;17:747-759.
red meat infrequently (substitute chicken or fish without 14. Greenwald P, Lanza E, Eddy GA. Dietary fiber in the reduction of
colon cancer risk. J Am Diet Assoc 1987;87:1178-1188.
skin); (5) eat more fiber-rich foods such as whole-grain
15. Englyst HN. Cummings JH. Non-starch polysaccharides (dietary
cereals, fruits, and vegetables (daily tot-al of 20-30 g fiber) and resistant starch. In: Furda I, Brine CJ, eds. New
fiber); (6) avoid obesity; (7) eat salt-cured, smoked, and development in dietary fiber. New York: Plenum, 1990:205-
nitrite-cure foods in moderation; (S) keep alcohol con- 225.
16. Burkitt DP, Trowel1 H. Some implications of dietaryfibre. London:
sumption moderate; (9) participate in daily physical Academic, 1975.
activity; and (10) do not smoke. Increasing total fiber 17. Pilch SM, ,ed. Physiological effects and health consequences of
intake to >30 g/day from the standard 10-g North dietary fiber. Bethesda, MD: Life Sciences Research Office,
Federation of American Societies for Experimental Biology,
American diet can not only protect against CRC but also
potentially decrease cholesterol levels,12’ improve insulin 18. Harris PJ, Ferguson LR. Dietary fibre: its composition and role in
resistance,12” reduce blood pressure,‘29p130 and prevent protection against colorectal cancer. Mutat Res 1993;290:97-
heart disease.‘j’
19. Stephen AM, Haddad AC, Phillips S. Passage of carbohydrate
YOUNG-IN KIM, M.D., FRCP(C) into the colon: direct measurements in humans. Gastroenterol-
ogy 1983;85:589-595.
Division of Gastroenterology
20. Prosky L, Asp N-G, SchweizerTF, deVries JW, Furda I, Determina-
Department of Alea’icine tion of insoluble, soluble dietary fiber in foods and food
University of Tot-onto and St. AIichael’J Hospital products: interlaboratory study. J Assoc Off Anal Chem 1988;71:
Department of Nutritiona/ Sciences 1017-1023.
21. Englyst HN, Cummings JH. Improved method for measurement
University of Toronto of dietary fiber as the non-starch polysaccharides in plant foods.
Toroilto, Callada J Assoc Off Anal Chem 1988;71:808-814.
22. Englyst HN, Kingman SM. Dietary fiber and resistant starch: a
nutritional classification of plant polysaccharides. In: Kirtchev-
References sky D, Bonfield C, Anderson JW, eds. Dietary fiber chemistry,
1. Cancer facts and figures: 1998. Atlanta, GA: American Cancer physiology and health effects. New York: Plenum, 1990:49-65.
Society, 1998. 23. Lanza E, Jones DY, Block G, Kessler L. Dietary fiber intake in the
2. Potter JD, Slattery ML, Bostick RM, Gapstur SM. Colon cancer: a US population. Am J Clin Nutr 1987;46:790-797,
review of the epidemiology. Epidemiol Rev 1993;15:499-545. 24. Southgate DAT. How much and what classes of carbohydrate
3. Sandler RS. Epidemiology and risk factors for colorectal cancer. reach the colon. Eur J Cancer Prev 1998;7(suppl 2):S81-S82.
Gastroenterol Clin North Am 1996;25:717-735. 25. Briefel RR. Assessment of the US diet in national nutrition
4. Doll R, Peto R. The causes of cancer: quantitative estimates of surveys: national collaborative efforts and NHANES. Am J Clin
avoidable risks of cancer in the United States today. J Natl Nutr 1994;59(suppl):164S-167s.
Cancer lnst 1981;66:1191-1308. 26. Btirkitt DP. Relationship as a clue to causation. Lancet 1970;2:
5. Chu KC, Tarone RE, Chow W-H, Hankey BF, Gloeckler Ries IA. 1237-1240.
Temporal patterns in colorectal cancer incidence, survival, and 27. Burkitt DP. Epidemiology of cancer of the colon and rectum.
mortality from 1950 through 1990. J Nat1 Cancer lnst 1994;86: Cancer 1971;28:3-13.
997-1006. 28. Track B, Lanza E, Greenwald P. Dietary fiber, vegetables, and
6. Heiisouer KJ, Block G, Blumberg J, Diplock AT, Levine M, Marnett colon cancer: Critical review and meta-analyses of the epidemio-
LJ, Schulplein RJ, Spence JT, Simic MG. Summary of the round logic evidence. J Natl Cancer lnst 1990;82:650-661.
table discussion on strategies for cancer prevention: diet, food, 29. Howe GR, Benito E. Castelleto R, Cornee J, Eteve J, Gallagher
additives, supplements, and drugs. Cancer Res 1994;54(suppl): RP, lscovich JM, Deng-ao J, Kaaks R, Kune GA, Kune S, L’Abbe
2044S-2051s. K4, Lee HP, Lee hl, Miller AB, Peters RK, Potter JD, Riboli E,
7. McKeown-Eyssen G. Epidemiology of colorectal cancer revisited: Slattery ML, Trichopoulos D, Tuyns A, Tzonou A, Whittemore AS,
are serum triglycerides and/or plasma glucose associated with Wu-Williams AH, Shu Z. Dietary intake of fiber and decreased
risk? Cancer Epidemiol Biomarkers Prev 1994;3:687-695. risk of cancers of the colon and rectum: evidence from the
8. Giovannucci E. Insulin and colon cancer. Cancer Causes Control combined analysis of 13 case-control studies. J Natl Cancer lnst
1995;6:164-179. 1992;84:1887-1896.
9. Schatzkin A, Lanza E, Freedman LS, Tangrea J. Coopoei MR, 30. Friedenrech CM, Brant RF, Riboli E. Influence of methodologic
Marshall JR, Murphy PA, Selby JV, Shike M. Schade RR, Burt RW, factors in a pooled analysis of 13 casecontrol studies of
Kikendall W, Cahill J, PPT Study Group. The Polyp Prevention Trial colorectal cancer and dietary fiber. Epidemiology 1994:5:66-
I: rationale. design, recruitment, and baseline participant charac- 79.
teristics. Cancer Epidemiol Biomarkers Prev 1996;5:37!5-383. 31. Winawer SJ, Fletcher RH, Miller L, Godlee F, Stola: MH, Mulrow
10. ECP consensus panel on cereals and cancer. Consensus state- CD, Woolf SH, Glick SN, Ganiats TG, Bond JH, Rosen L. Zapka
ment on cereals, fibre and colorectal cancer and breast cancers. JG, Olsen SJ, Giardiello FM, Sisk JE, Van Antwerp R. Brown-Davis
Eur J Cancer Prev 1998;7(suppl2):Sl-S3. C, Marciniak DA, Mayer RJ. Colorectal cancer screening: clinical
11. Hill MJ. Cereals, cereal fibre and colorectal cancer risk: a review guidelines and rationale. Gastroenterology 1997;112:594-
of the epidemiologic literature. Eur J Cancer Prev 1998;7(suppl 642.
2):S5-SlO. 32. Hoff G. Moen IE. Trygg K. Frlich W. Sauar J, Vatn M, Gjone E.
12. Caygill CPJ, Charlett A, Hill MJ. Relationship between the intake Larsen S. Epidemiology of polyps of the rectum and sigmoid

colon: evaluation of nutritional factors. Gastroenterology 1986; naire: the effects of week-to-week variation in food consump
21:199-204. tion. Int J Epidemiol1989;18:858-867.
33. Macquart-Moulin G, Riboli E, Cornee J, Kaaks R. Berthezene P. 52. Rrmm EB, Giovannucci EL, Stampfer MJ, Cold& GA, Litin LB,
Colorectal polyps and diet: a case-control study in Marseilles. Int Willett WC. Reproducibility and validity of an expanded self.
J Cancer 1987;40:179-188. administered semiquantitative food frequency questionnaire
34. Kune GA, Kune S, Read A. McGowan K. Penfold C. Watson LF. among male health professionals. Am J Epidemiol 1992;135:
Colorectal polyps, diet, and family history of colorectal cancer: a 1114-1126.
case-control study. Nutr Cancer 1991;16:25-30. 53. Eastwood MA. The physiological effect of dietary fiber: an
35. Neugut AI, Garbowski GC, Lee WC, Murray T, Nieves JW, Forde update. Annu Rev Nutr 1992;12:19-35.
KA, Trat MR, Waye JD, Fenoglio-Preiser C. Dietary risk factors for 54. Schatzkin A, Freedman LS, Dorgan J, McShane LM, Schiffman
the incidence and recurrence of colorectal adenomatous polyps: MH, Dawsey SM. Surrogate end points in cancer research: a
a casecontrol study. Ann Intern Med 1993;118:91-95. critique. Cancer Epidemiol Biomarkers Prev 1996;5:947-9.53.
36. Sandler RS, Lyles CM, Peipins LA, McAuliffe CA, Woosley JT, 55. Einspahr JG, Alberts DS, Gapstur SM, Bostick RM, Emerson SS,
Kupper LL. Diet and risk of colorectal adenomas: macronutri- Gerner EW. Surrogate end-point biomarkers as measures of
ents, cholesterol, and fiber. J Nat1 Cancer lnst 1993;85:884- colon cancer risk and their use in cancer prevention trials.
891. Cancer Epidmiol Biomarkers Prev 1997;6:37-48.
37. Haile RW, Witte JW, Longnecker MP, Probst-Hensch N, Chen MJ, 56. Atkin WS, Morson BC, Cuzick J. Long-term risk of colorectal
Harper J, Frank1 HD. Lee ER. A sigmoidoscopy-based case cancer after excision of rectosigmoid adenomas. N Engl J Med
control study of polyps: macronutrients, fiber and meat consump 1992;326:658-662.
tion. Int J Cancer 1997;73:497-502. 57. Winawer SJ, Zauber AG, Ho MN, O’Brien MJ, Gottlieb LS,
38. Hirayama T. A large-scale cohort study on the relationship Sternberg SS, Waye JD, Schapiro M, Bond JH, Panish JF, Ackroyd
between diet and selected cancers of digestive organs, In: F, Shike M. Kurtz RC, Hornsby-Lewis L, Stewart ET, National
Bruce WR, Correa P, Lipkin M,Tannenbaum, Wilkins TD, eds. Polyp Study Workgroup. Prevention of colorectal cancer by
Banbury report 7. Gastrointestinal cancer: endogenous factors. colonoscopic polypectomy. N Engl J Med 1993;329:1977-
Cold Spring Harbor, New York: Cold Spring Harbor Laboratory, 1981.
198k409-429. 58. DeCosse JJ, Miller HH. Lesser ML. Effect of wheat fiber and
39. Kromhout D, Bosschieter EB, de Lezenne Coulander C. Dietary vitamin C and E on rectal polyps in patients with familial
fibre and l@year mortality from coronary heart disease, cancer, adenomatous polyps. J Natl Cancer lnst 1989;81:129C-1297.
and all causes. The Zutphen study. Lancet 1982;2:518-522. 59. Alberts DS. Einspahr J, Rees-McGee S, Ramanujam P. Buller
40. Phillips RN, Snowdon DA. Dietary relationships with fatal colorec- MK, Clark L, Ritenbaugh C, Atwood J, Pethigal P, Earnest 0, Villar
tal cancer among Seventh-Day Adventists. J Nati Cancer lnst H, Phelps J, Lipkin M. Wargovich M. Meyskens FL Jr. Effects of
1985;74:307-317. dietary wheat bran on rectal epithelial cell proliferation in
41. Stemmermann GN, Heilbrun LK. Nomura AM. Association of diet patients with resection for colorectal cancer. J Natl Cancer lnst
and other factors with adenomatous polyps of the large bowel: a 1990;82:1280-1285.
prospective autopsy study. Am J Clin Nutr 1988;47:312-317. 60. Alberts DS, Einspahr J, Ritenbaugh C, Aickin M, Rees-McGee S,
42. Thun MJ, Calle EE, Namboodiri MM, Flanders WD, Coates RJ, Atwood J, Emerson S, Mason-Liddil N, Bettinger L, Pate1 J,
Byers T, Boffetta P, Garfinkel L, Heath CW Jr. Risk factors for Bellapravalu S, Ramanujam PS, Phelps J, Clark L. The effect of
fatal colon cancer in large prospective study. J Nat1 Cancer lnst wheat bran fiber and calcium supplementation on rectal muco
1992;84:1491-1500. sal proliferation rates in patients with resected adenomatous
43. Willett WC, Stampfer MJ, Colditz GA, Rosner BA, Speizer FE. colorectal polyps. Cancer Epidemiol Biomarkers Prev 1997;6:
Relation of meat, fat, and fiber intake to the risk of colon cancer 161-169.
in a prospective study among women. N Engl J Med 1990;323: 61. Alberts DS. Ritenbaugh C, Story JA, Aickin bl, Rees-McGee S,
1664-1672. Buller MK, Atwood J. Phelps J, Ramanujam PS. Bellapravalu S,
44. Fuchs CS, Giovannucci EL, Colditz GA, Hunter DJ. Stampfer MJ, Pate1 J, Bettinger L, Clark L. Randomized, double-blinded placebo
Rosner B, Speizer FE, Willett WC. Dietary fiber and the risk of controlled study of effect of wheat bran fiber and calcium on
colorectal cancer and adenoma in women. N Engl J Med fecal bile acids in patients with resected adenomatous colon
1999;340:169-176. polyps. J Natl Cancer lnst 1996;88:81-92.
45. Giovannucci E, Rimm EB, Stampfer MJ, Colditz GA, Ascherio A, 62. Martinez ME, Reid ME, Guillen-Rodriguez J, hlashall JR, Sam-
Willett WC. Intake of fat, meat, and fiber in relation to risk of pliner R, Aickin hl, Ritenbaugh C, van Leeuwen B. Mason-Liddil
colon cancer in men. Cancer Res 1994;54:239C-2397. N, Gluliano A, Vargas PA, Alberts DS. Design and baseline
46. Giovannucci E. Stampfer MJ, Colditz G, Rimm EB, Willett WC. characteristics of study participants in the Wheat Bran Fiber
Relationship of diet to risk of colorectal adenoma in men. J Natl Trial. Cancer Epidemiol Biomarkers Prev 1998:7:813-816.
Cancer lnst 1992;84:91-98. 63. McKeown-Eyssen GE, Bright-See E, Bruce WR. Jazmaji V, the
47. Platz EA. Giovannucci E, Rimm EB, Rocked HRH, Stampfer MJ, Toronto Polyp Prevention Group. A randomized trial of a low fat
Cold& GA, Willett WC. Dietary fiber and distal colorectal high fibre diet in the recurrence of colorectal polyps. J Clin
adenoma in men. Cancer Epidemiol Biomarkers Prev 1997;6: Epidemiol 1994;47:525-536.
661-670. 64. MacLenna R, Macrae F, Bain C, Battistutta D. Chapuis P, Gratten
48. Steinmetz KA, Kushi LH, 6ostick RM, Folsom AR, Potter JD. H, Lambert J, Newland RC, Ngu M, Russell A, Ward M, Wahlqvist
Vegetables, fruit, and colon cancer in the lowa Women’s Health ML, Austra!ian Polyp Prevention Project. Randomized trial Of
Study. Am J Epidemiol 1994;139:1-15. intake of fat, fiber, and beta carotene to prevent colorectal
49. Wllett WC. Nutritional epidemiology. New York: Oxford, 1989. adenomas. J Natl Cancer lnst 1995;87:1760-1766.
50. Willett WC, Sampson L, Browne ML, Stampfer MJ, Rosner B, 65. Faivre J, Giacosa A. Primary prevention of colorectal cancer
Hennekens CH, Speizer FE. The use of a self-administered through fibre supplementation. Eur J Cancer Prev 1998:7(suppl
questionnaire to assess diet four years in the past. Am J 2):S29-s32.
Epidemiol1988;127:188-199. 66. Burn J, Chapman PD, Bertario L, Bishop DT, BUlow S, Cummings
51. Salvini S. Huner DJ. Sampson L, Stampfer MJ, Colditz GA, J, Mathers J, Phillips R, Vasen H. The protocol for a European
Rosner B, Willett WC. Food-based validation of a dietaryquestion- double-blind trial of aspririn and resistant starch in familial

adenomatous polyposis: The CAPP Study. Eur J Cancer 1995; 86. Roberton AM, Ferguson LR, Hollands HJ, Harris PJ. Adsorption
31A:1385-1386. of a hydrophobic mutagen to five contrasting dietary fiber
67. Englyst HN, Kingman SM. Hudson GJ, Cummings JH. Measure- preparations. Mutat Res 1991;262:195-202.
ment of resistant starch in vitro and in vivo. Br J Nutr 1996;75: 87. Harris PJ, Roberton AM, Watson ME, Triggs CM, Ferguson LR.
749-755. The effects of soiuble-fiber polysaccharides on the adsorption of
68. Phillips J, Muir JG, Birkett A, Lu ZX, Jones GP, O’Dea K, Young GP. a hydrophobic carcinogen to an insoluble dietary fiber. Nutr
Effect of resistant starch on fecal bulk and fermentation- Cancer 1993;19:43-54.
dependent events in humans. Am J Clin Nutr 1995;62:121- 88. Story JA, Kirtchevsky D. Comparison of the binding of various
130. bile acids and bile salts in vitro by several types of fiber. J Nutr
69. Cummings JH, Beatty ER, Kingman SM, Brngham SA, Englyst 1976;106:1292-1294.
HN. Digestion and physiologic properties of resistant starch in 89. Reddy BS, Sharma C, Simi 8, Engle A, Laakso K, Puska P,
the human large bowel. Br J Nutr 1996;75:733-747. Korpela R. Metabolic epidemiology of colon cancer: effect of
70. Hylla S, Gostner A, Dusel G. Anger H. Bartram HP, Christ1 SU, dietary fiber on fecal mutagens and bile acids in healthy
Kasper H, Scheppach W. Effects of resistant starch on the colon subjects. Cancer Res 1987;47:644-648.
in healthy volunteers: possible implications for cancer preven- 90. Bruce WR. Recent hypotheses for the origin of colon cancer.
tion. Am J Clin Nutr 1998;67:136-142. Cancer Res 1987;47:4237-4242.
71. Cassidy A, Bingham SA. Cummings JH. Starch intake and 91. MacDonald IA, Singh G, Mahony DE, Meier CE. Effect of pH on
colorectal cancer risk: an international comparison. Br J Cancer bile salt degradation by mixed fecal cultures. Steroids 1978;32:
1994;69:937-942. 245-256.
72. Caderni G, Luceri C, Spagnesi MT, Giannini A, Biggeri A, Dolara P. 92. Thorton JR. High colonic pH promotes colorectal cancer. Lancet
Dietary carbohydrates modify azoxymethane-induced intestinal 1981;1081-1082.
carcinogenesis in rats. J Nutr 1994;124:517-523. 93. Wargovich MJ, Eng VWS, Newmark HL. Calcium inhibits the
73. Sakamoto J, Nakaji S, Sugawara K, lwane S. Munakata A. damaging and compensatory proliferative effects of fatty acids
Comparison of resistant starch with cellulose diet on 1,2- on mouse colonic epithelium. Cancer Lett 1984;23:253-258.
dimethylhydrazine-induced colon carcinogenesis in rats. Gastro 94. Walker ARP, Walker BF, Walker NV. Faecal pH, dietary fiber intake
enterology 1996:11(3:116-120. and proneness to colon cancer in four south African populations.
74. Young GP, McIntyre A, Albert V, Folino M, Muir JG, Gibson PR. Br J Cancer 1986;53:489-495.
Wheat bran suppresses potato starch-potentiated colorectal 95. Jacobs LR. Influence of soluble fibers on experimental colon
tumorigenesis at the aberrant crypt stage in a rat model. carcinogenesis. In: Kritchevsky D, Bonfield C, Anderson JW, eds.
Gastroenterology 1996;110:508-514. Dietary fiber chemistry, physiology, and health effects. New
75. Fodde R, Edelmann W, Yang K. van Leeuwen C, Carlson C, York: Plenum, 1990:389-410.
Renault 8, Breukel C, Alt E, Lipkin M, Meera Khan P, Kucher- 96. Goldin BR, Gorbach SL. The relationship between diet and rat
lapati R. A targeted chain-termination in the mouse Apt gene fecal bacterial enzymes implicated in colon cancer. J Nati Cancer
results in multiple intestinal tumours. Proc Natl Acad Sci U S A lnst 1976:57:371-375.
1994;91:8969-8973. 97. Stephen AM, Cummings JH. Mechanism of action of dietary fibre
76. Burn J, Kartheuse A, Fodde R, Coaker J, Chapman P, Mathers JC. in the human colon. Nature 1980;284:283-284.
Intestinal tumours in the Apt 1638N mouse: aspirin not protec- 98. Hagopian HK, Riggs MG, Swartz LA, Ingram VM. Effect of
tive and resistant starch increases small bowel tumours (abstr). n-butyrate on DNA synthesis in chick fibroblasts and HeLa cells,
Eur J Hum Genet 1996;4:13. Cell 1977;12:855-860.
77. Scheppach W, Bartram HP, Richter F. Role of short-chain fatty 99. Kim YS, Tsao D, Siddiqui B, Whitehead JS, Arnstein P. Bennett J,
acids in the prevention of colorectal cancer. Eur J Cancer Hicks J. Effects of sodium butyrate and dimethylsulfoxide on
1995;31A:l077-1080. biochemical properties of human colon cancer cells. Cancer
78. Mortensen PB, Clausen MR. Short-chain fatty acids in the 1980;45:1185-1192.
human colon: relation to gastrointestinal health and disease. 100. Bernard JA. Warwick G. Butyrate rapidly induces growth inhibi-
Stand J Gastroenterol 1996;31:132-148. tion and differentiation in HT29 cells. Cell Growth Differ 1993;4:
79. McIntyre A, Gibson PR, Young GP. Butyrate production from 495-501.
dietary fibre and protection against large bowel cancer in a rat 101. Hague A, Manning AM, Hanlon KA, Huschtscha LI, Hart D,
model. Gut 1993;34:386-391. Paraskeva C. Sodium butyrate induces apoptosis in human
80. hledina V, Afonso JJ, Alvarez-Arguelles H. Hernandez C, Gonzalez colonic tumour cell lines in a p53independent pathway-
F. Sodium butyrate inhibits carcinoma development in a 1,2- implications for the possible role of dietary fibre in the preven-
dimethylhydrazine-induced rat colon cancer. JPEN 1998;22:14- tion of large bowel cancer. Int J Cancer 1993;55:498-505.
II. 102. Whitlock JP, Galeazzi D, Schulman H. Acetylation and calcium
81. D’Argenio G, Cosenza V, Cave MD, lovino P, Valle ND, Lombardi dependent phosphorylation of histone H3 in nuclei from butyrate-
G, Mazzacca G. Butyrate enemas in experimental colitis and treated HeLa cells. J Biol Chem 1983;258:1299-1304.
protection against large bowel cancer in a rat model. Gastroen- 103. Toscani A, Soprano DR, Soprano KJ. Molecular analysis of
terology 1996;110:1727-1734. sodium butyrate-induced growth arrest. Oncogene Res 1988;3:
82. Zoran DL, Turner ND, Taddeo SS. Chapkin RS, Lupton JR. Wheat 223-228.
bran diet reduces tumor incidence in a rat model of colon cancer 104. Kruh J, Tichonicky L. Defer N. Effect of butyrate on gene
independent of effects of distal luminal butyrate concentrations. expression. In: Binder HJ, Cummings JH, Soergel K, eds. Short
J Nutr 1997;127:2217-2225. chain fatty acids. Dordrecht, the Netherlands: Kluwer, 1934:135-
83. Klurfeld DM. Dietary fiber-mediated mechanisms in carcinogen- 147.
esis. Cancer Res 1992;52(suppl):2055s-2059s. 105. Watkins LF, Lewis LR, Levine AE. Characterization of the synergis-
84. Kritchevsky D. Cereal fibres and colorectal cancer: a search for tic effects of insulin and transferrin and the regulation of their
mechanisms. Eur J Cancer Prev 1998:7(suppl2):S33-S39. receptors on a human colon carcinoma cell line. Int J Cancer
85. Smith-Barbaro PD. Hanse D, Reddy BS. Carcinogen binding to 1990;45:372:375.
various types of dietary fiber. J Natl Cancer lnst 1981;67:495- 106. Bjork J, Nilsson J, Hultcrant R, Johnasson C. Growth-regulatory
497. effects of sensory neuropeptides, epidermal growth factor,

insulin, and somatostatin on the nontransformed intestinal 123. Barbone E, Austin H. Partridge EE. Diet and endometrial cancer:
epithelial cell line IEC-6 and the colon cancer cell line HT 29. a casecontrol study. Am J Epidemiol 1993;137:393-403.
Stand J Gastroenterol1993;28:879-884. 124. Potischman N, Swanson CA, Brinton LA, McAdams M, Barrett
107. MacDonald RS, Thornton WH, Bean TL. Insulin and IGF-1 RJ, Berman ML, Mortel R, Twiggs LB, Wilbank GO, Hoover RN.
receptors in a human intestinal adenocarcinoma cell line (CACO- Dietary associations in a case-control study of endometrial
2): regulation of NA+ glucose transport across the brush border. cancer. Cancer Causes Control 1993;4:239-250.
J Recept Res 1993;13:1093-1113. 125. Risch HA. Jain M. Marrett LD, Howe GR. Dietary fat intake and
108. Guo YS, Narayan S, Yallampalli C, Singh P. Characterization of risk of epithelial ovarian cancer. J Nati Cancer lnst 1994;86:
insulin-like growth factor 1 receptors in human colon cancer. 1409-1415.
Gastroenterology 1992;102:1101-1118. 126. Anderson S-O, Wolk A, Bergstrom R. Giovannucci E. Lindgren C,
109. Macaulay VM. Insulin-like growth factors and cancer. BrJ Cancer Baron J, Adami H-O. Energy, nutrient intake and prostate cancer
1992;65:311-320. risk: a population-based case-control study in Sweden. Int J
110. Cullen KJ, Yee D, Rosen N. Insulin-like growth factors in human Cancer 1996;68:716-722.
malignancy. Cancer invest 1991;9:443-454. 127. Jenkins DJ, Wolever TM, Kalmusky J, Guidici S, Giordano C,
111. Culouscou JM, Shoyah M. Purification of a colon cancer cell Pattern R, Wong GS, Bird JN, Hall M. Buckley G, et al.
growth inhibitor and its identification as an insulin-like growth Low-glycemic index diet in hyperlipidemia: use of traditional
factor binding protein. Cancer Res 1991;51:2813-2819. starchy foods. Am J Chin Nutr 1987;46:66-71.
112. Ritter MM, Richter WO, Schwandt P. Acromegaly and colon 128. Salmeron J, Manson JE. Stampfer MJ, Colditz GA, Wing AL,
cancer. Ann Intern Med 1987;106:636-637. Willett WC. Dietary fiber, glycemic load, and risk of non-insulin-
113. Kim YI. Diet, lifestyle, and colorectal cancer: is hyperinsulinemia dependent diabetes mellitus in women. JAMA 1997;277:472-
the missing link? Nutr Rev 1998;56:275-279. 477.
114. Weiderpas E, Gridley G, Nyren 0, Ekbom A, Persson I, Adami 129. Ascherio A, Hennekens C, Willett WC, Sacks F, Rosner 6,
H-O. Diabetes mellitus and risk of large bowel cancer. J Nat1 Manson J, Witteman J. Stampfer MJ. Prospective study of
Cancer lnst 1997;89:660-661. nutritional factor, blood pressure, and hypertension among US
115. Will JC, Galuska DA, Vinicor F, Calle EE. Colorectal cancer: women. Hypertension 1996;27:1065-1072.
another complication of diabetes mellitus? Am J Epidemiol 130. Ascherio A, Rimm EB, Giovannucci EL, Colditz GA, Rosner B,
1998;147:816-825. Willett WC, Sachs F, Stampfer MJ. A prospective study of
116. Tran TT, Medline A, Bruce WR. Insulin promotion of colon tumors nutritional factors and hypertension among US men. Circulation
in rats. Cancer Epidemiol Biomarkers Prev 1997;5:1013-1015. 1992;86:1475-1484.
117. Corpet DE, Jacquinet C, Peiffer G. Tache S. Insulin injections 131. Rimm EB, Ascherio A. Giovannucci E. Spiegelman D, Stampfer
promote the growth of aberrant crypt foci in the colon of rats. MJ, Willett WC. Vegetable, fruit, and cereal fiber intake and risk
Nutr Cancer 1997;27:316-320. of coronary heart disease among men. JAMA 1996;275:447-
118. Riccardi G, Rivellese AA. Effects of dietary fiber and carbohy- 451.
drate on glucose and lipoprotein metabolism in diabetic pa-
tients. Diabetes Care 1991;14:1115-1125.
119. WCRF Panel. Diet, nutrition, and the prevention of cancer: a
global perspective. Washington, DC: World Cancer Research Address requests for reprints to: Chair, Clinical Practice and
Fund/American Institute for Cancer Research, 1997. Practice Economics Committee, AGA National Office, c/o Member-
120. Potter JO. Fiber and colorectal cancer-Where to now? N Engl J ship Department, 7910 Woodmont Avenue, 7th Floor, Bethesda,
Med 1999;340:223-224. Maryland 20815. Fax: (301) 654-5920.
121. Steinmetz KA, Potter JD. Vegetables, fruit, and cancer. I. The Clinical Practice and Practice Economics Committee acknowl-
Epidemiology. Cancer Causes Control 1991;2:325-357. edges the following individuals, whose critiques of this review paper
122. Grber M. Fibre and breast cancer. Eur J Cancer Prev 1998; provided valuable guidance to the authors: Graeme P. Young, M.D.,
7(suppl2):S63-S67. Joel Mason, M.D., and James J. Cerda, M.D.
The New England. Journal o’f Mcdicinc
,--.. .. P..
-.. . __-_. .,.- P.

Correspondence summary, our study’providcd no cvidc
- in fat and high in libcr, fruits, and VI:
risk of recurrent colorcctal adcnomu.
thL effect appcarcd widely kl the media,
pie u-ich the b&t that diet makes no di
fact, WCdo not yet know whcchcr this is

High-Fiber Diet and Colorectal Adenomas 1. Schatzkin A, L;lnl.a E, Corlc D, et al. Lack of cffecr
libcr dicr on the rcwncnce of colorccu1 adcnoma. N
To rh Edirvr: I am conccmcd that the report by Schaakin 342:1149.$5.
and ~llcagucs (April 20 issue)’ may lcad many pcoplc KO
conclude ctroncously that diet does not affccr Rxcptibiliry
to colon cancer, rvcn though the hypothesis was not adc-
quatcly tcstcd. A--or iiny KOTable 3 of the report, dices fit
dccrcascd from 35.6 percent at randomization to 23.8 per
cent at four ycxs in the inccrvenrion group and from 36.0
to 33.9 pcrccn~ in the control group. Clonsumptinn of red
meat dccrcucd from 93.2 to 74.5 g per day in the inrcrvcn-
rion group and from 97.9 to 94.9 g per day in the control
group. Howcvcr, plasma total cholcsrcrol decrcascd only
from 5.30 to 5.27 mg per da3litc.r in the intcrvcntion goup
and from 5.29 to 5.27 mg per deciliter in the control goup itux of calories. Rrducrions in activity are
(Irjg-uansformcd valets); thcsc drmascs rcprwnt 2 per.
cent and 1 pcrccnt in rhc absolute cllolestcrol conccntra-
rions, rcspectivcly. Reductions ofthis magnirudc in the in-
take of dictxy fit and nd meat in the intcmntion group (if
they really occurred) .should have caused a grcatcr rcducrion
in the plasma total cholesterol lcvd- in any cvcnt, grcxcr
than that in the control group, which did not occur. Also,
weight was csscnti$ unchanged in both groups.
The logical conclusion is that the patients in the inter.
vcntion group u’crc actually consuming a diet very similar
to that of tic control group. This would not bc surprising,
Gncc it is difficult to motivate people to m&c and maintain
dietary changes in large-scale studies, and it il; equally dif-
ficult to obtain accuc:c dic~ary information in clinical trials.
Hut it is as crron;xA tu claim char dietary fat and choles-
rrrol intic ha\< no cffcct vn colon cmccr as it is to say that
they have no cffcct cm plasma tc)tal cholcstrrul. 1. Rollo SJ, Bell EA. Diary apprwckr tv the ucxmcn
‘It js a prcat disservice for the authors to conclude, “In CL-8North Am 2000;84:463 -18.

--. --- .- - --. .. I’
Lcrur~ tv rhc Editor arc considcxd for publication (rubjccr to editing and nbri&mcm) prwidcd they do not contin maria1
wbmirtcd or publrth:d clsr;whcrc. Plcarc noa rhc folluwing: -Yrwr lctw must bc typcwritwn and triple-spaced. l Irs text, not
rcfcrcnccs, mutt not cxcccd 400 wor& (plrwr include 1 uwd count). *It met hwc no more thw live rcfcrenccr and one Rgurc or table.
-It should not be signed by mnrc than rhrcc authors. l Lcrtcn- referring to a ~CCCIILJwnul article mw be rcccivcd within four WC& of its
publication. *PIcare include your full addrrnc, t&phone numbcz, and fir number (if you ha\r one). *You may send us )vur lctcc by port, hx,
or clcclnnic mail.
Our ad&or, Lcrtcrs to the Editor l NW EngLandJournul ufMsdicinc * 10 Shattuck St.. Roston, hfA 02115
Our Lx numhcrs; 617-739-Y864 and 617-734-4457
Our c-mail address:
WC cznwt ~knowlcdgc receipt of your Icrtcr, but w: will notify you when WChrvc mdc a d&ion about publication. 1%‘~1%
provide prcpublicarion proofs. Plcasc cnolusc L rc~mpd, ielf-addrcasxl cn~~lopc if you UWII unpublLhcd marcri.ll mwmcd zo y
Finnncid associa[iw1s or orhcr pvrdblc conflicts of intercv tnu~t bc disclosed. Submit&n uf a lcrtcr cimrritutcr pcrmir&n for
rats Mcdicll Society, its liccnscclr, 2nd itu rstipnccs to use ic in the ]ourxd’t various editions (print, dlta base, and oprical disk) .nd ilr
wxhologicr, rcvirions w-v2 any oticr fnrm or medium.

736 * Scprcmhcr 7, 2000
--.. . I. - . ._--- _._..

7t C/JCJ:'dihw Most of the explanations for the disap- cd number cjfcxpeacd cancers in chc combi
pointing rcsulrs of the studies rcportcd by Schatzkin ct al.” p)Ups WdS OOly 6.4, as
and flbcrts ct ,Q.).~ arc ncjred in the accompanying. editorial bcr ofl7, ~+Miny a risk rat
by Bycrs.’ One qucsrion has not been addrcsscd:,‘vVas cum- number of bowtl car~ct’rs
pliancc with the rcgimcn sufficicntJy asccrnincd? In both to bc significant at the 95
rrudics, the asscssmcnt of compliance rclicd mainly on the that the risk. of hl,wcl
srudy pnrticipants, whu eitllcr r~~urncd scroll bores and rc-
c&cd supplcmcnr cansumprior\ on a calcndar,2 or rcport-
cd daily intake ot’ food and srlppkment\.’
The incidcncc ofsomc advcrsc puointcstinal cffccts can
bc considcrcd an indcpcndcnt marker of compliance with
rhc USCof the high-fiber supplcmcnt in the srudy by Albcru
cl: a!.; the incidcncc was sipnifitandy hitier in the high-fiber
grwp than in the low-fiber group. ‘1l’hus, in this study, one
can say thar insoluble fiber alone dots not account for all
the effcrts that have been attributed to a diet high in fruit
and vcgcrablcs in obscrvJciona1 srudics and has no short-
term bcnefir in reducing the risk of color~ctd adcnoma.
In chc study by Schatzkjn ct al., thcrc wcrc two indcpcnd-
cm markers of compliance: wci&c loss and sccum total ca-
rorcnoids. (The cntoccnoid vJucs in Tahlc 3 oftheir article To rbr Ediror: Is it possibl
appear co lrc 100 time: the usual v~lucs.) A!thouf;h sipnif- ct al. rhc chvicc ofsubjects
icant, rhc changes ..CK very small after four years, and it is adcnonlas that wcrc potcn
unfkcunarc th3; nc, additional biolc@c markers were inv~~- ccr, had a lcvcling ctYcc~c
tigarcd and thar such invcscigations wcrc not pcrformcd tnced by dicrtary change?
mote of&. Thus, it is difficult to say that compliance was

75 rhc .i%%ur:Although the subjcc
group in the srudy by Schatzkin ct al.
al information and cuuoxling in Or&
percent of toal calories from fat, ch
the report dots not star whcthcr the
To rhc Hiray: 1s it possible rhat a high-fiber diet may bc
harmful? The main cncl point in both the study by Schatz-
kin ct al. and the study by Alhcrts tr al. was the recurrcncc
of polyps, but an PC~ norc imporranr consideration is the
cffcct of a hiyh-fiber diet on the occucrcncc of bowel can-
CU. In each study, rhcrc appcarcd to be more bowel can.
ccrs in.thc high-fiber group than in the low-fiber or COI>UO~
01, in the control and intcrvcnrion groups, h
group: 10 casts as compared with 4 in tic study by Schara- lipid> 315 not n0tCii.
kin cr al. 3rd 7 cnuos as co,mparcd \virh 2 in the study by I think an as?crsmcnr ot’n-3 lipids must
,%lbcrrr cr ai. account in evaluating the cffcct~ of dictzry i
In the two studies, the risk ratios for bowel cancer in on the rate of rccurrcncc of colonic adcnoma
the high-fiber groups, as compared with the low-fiber and
control groups, appear to bc similti, and rhc charactcris-
tics of chc paticncs in chc two group, wcrc similar. ‘Thcrc-
fox, WCasrusscd ~hc risk of bowel anccr in rhc high-fiber
groups and in the I~Jw-tih;r and control groups orb com-
bining the data from the two studies. R’c used the avail-
nblc information in the two studies to cstimarc the approx-;c number ofpcrson-years for rhc combined high-fiber
groups and the combined low-fiber and cont.rol groups.
WC awmwl that the ccjmbincd high-libcr groups should
have had the snmc frcqucncy of bowel cnnccr as the COIII-
bincd low-fiber and control groups. Howcvcr, the csrimat-
.__ New England
-.--A-- Journal of 1_.-
Mcdicinc -‘-.‘-.

rrol group5 was minimaL As w scared in Orlr report, our h tit
study could not rule out rhc possibility rhar greater rcduc.
tions in fat and red mcar or further increases in fiber and
fruits and vcsctablcs might bc rcquictd to rcducc rhc risk of of clinici! Ui.
colorectal ncoplaric.
‘Jhc absoluts di&cncc in the change tl caloric intake hc- Dicrary factors
twccn the inurvcntion and control groups uas -25 Cal (95
pcrscnt coniidcncc interval, -72 to 22). Thcrcforc, tic trial
did not establish a significanr.bct~vecn-proup diffcrcncc in To tbz Edir~r: WC agree wirh Dr. Gcrbcl rhc gastro-
the change in rncrgy consumption. hlurca~, the lilrrlihood intestinal side cffccts caused by rhc bily ir of 13.5 g of
ofundctrcporc;~g c#cncrgy intake in dicrary-intcflenrivn wheat-bran cereal arc an indcpcndcnr rr& ‘compliance
stud& would explain, at least in pat;, the apparent discrcp- with chc use of the high-fibs supplcmcnr il trial. In our
ancy bcrwcen changes in rcportcd cncrgy inrakc and ob- phase I and 2 studies of intcrvcntions invo wheat-bran
scrvcd weight loss in this and other dietary trials.’ fiber, wc documented that 13.5 y of this su lent per &y
WC arc conducting the rt;pr trfobscnabnnai analysessug- could bc takrn with a rcasonablc lcvcl of CC mc by old-
gcstcd by Dr. Davis. All such analysts, homer, arc subject cr study pazricipants for periods of a few :hs.‘azHow-
to confounding: pcoplc who adhcrc to an intervention arc cvcr, some older persons may nor bc able >lrratc high
often found co be sysrcmarically diffcrcnt from rhosc who doses of wheat-bran fiber on a daily basis xrai years.
du not in ways thar arc rclarcd to rhc clinical o\ncomc. WC Lowcnfds and Maisonne~c ague that yh-hbcr cc-
prcsrntcd the K.&J of an inrcnrion-to-rrcat analysis, which ml promoccd the dczrlopmrnr of colorer UICCI.How-
is an intcrnariondly acccptcd method of analysis.’ cvcr, our findings do not support this a~ lr. Scvcn of
R’c did find significant net incrcascs in lutein, alpha car- the nine colorc:d cancers that wcrc dctc occurred in
otcnc, and bcu carot.cnc in the intcrvcncion group. The rcl- members of the h&h-fiber group, but rhr rhcsc scvcn
ativcly small (though statistically significant) incccasc in ca- ancrrs wcrc diagnosed 8,10, and 11 mnn tcr random-
rotcnoids may rcflcct thr facts that carutcaoid-rich fruits ization; thcsc cancers wcrc probably missed yw~
and vcgctablcs accounted for only abuut half the total in- colonoscopy (i.e., bcforc tic start nfchc inc
crcasc in fruits and vcgctablcs in the intctvcntion group and only 4 of the 719 patients in the high-fibc up and 2 oi
that catotcnoidp from fruit< and vrgctablcs are subctanrial- rhc 584 in the low-fiber group had collm :anccrs that
Jy less bjoavailablc than rh~c f-rum supplcmcncs.’ Dr. Gcrbcr wcrc detected at lcasc 1 year after randomi ‘i (range. 19
corrccrly pvinu out two crtors in the unirs ofmcaswcmcnt to 39 months). The cliffcrcncc is nor statis’ I significant.
for carotcnoid\ in our article: in chc rhird fclocnotc ro Ta-
blc 1, the unit ofmcasurcmcnc should be micromoles per
liter, and in Tabl: 3 it should be micrograms per dccilitcr.
If WCclimi:.rtc the cancers diagnosed within the first year AILsH
after cnrollmcnt, which. wcrc likely to have bcrn prc%cnr
when the intcrvcncions bcgn (six in the intcr+rntion group .kSOI
and cwo in the control group in our study, and three in the
high-fiber group in the study by Albcrts CCal.). thcrc wcrc
a total of nine cancers in rhc high-fiber groups and four in
rhc control and low-fiber groups -- nor a srarktically siy-
niticanr diffcrcncc. A thorough invcstigaciun of rhc effects
of&r on chc risk ofcof(>rcctal cancer in inccrvention stud-
its rcquircs rhc poolins of data From all rhc trials that have
lvnkcd at rhis issue.
WC agtcc with Dr. Mullet: our trial could not determine
whcthcr llictary modification at%crs the risk of c~lotr~~l Echiiococcosis - An Emerging iseasc
adcnoma in pcrhons who have not had a prc.vic~s ;ulenoma. is Farmers
In bnc wirh Dr. lXptcy’s suggcsdon, wc will be anaJyzi.ny P
the dietary data from our study for n-3 fatty acids. Thcrc To rbc Edirov: Two echinococcus spec s - Echiocuccus
was cdy a smatl, though sratisrically significanr, di&cncc multiloculati and E. panulosur - arc ncnvn to exist in
in the consumption of fish bctwccn the inccrvcntion and ccnrral Europe and to cwsc alveolar an % cysdc cchinocor-
conrrol groups (21.5 and I X.6 y, rcsp+~cly). cosis, rcsp&ivcly in humus. 5% rcpvr a’high ptctalcn~c
of antibodies aeainsr thcsc orcanisms i d farmers.
Kational Caner Insrirute
Y,c&Jda, bill 20832-7232 women; mean 3sc, 41 years

L~u~~scs FXEEDXW,PH.D.
Rar Il?n U~+~rsiry
Kamar Gm 52900, Irncl

1. Hcgrtcd D>l, Kritchcvzky 11.Dicr and $cru~nlipid conccntr~ionr:
whcr; WCwe! Am J Clin Nutr 1997;65:lYYZ~6.
As~~~t~~s~m RESEARCH 19:36SI-3684 (1999)

Prevention of Colon Carcinogenesis by Components
of Dietary Fiber

Di\Gon ofNulltritiotla1 Carcinogenesis, American Heal/h Fowrdafiotl, C’alhalia, NtwmYork, 10595, U.S.A.

Al)5[t.iiCt. CfltlCer of thecob k olleof he hdhg Callses of studies have demonstrated that -increased consumption of
catIcer death in Westem counlries and is increasing rapidlj itI fruits and vegetables and high intake of dietary fiber reduce
J~,)~~~~. Epidetniological and laborator) atlima made1 studies the risk of colon cancer [3]. Interest in the concept of cancer
ha1.eslfsesred an imlerse relationship behveen colon cancer risk prevention is growing rapidly because the utilization of
atlll itltake of fiber-rich foods. The protective qffecr af dietaty nutritional factors and naturally-occurring and synthetic
fiber which comprises a heterogeneous group of nonsrarch agents that can protect against the development and
,t~;!\‘accllaridf,s such as cellulose, hemicellrdose, and pectin atui progression of carcinogenic process is not only an attractive
trot!: rrrl~oh~drate substances s~lch as phytic acid depends 011the but plausible approach to either inhibit or reverse
na/i,rc attd source ofjiber it1 the diet. Laboraroty animal tnadr!s carcinogenesis.
hare coruisletll[) showw that dietary administration of Icheat
brat! reduced colors twnarigenrsis. Human diet in~et~etltion Dietary Fiber and Colon Cancer
snrdies halIe demanstratrd that srlpplemet~tal wheul bran in the
die{ decreased the fonnaliotr of prltatir*e metabolites swh as The hypothesis that a diet high in fiber may protect against
secot&r>, bile acids and diacylg~vcerol in the colon thol hew colon cancer was first proposed by Burkitt 141who observed
bc,eti s/lo)\ tl to act as tumor promoters itI [he colon. Among Ihe that African Blacks consuming high fibrous and
corrtyuents of dietat), fiber; especial4 \\,hear bt.atz, phyric acid lo\v-fat foods had lower death rates due to colon cancer
(itrrnilo! hexaplrospha~e) has been studied errensiL,eb for its compared to their uhite counterparts eating a lou-fiber and
clicttioprevet~~i~~eprop&es agaitirr colotr carcitiogenesis iti rhe high fat diets. Subsequent studies demonstrated that, in
labotnroty atzimal models. 61 siudies cat-tied out to date, di?taty populations consuming diets high in total fat, the intake of
phyic acid reduced the incidetlce of colottic abetTatlr ctypt foci, diets high in total fiber, fibrous foods, and certain bvholegrain
palaliw pretleoplastic lesions iti rats. 01x21 atimitiisWaliot~ of foods reduce risk for colon cancer [5,6]. Intracountry
phyic acid was shonw 10 itlhibit colon carcin0genesi.r it1 rodetlts comparisons of dietary fiber and colon cancer mortality rates
dwitlg rile initiation and porrinitiariotl stages. These srudier strongly supported the hypothesis that dietary fiber, especially
pr0 i!lc elidetrce for po~etiCa1 clrettloprel,etrril,e pt.operries of fiber from cereal sources and pulses, protects against colon
ph)lic acid agaitis! colon cancer. ItS!h regard I0 mode of acfion, cancer [7]. Case-control studies on the relationship behveen
phyic acid acts as an anCoxidatlt, to reduce the rate of cell the dietary fiber and colon cancer provided convincing
prol$rarion atrd lo arrgtnetJ[ rhe imttwne response b), etlhancit~g results. Out of 19 case-control studies to assessthe role of
Ihe flcriCq, of natltral killer (h’K) cellr. fiber and fiber-containing foods, 3 studies reported no
protective effect, 2 found an increased risk. and 13 studies
Cancer of the colon and rectum is the fourth most common reported a protective effect of fiber-containing foods and
cause of cancer deaths world~vidc [l]. Cancer of the colon vegetables [S]. Hoi5.e et al [9] examined the results of
uhi:‘- k one of the leading causes of cancer deaths in both combined analysis of 13 case control studies of diet and colon
mell :~nd women in the LYestern count& including A’orth cancer ivith respect to the intakes of dietan, fibsr. In this
America [2] is generallv increasing rapidly in Japan including analysis, the individual data record’s for 5X7 colon cancer
the urban area’s of tie developing world. Epidemiological cases and 10170 control subjects have been pooled for a
common analysis which provided substantive evidence that
intake of fiber-rich foods is inversely related to colon cancer
risk with odds ratios of 1.0, O.S,0.7, 0.6. 0.5 for each quintilc
of consumption from lowest to highest. Similar findings have
~~n-~vuxufo~cc 10; Dr. Bandaru S. Reddv, American Health
Founi!tion, I Dana Road Valhalla. N.>‘. ld595, USA. been reported for a meta-analysis of 16 case-control studies.
15ith odds ratio of 0.6 for the highest versus loives intake of
J$ Ilbrdr: co1 on cancer, dicta9 fitxr, phytic acid. fiber [3].

‘-‘250-7005199$2.00+ .40 3681

Laboratory animal model studies also indicated that the grains and legumes is inversely associated \vith colon cancel c
protective effects of dietav fiber depends on the type of risk. This finding is significant because cereals, grains and it
fiber; wheat bran, hut neither corn bran nor oat bran, appears legumes are a rich source of ph! tic acid. It is possible that one iI
to inhibit color. tumor development [IO-111. The effect of of the mechanisms by bvhich dietary fiber inhibits colon ;I
dietary ivheat bran at Ii% level or corn bran plus 5% dietaq carcinogenesis is through the effects of phytic acid on cell \i
fat on colon carcinogenesis induced by azoqmethane (A051) proliferation and differentiation. I1
or 3,2’-dimethyl--l-aminobiphenyl (DhiBA) was studied in Ph!,tic acid and inositol have been tested ;I; (I
male F3-N rats. The composition of diets was adjusted so that chemoprecentive agents in ijf Iitta systems and laborator! (!
all the animals in different experimental groups consumed animal moJels for colon cancer. Sakamoto et al [?I] b
approximately the same amount of protein, fat, minerals, and in\.estigated the effect of phytic acid on proliferation and ir
vitamins. The animals fed wheat bran had a loucr incidence differentiation of human cancer cell line, HT-29 01 \*irro. 0
(number of animals bvith tumors) and multiplicity (number of These results shelved that phytic acid inhibits cell L-
tumors.!animal) of colon tumors than did those fed the control proliferation and concomitanti! increases differentiation I’
diet ivhereas corn bran or oat bran had no effect. Thus animal suggesting that it suppresses not only the malignant \i
model studies clearI), suggest that wheat bran consistentl} phenotype but also allo\vs the maturation of human colk)n ir;
inhibits colon carcinogenesis associated with administration cancer cells to structurally and behaviorally resemble normal I LI

of colon-specific carcinogens. cells. In ill l*irm studies, phytic acid reduced cell proliferation P!
In human clinical trials, supplements of ivheat bran of all human and rodent cell lines tested, including MC-7 di
produced a reduction in the incidence of rectal polyps among human breast carcinoma cells (201. Enhanced differentiation \I (
the individuals genetically predisposed to these lesions [Is]. of cancer cells to the point of reversion back to norrnal II :
hietabolic epidemiologic studies demonstrated that the phenotype was also observed in several lines, including the dr
individuals consuming high fat and low fiber diets excrete HT-29 human colon carcinoma cell line [2]. These studies iii,
increased levels of fecal mutagens and bile acids compared provide evidence for many poiential beneficial actions of
with those consuming IOX fat and high fiber or high fat and phytic acid.
high fiber diets [16,17]. Additional studies have also provided The exact mechanisms by ivhich phytic acid exert its
evidence that wheat-bran suppletnentation favorably altered a chemopreventive effects have not been clearly demonstrated. AI
number of biomarkers that are related to the risk of Because of the highly charged nature of phytic acid, it \vas CO
colorectal cancer including fecal mutagenicity [ 161, fecal thought that it could not be transported inside the cell [ZO]; St1

secondaq bile acids and bacterial 7u-dehydroxylase [17,1S] holyever, Sakamoto er al (31demonstrated that intragastricallq rc(
and rectal cell proliferation [19]. Dietary oat bran had no administered [3H]phytic acid \vas absorbed from the stomach iii:
effect on fecal secondary bile acids or 7a-dzhydroqlase and upper small intestine, distributed into various organs and
activity, ivhereas dietary .corn bran increased the levels of appeared in the plasma and urine as inositol and inositol PI.
secondary bile acids and 7u-dehydroxylase activity. hlore indicating metabolism of the parent compo.und phytic acid. hi:
recent studies have compared the effects of altering both fiber Phytic acid has been sho\vn to act as an antioxidant, to control gri
and fat content on fecal secondary bile acids. In this study, cell division and reduce the rate of cell proliferation, and to
healthy subjects who had consumed a typical high fat, low- enhance the activity of natural killer cells, \vhich play an Ke
fiber Western diet and were sivitched to a lokv-fat, very-low- important role in the host defense against neoplasia [20].
fiber diet and then to a lo\v-fat, high-fiber diet shoaed a Chemopreventive activity of phytic acid has been evaluawd
dramatic reduction in secondar) bile acids during the low-fat in preclinical animal models. Aberrant cvpt foci (ACF) are
and high fiber period, compared with the highfat and low- recognized as early preneoplastic lesions in the colon from
fiber period. In this connection, several lines of evidence shoM tvhich adenomas and adenocarcinomas may develop in the
that dietary fiber affects the metabolic activity of gut colon of both rodents and humans. There is evidence that
microflora; this effect also depends on the t>.pe of fiber several inhibitors of ACF formation reduce the incidence of
consumed 1171. There is con\,incing evidence that these colon tumors in laboratory animal models supgesting that
secondan, bile acids such as deox>,cholic acid and lithocholic ACF can be used to evaluate novel agents for their potenrial .i .I

acid act as colon tumor promoters. The evidence thus far chemoprerentive activities against colon cancer [23]. In this
generated suggests that high dietary fiber including \vheat connecrion, Pretlow et al [21] demonstrated that the (1 I
bran reduce the risk of color, cancer. development of larger ACF with I .or mow aberrant c

cn,pt!focur \vas significantly inhibited in F3-M rats t

Inositol Hexaphosphate administered AOhl and given 2% phytic acid in drinkin:I 7 >

\vater. Phytic acid at 1 and 2% isvels in the diet significantl.\; ti;
Inositol hexaphosphate (InsPh. phytic acid) is a natural]!, decreased the number of ACF in the colon [Xl. Results also I

occurring compound found in substantial amounts in cereals showed that 2% phytic acid administered in combination \\ith 9 1

and legumes [20]. As discussed above, intake of several 2% green tea extract had a synergistic effect exhibiting a total
classesof foods with high fiber content, and intake of cereals, of about 30% reduction in ACF (~~0.02) ivhereas green tea


czIr;,cI nlone had marginal effect (p<O.ll). Colon tumor-
inhil,it<>r\ activity of ph!tic \vid in
has alSo been evaluated
;llli,li;tl hodcls. Ullah and Shnmsuddin [?O] sho\Ved that
nt 0. I and I .O’T;phytic acid in drinkinp water
,i~i,itic;lnt/) inhibited AOM-induced colon tumor incidence, I I \Vatanahc lip RcJ$ BS. \!‘eicburger JH. Kritshc\sk! D: Effect (if
t,il,l ,Tlisity and size. Administration of 1% phytic acid in dietap alfalfa. pcc’~n and uheat bran on or m+l-
nitr~)sotIrc’;t-illclttced colt)n cnrcinogene\ir in F3-l-l rat,. J N;,tl Cancer
drilj!iill~ \vater reduced colon tumor multiplicity
by 52%
as lnst63. 151.15h. IY7Y.
.itorv (p<l).~)~). tumor frequency by 56% (p<O.OOI) and tumor size 12 Kedd) BS. hlori H: Effect of dietny bhsat bran and dch~drated
by f,zc; (~<O.OOl): 0.1% phltic acid eshibitecl only reduction citrus fiber on .:.2’-dinicth!I--l-aminirhiphen~l-induced intestinal carci-
in tumor size ‘P).71% (p<O.OOl). In another study. the effect nopenesic in F3-IJ rats. Carcinogenesis 2: 21-25. IYSI.
and 13 Redd! BS. hlori H. h’icolak hl: Effect of dietary \$hent bran and
I.ill.0. ,,f pl\y\ic acid administered during the postinitiation stage of
dehvdrated citrus fiha on azo?msthnne-induced intrctinol
,-ol~ltl carcinogenesis ~vas investigated by Shamsuddin and
CCll carsinogsnesis in F?-!l r,+ts. J Xatl Canca Inst 66; 553-557, IV,YI.
~~11;,1~[37]. Phytic acid when adminktered in drinking water 2 I-I Reddy BS. hlacura 1.. \\‘ayman hl: Effect of dictan corn bran and
NC‘\ .: or 5 months after AOM treatment significantI} autohydrolyzed lignin on 3.2.-dim~th!-l-1-aIminohiphen~l~induced
colon inllil>ilcd colon tumor multiplicity. tumor incidence and intestinal carcinogenesis in male Fischa 3-11 rats. J Katl Cancer Inst
tunior size in F344 rats suggesting that the beneficial action of 71: 41Y-1’3. lYS3.
rmal 15 DrCosse J. hlila HEI. Lesser hlL: Effect of wheat fiber and vitamins
ation ph!tic acid is not restricted to the prevention of tumor C and E on rectal pol!ps in patknts with familial adenomatous
development but perhaps to treatment of existing tumors as polyposis, J Nat1 Cancer Inst 81: 1290-1297. 19S9.
\vell [27]. In support of these results, Pretlow CI nl 1241 have 16 Reddy 8. Engle A. Katsifis S, Simi B. el ai Biochemical epidemiolog)
aLso demonstrated that administration of 2% phytic acid in of colon cancer: effect of types of dietary fiber on fecal mutagens, and
rrnal acid, and neutral sterols in health! subjects. Cancer Res 49: 4629.
(frillking Lvater during posiinitiation
stage suppressed AOM-
; the 4635. 1980.
int\il.:cd colon tumor incidence (p<O.O04). in F314 rats. I7 Red+ BS, Engle A, Simi B, Goldman hf: Effect of dietary fiber on
IS of colonic bacterial enqmes and bile acids in relation to colon cancer.
Conclusions Gnstroenterol 102: I-t?%1-W. 1992.
IS Atherts DS, Ritenbaugh C, Story JA. er nl: Randomized double-
.t its blinded. placebo- controlled study of effect of wheat bran fiber and
ated. Animal model studies clearly suggest that wheat bran
calcium on fecal bile acid, in patient> with resected adenomatous
\vas consistently inhibits colon carcinogens& Case-control colon polyps. 3 NatI Cancer lnst SS: Sl-92 1996.
[X]; studies show reasonnbl~ strone evidence that dietary fiber 19 Arhcrti DS, Einspahr 1, Rees hlcGee S. C( a/: Effects of dietary wheat

ically rctluccs the risk of colon cancer in humans. Dietary hran fiber on rectal epithctlial cell proliferation in patients with
ini,,,?.cntion studies provide evidence that Icheat bran resection for colorectat cwcers. J Kiatl Cancer Inst 82: 1250-1285,
nxh 1990.
> and suppkmcntation decreases the levels of several putative
20 Shamsuddin Ahl: Inositol phosphates have novel anticancer function.
tumor promoters in the colon. Administration of phytic acid, J Kutr 12-i: 72X-732s. lYY5.
11 PI,
hiph Icvels of which are present in wheat bran and other 21 Sakamoto K. Venkatraman G, Shnmsuddin A>l: Gmnth inhibition
grains inhibits colon carcinogenesis in animal models. and diffsrentiatinn of F-IT-29 crltr i/r ~.irjo by inositol hexaphosphatz
(phytic acid). Carcinogenzsis l-1; ISl.S-ISI9. 1993.
Id to 22 Sakamoto K;. Vucenik 1. Shamcuddln AXI: (‘H]Phytic acid (inositol
y an References
hexaphosphats) ic alxorbcd and dictrihuted to various tissues in rats.
J Nutr I?.?: 7l?-7211, IY93.
1 \’ tlr1d ~lcalth Organization. The \Vorld Heath Report. LVorld Health :3 Warpovich hlli. Chen DD. Jimencz A. et nl: Aberrant crxpts as a
fated ( jr+lniz;ltion. Geneva. S~xitzerland. IYO7. hiomarker for colnn cancer: evaluation of potential chemopreventive
) are 1 I’;lrkc;l SL. Tong T, B~~IJen S. Q’inpo PA: Cancer Statistics 1997. A agents in the rat. Cancer Epidemiol Biomarkers Prsv 5: 355-360,
from CA Cancer J Clin -/? i-27. lYY7.
I thz 3 1 rock 13. Lanza E. Grcsnwald. Dietuw fiber, vegetahlss, and colon 2J Pretlw I?. O’Riordw &IA, Somich G.-k Amini SB, Pre:lo\s TG:
Ctlfll~Cr: critical rcvieu and meta-anal>Tes of the epidemiologic
that ciidmcc. J Nat1 Cancer lnrt S2: 651~~661, IYYU.
Aberrant crxpt$ corrclatc v.ith tumor incidence in FXI rats treated
with azoymeth.lne and ph>tate. Carcinogznesis IX 1509-1512, lY92.
:e of 4 I3urkilt DP: Epidcmic~lil~~x
Z. of cancer of the co1011and rectum. Cancer ‘5
-. Chnlla A. Rae DR. R&l\ BS: Interacti\s suppression of aberrant
that ZS:.Sl.~. 1071.
cnpt foci induced b> azosymeth:lnc in rat colon by phitic acid and
,lThl green te<I. Carcinogenecis IS. 21)2.~-2111fk lYY7.
this !h Ullah A. Sh;tm~utltl~n A\f: Dose-dependent inhibition of large
the inte\tin;ll c;tncer by inocitol heynphocphdte in F3J-I r<lts. Carcino-
eenssis II: 27lY-2222, IYYO.
r1 tea

The New England
Journal of Medicine
0Cop)right, 2000, by the htassachusctrs hlcdical Society

VOLUME 342 APRIL 20, 2000 NUMBER 16



ABSTRACT WEALTH of laboratory, nutritional, and
Bnckpwtnd We tested the hypothesis that die- epidemiologic evidence implicates dietary
tary intervention can inhibit the development of re- factors in the pathogenesis of colorectal
current colorcctal adenomas, which are precursors cancer.’ International variation in the inci-
of most large-bowel cancers. dence of and mortality due to large-bowel cancer,2
lIfe&ods We randomly assigned 2079 men and rapid increases in the incidence of colorectal cancer
women who ware 35 years of age or older and who in several countries,3 and data on migration* are con-
had had one or more histologically confirmed colo- sistent with a role of diet in the causation of colorcctal
rectal adenomas removed within six months before cancer. Moreover, altering the proportions of dietary
randomization to one of two groups: an intervention fats and fiber6 influences the development of colon
grorJp given intensive counseling and assigned to
follow a diet that was low in fat (20 percent of total
rumors in animals. In humans, diet affects the pro-
calories) and high in fiber (18 g of dietary fiber per duction of intTacolonic metabolic byproducts that may
1000 kcal) and fruits and vegetables (3.5 servings per influence carcinogenesis.‘-9 Observational epidemio-
1000 kcal!, and a control group given a standard bro- logic studies suggest that the ingestion of red meat
chure on healthy eating and assigned to follow their and dietary fat increases the risk of colorectal cancer,
usua! diet. Subjects entered the study after uildergo- whereas the ingestion of vegetables, dietary fiber, and
ing complete colonoscopy and removal of adenom- certain micronutrients lowers the risk.‘@14These re-
atous polyps; they remained in the study for approx- sults, holrpever, are inconsistent,lj and the evidence that
imately four years, undergoing colonoscopy one and diet contributes to causing colorectal cancer is hardly
four years after randomization. conclusive.
Res&r A total of 1905 of the randomized subjects
We studied whether adults can reduce their risk of
(91.6 percent) completed the study. Of the 958 subjects
in the intervention group and the 947 in the control
colorectal cancer by modivtng their diet. Because ad-
group who completed the study, 39.7 percent and enomatous polyps are considered precursors of most
39.5 percent, respectively, had at least one recurrent large-bo\vel cancers, n.e chose recurrence of adenomas
adenoma; the unadjusted risk ratio was 1.00 (95 per- as the primary end point.16
cent confidence interval, 0.90 to 1.12). Among subjects
with recurrent adenomas, the mean (%SE) number of
. such lesions was 1.8520.08 in the intervention group From the h’xional CancerJnstiture,Bethesda, Md. (AX, E.L., DC.);
and 1.84~0.07 in the control group. The rate of recur- the School of hlcdicinc and Biomedical Sciences,ScaccUnivcrsiv of New
rence of large adenomas (with a maximal diameter York at BufXo, Ruffalo (P.L.); EdwardHines, J:., Hospital, VcreransAf6irs
of at least 1 cm) and advanced adenomas (defined as Mcdicd Ccncer, Hines, Ill. (F.I.); the Kgiscr Foundation Rcscar:h hstit~t~,
lesions that had a maximal diameter of at least 1 cm Oakland, Calif. (B.C.); hlemorial Sloan-Kctrering Cancer Ccnr:r, New York
(M.S.); chc Univcrsit) of I’ictsburgh, Pitrsburgh (I.\\‘.); the University of Utah,
uibb h,x or at least 25 percent villous elements or evidence of Salt L&c Citv (R.B.); rVakc Forest University Baptist hlcdicd CCIXC~, %%I-
high-grade dysplasia, including carcinoma) did not ston-S&m, 2.C. (h1.R.C.); JValccrReed Amy hIcdical Gnicr, %shington,
icines. differ significantly between the two groups. D.C. (J.5V.K.); and Wcsrar, Rockvillc, bid. (J.C.).
0th~ authors \vcre Laurence Frccdmq Ph.D., National Cancer Instimtc,
-d aboxt Conclmions Adopting a diet that is low in fat and Bcthcsda, bid.; JamesMarshall, Ph.D., UN:ycrsir).ofArizona, Txson; Rob-
high in fiber, fruits, and vegetables does not influ- ert E. Schoen, M.D., MPH., Zlnivcrsity of Patsburgh, Pittsburgh; and
Learr: ence the risk of recurrence of colorectal adenomas. hl.mha Slanery, Ph.D., Univccsicy of Utah, Salr Lake City.
(N Engl J Med 2000;342:1149-55.) ‘Orhcr mcmbcrs of the Polyp Prcvcndon Trial Study Group an listed in
02000, Massachusetts Medical Society. the Appendix.

Volume 342 Number 16 * 1149
The New England Journal of hledicine
__~ -.. -

Some earlier trials tested the effects of dietary sup- its and History Questionnaire, *~a which was modified slightly to
plements, rather than an explicit dietary change, on rcflcct the intake of low-far and high-fiber foods. In addition, sub.
jccts in the intervention group complcrcd a four-day food record
the recurrence of adenomas.17-22Tko pioneering stud- six months after randomization. Each year we administered un.
ies did not find that low-fat diets (coupled with fiber scheduled 24.hour dietary-recall questionnaires to a newly sclc;-.
supplementation) reduced the recurrence of adeno- cd random sample of 10 percent of subjects.
mas,aaJJ but these small trials had limited statistical
pen-er. We report the results of the Polyp Prevention
Tria!, a large multicenter, randomized, controlled trial Subjects returned to their usual endoscopist for colonoscoF!
one and four years after randomization. The one-par colonoscop,
of the effect of a comprehensive dietary interven- had to be pcrformcd at least 180 days after randomization but lcsi
tion - counseling of patients and assignment to a than 2 years aftenvard. This colonoscopy served to detect and rc.
diet low in fat and high in fiber, fruits, and vegeta- move any lesions missed by the base-line colonoscopy. We obta&s
bles - on the recurrence of large-bowel adenomas. data on any unscheduled endoscopic procedure carried out in ad.
dition to the follow-up procedures at one and four years. ~VCa&c?
METHODS all investigators and subjects not to discuss a subject’s randomiza.
tion scams with the cndoscopists.
Study Design and Subjects
Assessment of Adenomas
Details of the study design, eligibility criteria, randomization pro-
cedures, dietary intervention, and end-point assessment have been Two central pathologists, who were unaware of the subjects’
previously reported. as.a6In brief, wc recruited subjects who were group assignment, determined the histologic features and degree of
at least 35 years old and who had had one or more histologically arypia (low-grade vs. high-grade) of all lesions. The cndoscopists’
confirmed colorcctal adcnomas removed during a qualifying co- reports provided information on the size, number, and location
lonoscopy (in which the cccum was visualized, all polyps were rc- of all po!yps.
moved, and the bowel was adequately prcparcd) svithin six months WC d&cd an adcnoma as recurrent if it was found during an;,
before randomization. Eligible subjects had no history ofcolorcctal cndoscopic proccdurc after the one-year colonoscopy or, for sub-
cancer, surgical resection of adcnomas, bowel resection, the poly- jects who missed the one-year colonoscopy, during any cndoscopi:
posis syndrome, or inflammat&y bowel disease; weighed no more procedure performed at least two years atier randomizxion. AL-
than 150 percent of the recommended level; were taking no lipid- cnomas found during the one-year colonoscopy were not consid-
lowcring drugs; and had no medical condition or dietary rcsnictions ered recurrent. An end-points committee of gastrocntcrologisrc
or practices that would substantially limit compliance with the pro- who were unaware of the subjects’ group assignment evaluated com-
tocol. The institutional review boards of the National Cancer In- plicated cases, including those involving lost tissue specimens c:
stitute and each participating center approved the study. All sub- failure to reach the cccum. The few colorectal cancers diagnosed
jects provided written informed consent. after the one-year colonoscopy were counted as recurrent lesion,.
Staff mcmbcrs at eight clinical centers (listed in the Appendix)
identified potential subjects through referrals by endoscopists or rc- Statistical Analysis
views of the records of the cndoscopy service. Of 35,277 poten- We used the intention-to-treat principle to compare the inter-
tial subjects, we enrolled 2079 (5.4 percent) in the trial. A total of vention and control groups, defining groups according to the ini-
1037 were randomly assigned to adopt a diet that was low in fat tial random assignment rather than according to actual or report.
and high in fiber, fruits, and vegetables (the intervention group), cd compliance with the protocol.a9 The primary end point was the
and 1042 were randomly assigned to follow their usual diet (the recurrence of adenomas during the interval from the one-year to
control group). The base-line characteristics of thcsc subjects the four-year colonoscopy. Secondary end points were the num-
have been reported prcviously.asta6 ber, size, location, and histologic fcarures of the adenomas that
were found. We calculated risk ratios and 95 percent confidence in.
Collection of Data tervals in order to compare end-point events in the two gr0ups.r:
At one of two clinic visits before randomization, we measured WC used logistic regression to adjust the effect of intervention for
each subject’s weight and height. At the base-line visit and at sub- base-line prognostic factors. We used logistic-regression models to
sequent annual visits at years 1,2, 3, and 4, each subject answered determine whether there was an interaction bcmeen dietary ic-
a qucstionnairc assessing a variety of demographic, clinical, and be- tcrvcntion and various covariatcs, and where appropriate, we pcr-
havioral characteristics and provided a venous blood specimen after formed covariate stratum-specific analyses.
an overnight fast.
Dietary Goals and Follow-up Characteristics of the Subjects
For subjects in the intervention group, the dictq goals were The base-line demographic, clinical, nutritional,
to provide 20 percent of total calories from fat, 18 g of dictq
fiber per 1000 kcal, and 3.5 servings of fruits and vegetables per
and behavioral characteristics \vere similar in the 95s
1000 kcal (range, 5 to 8 daily servings, depending on total energy subjects in the intervention group and the 947 sub-
intake). The intervention program included nutritional informa- jects in the control group who completed the stud!
tion and behavior-modification techniques. WC offered each sub- (Table 1). Of th ese 1905 subjects, 1768 (92.8 per-
ject more than 50 hours of counseling sessions during the four- cent) underwent a colonoscopy during year 1; the pro-
year intervention period, including 20 hours in the fist year. Each
subject in the intervention group was assigned to one nutritionist cedure was performed in 93.S percent of the subjects
for counseling and another for dietary assessment. We provided in the intervention group and 91.8 percent of the
subjects in the control group with general dietary guidelines from subjects in the control group (Table 2). The median
the liational Dairy Council but gave them no additional nutrition- observation period (3.05 years) and the mean num-
al or behavioral information.
WC followed the subjects for approximately four years after ran-
ber of colonoscopic examinations after randomization
domization. Each year all subjects completed a four-day food record (2.31) were the same in both groups (Table 2).
followed by a food-frequency questionnaire, the Block Health Hab- Subjects in the intervention group reduced their

1150 - April 20, 2000
--- --

ghtly to
:red un-
y selcct-
CHaRAcTERlsnc (N=9581 (N=947)
Ko. randomized 1037 1042
noscop) Age (rr) 61.0+0.3 61.1~0.3 So adcnoma at base lmc - no. (%) 3 (0.3) 1 10.1)
noscopy hlals sex (96) 4s.s 63.2 Lost to followup - no. (%) 76 ;7.3; 94 i9.0;
but less hlinority race or ethnic group (X) 11.7 9.2 Withdrew’ 34 (44.7) 48 (51.1)
: and re- hlorc than high school cducatio” (%) 65.3 65.2 Died before follow-up colonoscopy 42 (55.3) 46 (48.9)
Jbrained hlxricd (96) 78.2 80.8 Follow-up colonoscopy - no. (%)t 958 (92.4; 947 i90.9;
It in ad- Colo”oscopy at year I$ 899 (93.8) 869 (91.8)
Current smoker (%) 13.4 13.2
Vc asked Colo”oscopg at year 4 638 (71.0) 550 (63.3)
Alcohol intake (g/day) 7.420.4 8.0+0.5 Colonoscopy at year 4 and un- 150 (16.7) 169 (19.4)
Body-ma index 27.6irO.l 27.5~0.1 scheduled coloooscop~
Vigorous or moderate activir)’ or both (hr/wk) 12.64-0.5 11.6ZO.4 Unscheduled colonoscopy only 111 (12.3) lSO(17.3)
I-co colo”osiopy at year 1 59 (6.2) 78 (8.2)
Current aspirin use (?A) 23.3 22.0
Colo”oscopy only at year 4 19 (32.2) 30 (38.5)
ubjects’ Use of calcium supplements (%) 15.4 14.1 Colonoscopy at year 4 and un- 26 (44.1) 23 (29.5)
agree of USCof vitamin E supplements (56) 18.8 15.1 scheduled colonoscopy
copists’ Plasma total cholesterol (mg/dl)t 202.6-cl.S 200.2’1.7 Unschcdulcd colonoscopy only 14 (23.7) 25 (32.1)
ocation Strum tqtal carotcnoidr (/&dl)t 92.9~2.0 92.422.0 hlcdian follou-up - yr 3.05 3.05
Serum a-tocophcrol @g/dl)§ 1442~39 1335227 No. of proccdurcs - mean -CSE 2.3120.02 2.31+0.03
mg any Family history of colorcctal cancer (%) 24.3 26.0
br sub- ‘The reasons for withdraxxl were as follows: no colonoscopy at yen 4 in
Adennrna 21 cm in maximal diameter (%) 27.2 31.5 29 subjects in th: intcrvendon group and 43 subjects in the control group;
~2 Adcnomas (%) 33.0 33.8 rchual to participate in the caseof5 and 4 subjects, rcspcctively; and illness
,n. Ad-
a1 Vious or tubulovillous adcnomas (%)I 19.2 21.0 in I subject in the control group.
ologists Advanced adenoma (%)I[ 36.0 39.1 fAmong subjects in the intervention group who underwent follow-up
History of adenomaswithin previous 5 yr (%) 19.6 16.9 colonoscopy, in 35 the cecum w;1snot visualized; in 22 the bowel was poor-
:d corn-
ly pcepxcd, which might hzve cawed small polyps to bc ovcrlookcd, and in
‘lens or 53 one or more tissue specimenswere lost during the procedure and thcrc-
gnoscd ‘Plus-minus values are means tSE. Body-mass index is calculated as the
weight in kilograms divided by the square of the height in meters. fore were not analyzed, no slidcs were available for pathological rcvicw, or
lesions. data on histologic tindings were unknown. The rcspcctivc numbers in the
tTo convert values for cholesrcrol to millimolcs per liter, multiply b: control group ~verc40, 25, and 44 Fire subjects (three in the interwxion
0.025Yh. A total of414 subjects in the i”tcrvcn:ion group and412 subjects group and t\vo in the control group) underlvent sigmoidoscopy as the fol-
in ths control group were asscsscdafter an overnight fast. low-up proccdurc.
e intcr-
$To convert values for carotenoidj to millimolcs per liter, multiply by SP=O.lO for the diffcrcncc between groups.
the ini- 0.0185. A total of415 subjects in the intervention group and 411 subjects
EpOCt- in the control group were asscss:d after an overnight fast.
$4 total of 418 subjects in the interzntion group and 415 subjects in
year to the control group \vcre assessedafter an overnight fast.
I num-
ias that IInformation is based on the histologic analysis conducted by the central Subjects in the intervention group raised their fi-
nce in- ber intake by nearly 75 percent; subjects in the con-
oups.‘Q I]Advanced adenoma was dcfincd as one that had a m&ximal divnetcr of
at least 1 cm or at least 25 percent villaus clcmcnts or cvidcncc of high- trol group had a slight increase (Table 3). By the end
ion for
dels to
grade dysplasia (including carcinoma). of the study, the difference between the two groups
a-y in. in the change in fiber consumption was 6.9 g of di-
ve pcr- etary fiber per 1000 kcal (95 percent confidence in-
terval, 6.4 to 7.3). As compared with subjects in the
fat intake from a mean (+SE) of 35.650.2 percent control group, those in the intervention group who
of calories at the beginning of the trial to 23.820.2 consumed 2000 kcal per day increased their fiber in-
percent at four years, according to data obtained from take by nearly 14 g on average. Data from the four-
:0d, the food-frequency questionnaire (Table 3). The val- da). food records were similar to those from the food-
: 958 ues from four-day food records from a random sam- frequency questionnaires.
sub- ple of 20 percent of subjects lvere 32.2 percent at The number of senings of fruits and vegetables per
stud) base line and 20.6 percent at four years. Fat intake 1000 kcal increased by about t\vo thirds in the in-
: per- in the control group declined from 36.OkO.2 per- tervention group; subjects in the control group raised
: pro- cent of calories at base line to 33.9’0.2 percent at their fruit and vegetable intake only slightly (Table 3).
ejects four years. The values from four-day food records in The difference benveen the ht’o groups in the change
*f the this group were 32.5 percent and 31.1 percent, re- in fruit and vegetable intake was 1.13 servings per
dian spectively. The absolute difference between the inter- 1000 kcal (95 percent confidence interval, 1.04 to
wm- vention and control groups in the change in dietary 1.21). As comp.ared with subjects in the control group,
:ation fat as a proportion of total calories over the four-year subjects in the intervention group who consumed
j. period was 9.7 percent (95 percent confidence inter- 2000 kcal per day increased their fruit and vegetable
their val, 9.0 to 10.3 percent). intake by approximately 2.25 servings. Data from the

Volume 342 Number 16 * 1151
The New England Journal of Medicine


*.T P”4.YDOM~
(x=903) @=947) (s=SS3)
Fx (X of calories) 35.6-CO.2 23.850.2 36.020.2 33.910.2 -9.7 (-10.3 to -9.0)
Fiber (g/1000 kcal) lO.OIO.1 17.4eO.2 9.5zO.l 10.0+0.1 6.9 (6.4 to 7.3)
Fruits and vcgctables (servings/ 2.05?0.03 3.41+0.04 2.00?0.03 2.23-cO.03 1.13 (1.04 to 1.21)
1000 kcal)
Calories (kcal/day) 1972119 1570~16 1981~20 1910?18 -25 (-72 to 22)
Red and processed meat (g/day) 93.2~1.7 74.521.4 97.9 2 1.8 94.9c1.7 -15.8 (-20.2 to -11.5)
Ratio of red meat to chicken 2.6~0.1 1.a+o.1 2.6eO.l 2.9~0.1 -1.0 (-1.3 to 0.7)
and fish
Whole grains (g/day) 83.4~2.0 115.3Z2.3 76.821.9 72.621.9 35.9 (30.3 to 41.6)
Lcgu*cs (g/d-v) 14.2~0.6 48.5t1.6 13.7~0.6 16.2~0.7 31.9 (28.9 to 35.0)
Crucifcrou vcgctablcs (g/day) 28.9~0.9 44.4?1.5 26.5tl.O 27.72 1.0 14.2 (10.8 to 17.5)
Calcium from food and supple. 1032?20 1193223 1002z20 1096223 77.2 (16.1 to 138.3)
mcnts (mg/day)
Folatc from food and supplc- 435.0C8.8 593.9212.7 423.919.0 487.5~12.5 95.4 (62.0 to 128.9)
*ems (/x/W
Multivitamin use (%) 36.6 42.2 36.4 41.7 -0.3 (-4.8 to 4.2)
No. of subjects 95s 921 947 912
Plasma total cholcstcrol (mg/dl)$ 5.30~0.01 5.27+0.01 5.29?0.01 5.27?0.01 -0.01 (-0.03 to 0.01)
h’o. of subjects 414 372 412 364
Serum total carotcnoids (mg/dl)$ 4.461tO.02 4.5O-cO.02 4.45t0.02 4.42ZO.02 0.06 (0.01 to 0.11)
I‘;o. of subjects 415 369 411 361
Weight (lb)1 179.9?1.1 178.5~1.1 178.3+1.1 179.321.1 -2.5 (-3.6 to -1.4)
fro. of subjects 9% 919 947 907

*Plus-minuc values arc mcxn~ zSE. CI denotes confidence intcrvsl. To convert v~!ucs for cholesterol to millimolcs per liter,
multiply by 0.025S6; to convert v&cs for carotcnoids to millimoles per liter, multiply by 0.0185. Cholesterol and carotenoids were
measured a&r an overnight fast.
tDiffcrcnces wwc calculated only for subjects who had values at randomization and at year 4.
$Log-transformed values arc shown. Thr log-transformed values of -0.02 mg per deciliter in the intcmcntion group and -0.01
mg per deciliter in the conuol group for the difference xvithin groups from randomization to year 4 rcflcct respective decreasesin
absolute cholcstcrol concentrations of approximately 2 pcrsent and 1 percent; the absolute differcncc in the change bctwccn groups
is about -1 percent (95 percent confidence interval, -3 pcrccnt to 1 percent).
§Log-transformed values are shown. The log-transformed values of 0.04 mg per deciliter in the intcrvcntion group and -0.01
mg pe: deciliter in the control group for the diffcrencc within groups from randomizxion to year 4 reflect an incrcasc of approx-
imatcly 5 percent in absolute carotenoid conccnuations in the intcrvcntion group and A dccreascof 1 pcrccnt in the control group;
the absolute difference in the change bcnvccn groups is about 6 pcrccnt (95 pcrccnt confidcncc interval, 1 pcrccnt to 11 pcrccnt).
ITo convert values for weight to kilograms, divide by 2.2.

four-day food records showed a difference in the Over the four-year period of observation, the sub-
change between groups of 1.8 servings per 1000 kcal. jects in the intervention group had a significant in-
Changes in the intake of fat, fiber, and fruits and crease in serum carotenoid concentrations and de-
vegetables generally occurred jvithin the first year and crease in weight (Table 3), as compared with changes
were subsequently maintained. Data from the food- measured in subjects in the control group. The smali
frequency questionnaire showed that during the first reductions in plasma total cholesterol concentrations
year subjects in the intervention group obtained 24.6 did not differ significantly between the two groups.
percent of calories from fat, consumed 177 g of die- The differences in the changes in total cholesterol,
tary fiber per 1000 kcal, and ate 3.3 servings of fruits total carotenoids, and weight (calculated as the change
and vegetables per 1000 kcal. These changes were in the control group over time minus the change in
similar for men and n-omen. As compared with sub- the intervention group time) \vere somewhat
jects in the control group, subjects in the interven- greater after one year than after four years.
tion group also significantly altered their intake of oth-
Recurrence of Adenomas
er nutrients and foods, including red and processed
meat, u,hole grains, legumes, calcium, and folate (Ta- Adenomatous polyps recurred in 754 of the 1905
ble 3). Data from the 24-hour dietary recall were sim- subjects who completed the study (39.6 percent). At
ilar to those from the four-day food records. least one recurrent adenoma was found in 39.7 per-

1152 . April 20, 2000


(N=9581 (N=947l (95% CII’ VALUE

no. of subjects (%I

X:0. of adcnomas
s1t 350(39.7) 374(39.5) 1.00 (0.90-1.12) 0.98
7 219 (22.9) 217(22.9) 1.00 (O.Sj-l.lSj 1.00
85 (9.2) 82 (8.7) 1.06 (0.80-1.41) 0.75
a3 73 (7.6) 75 (7.9) 0.96 (0.71-1.31j 0.87
Location of adcnomasS
Proximal 203 (21.2) 173 (18.3) 1.16 (0.97-1.39) 0.12
Distal 100 (10.4) 124 (13.1) 0.90 (0.62-1.02) 0.09
l’rotimal and dmal 69 (7.2) 72 (7.6) 0 95 (0.69-1.30) 0.81
UthOWl 8 (0.8) 5 (0.5) 1.58 (0.52-4.82) 0.59
Largest adcnoma 21 cm 47 (4.9) 53 (5.6) 0.88 (0.60-1.25) 0.57
Advanced adcnomas 60 (6.3) 66 (7.Oj 0.90 (0.64-1.26) 0.60

l CI denotes confidcncc inrcrval
tThc absolute dificrcncc between groups was 0.2 percent (95 percent confidence intctval, -4.2
percent to 4.6 pcrccnt). The mean (zSE) number of rccwrent adcnomas among chosewith a recur-
rence was 1.85~0.08 in the intctvntion group and 1.841tO.07 in the conrrol group. The disuibu-
tions of adcnomasaccording to size wcrc not significamly diffcrcnt in the two groups (P=O.77).
Q’roximal is defined as the portion of the large bowl from the cccum up to, but nor including,
the splcnic flswrc. Distal is dsfincd as chc portion of the large bowl from the splcnic flcxurc up to
and including the recmm. The distributions of adenomar according to location were not significand)
differcnr in the Two groups (P=O 17).
§.%nadvanced adcnoma was one that had a maxima! diameter ofat least 1 cm or at least 25 percent
villous clcmcnrs or c\idencc of high-grade dysplasia (including carcinoma).

cent of subjects in the intervention group and 39.5 Colorectal cancer was diagnosed in 14 subjects af-
percent of subjects in the control group; the unad- ter randomization (10 in the intervention group and
justed risk ratio \vas 1.00 (95 percent confidence in- 4 in the control group); the unadjusted risk ratio
terval, 0.90 to 1.12; P=O.98) (Table 4). Of these re- was 2.5 (95 percent confidence interval, 0.8 to 7.9;
current adenomas, the mean number was 1.85~0.08 P=O.19). Of these 14 subjects, 6 (4 in the intervention
in the intervention group and 1.84t0.07 in the con- group and 2 in the control group) were given a diag-
trol group (P=O.93). Among the 638 subjects in the nosis after the one-year colonoscopy; the unadjusted
intervention group and 550 subjects in the control risk ratio was 2.0 (95 percent confidence interval,
group who underwent colonoscopy only at year 1 and 0.4 to 10.8; P=O.69).
year 4 afier randomization, 36.7 percent and 35.8 To adjust for an imbalance in influential base-line
percent, respectively, had one or more recurrent ade- variables between the groups, we used logistic-regres-
nomas; the unadjusted risk ratio was 1.02 (95 per- sion models that included as covariates the random
cent confidence interval, 0.8s to 1.19; P=O.81). group assignment and the base-line characteristics list-
he sub- The intervention and control groups did not dif- ed in Table 3. Adjustment for these factors had no
2nt in- fer significantly with respect to the number with re- effect on the risk of recurrence.
nd de- current large adenomas (lvith a maximal diameter For all but one of the covariates listed in Table 1,
-hanges of at least 1 cm) or advanced adenomas (defined as n-e found on logistic-regression analysis that there was
.e small those that had a msyimal diameter of at least 1 cm no statistically significant (P<O.Ol) interaction with
rations or at least 25 percent villous elements or evidence of group assignment. We obsemcd a significant interac-
qoups. high-grade dysplasia, including carcinoma); this was tion (P=O.O05 before adjustment for multiple com-
esterol, true when the analysis included all those nho com- parisons) between the randomization group and sex.
change pleted the study (Table 4) as well as bvhcn it included We therefore examined the recurrence of adenomas
mge in those who underivent only the scheduled colonos- among men and women separately. Among men, the
newhat copies at year 1 and year 4 after randomization (data recurrence rate n-as lolver in the intervention group
not shown). In both groups, approximately 27 per- than in the control group (41.9 percent vs. 46.7 per-
cent of subjects had at least one recurrent adenoma cent); the unadjusted risk ratio was 0.89 (95 percent
proximal to the splenic flexure (Table 4). Sixty-three confidence interval, 0.79 to 1.02; P=O.ll). Among
e 1905 percent of recurrent adenomas were proximal to the women, the rate of recurrence was higher in the in-
:nt). At splenic flexure, whereas 58 percent of base-line ade- tervention group than in the control group (35.4 per-
j.7 per- nomas were distal to that site (data not shown). cent vs. 27.2 percent); the unadjusted risk ratio was

Volume 342 Number 16 * 1153

The Nex\- England Journal of hfedicine

1.30 (95 percent confidence interval, 1.04 to 1.63; based on the equation of Keys et al.32) These data,
P=O.O3). With respect to both large and advanced hon.ever, do not preclude the possibility that in the
recurrent lesions, the differences beoveen groups were light of the dietary expectations fostered by the trial,
not significant for either men or \t.omen; the inter- subjects in the intervention group systematically uh.
action benveen the randomization group and sex was derreported their intake of fat or overreported their
not significant for either end point. There ivere also no consumption of fiber or fruits and vegetables. An.
significant (PcO.05) differences betlveen the groups other possibility is that the dietary intervention \v;ij
in the number of either deaths or hospitalizations inadequate; a reduction in fat intake to no more thAtn
(for all causes and for specific diagnoses). 15 percent of calories or a greater intake of fiber or
fruits and vegetables might be required to reduce the
DISCUSSION risk of recurrent adenomas. Moreover, \ve may not
We found that the rate of recurrent adenomas was have chosen the optimal set of dietary targets. The 20
not changed by dietary inter\*ention. Our results are percent reduction in the consumption of red and
compatible with, at most, an absolute reduction re- processed meat among subjects in the inten-ention
lated to the intervention of about 4 percent in the group may have been too small to affect the risk 0:
incidence of recurrent adenomas (Table 4). We also recurrence of adenomas. The same may be true for
found no efiect of the dietary intervention on the in- reductions in the consumption of meat cooked at
cidence of large or advanced recurrent lesions. high temperatures (rvhich contains high concentra-
T%o previous trials also found that dietq changes tions of heterocyclic amines) or sugar.11
had no effect on the overall risk of recurrence of co- The mean age of the subjects at base line was 61
lorectal adenomas. The Toronto Polyp Prevention Tri- years. If nutritional factors influence critical events
al reported no significant difference in recurrence af- in colorectal neoplasia at the molecular, cellular, or
ter two years between subjects in the intervention tissue level only e;lrlier in life, then a change in diet
group and those in the control group (a total of 201 later in adult life may be ineffective. A relatively short
subjects) \vho reported ingesting 25 and 33 percent period of dietary intervention (four years) might also
of calories from fat and 35 and 16 g of fiber per day, fXl to reduce the risk of recurrent adenomas. A longer
respectively. 23 In the Australian Polyp Prevention period of intervention as well as foIlon,-up might al-
Project, Lvhich included 424 subjects, none of the ir- low the development of enough adenomas to reveal
terventions (a reduction in clietav fat, use of a n-heat- the protective effect of the intervention, if there were
bran-fiber supplement, and supplementation with one. In a recent clinical trial of calcium supplementa-
beta carotene) resulted in a statistically significant re- tion to prevent colorectal adenoma,l* however, the av-
duction in the risk of recurrence afier 4s months of erage age of the subjects, the duration of the inteRen-
observation.zJ The Australian trial did report a mar- tion, and the length of follow-up n-ere similar to those
ginally significant reduction in the recurrence of large in our study, but that study did find a loner recur-
adenomas (a1 cm in diameter) among subjects eat- rence rate among subjects in the intervention group.
ing a lo\v-fat diet, but in that study large recurrent Bias is an unlikely esplanation for our results. Sub-
adcnomas developed in only 17 subjects, as com- jects in the inten-ention and control groups \vho com-
pared with 100 in our study. pleted the study did not differ appreciably with re-
The straightfonvard interpretation of our finding is spect to base-line characteristics, and the main results
that a diet that is lo~v in fat, and high in fiber, fruits, did not change after adjustment for multiple covari-
and vegetables does not reduce the risk of recurrent ates in logistic-regression analysis. Although we could
adenomas or, by inference, colorectal cancer. Alter- not disguise the group assignments from the sub-
native explanations, however, merit consideration. jects or guarantee that the endoscopists Lvere unaware
Most recurrent adenomas nere small; only about of these assignments, \ve ha1.e no reason to suspect
5 percent of subjects had a recurrent lesion 1 cm or that endoscopists tended to search more diligenth
more in diameter (Table 4). Adopting a diet that was for - and therefore find more - adenomas among
lo\v in fLlt and high in fiber, fruits, and vegetables subjects in the intervention group than in the contra!
might affect only the gronth of small adenomas into group. A series of imputations based on the age and
large adenomas or the transformation of large ade- ses of subjects who did not undergo follotv-up CO-
nomas into im.asive carcinomas.31 lonoscopy made no appreciable difference in estimates
The dietary-assessment data indicated that the in- of recurrmce.3G
tervention and control groups differed substantiall! The higher rate of recurrent adenomas among
in the consumption of fat, fiber, and fruits and veg- nromen in the intewention group than among those
etables. The findings regarding carotenoid concentra- in the control group and the interaction between sex
tions and wyeight \i’ere consistent with such differenc- and group II’as not affected by a multil-ariate adjust-
es. (The changes in blood lipid concentrations were ment for age and the number of adenomas at base line
minimal but compatible with the results of other stud- (both of hvhich xvere predictive of the risk of recur-
ies of dietary intervention as well as nith predictions rence) and other covariates listed in Table 1. Never-

1154 - April 20, 2000

‘hex data, theless, we conclude that this interaction resulted from scum trig!ycerides and/or plasma glucose associatedwith risk? Cancer Ep-
idcmiol Biomarkcrs l’rcv 1994;3:687-95.
Ihat in the chance observations arising from repeated testing. 10. Giovannucci E, Rimm EB, Stampfcr hlJ, Colditz GA, Archcrio A,
y the trial, In the Toronto trial, the rate of recurrence \vas loiver \Villctt I\%. Intake of fat, meat, and fiber in relation to risk of colon cancer
in men. Cancer Rcs 1994;54.2390-7.
itically un- among women in the intervention group but higher 11. World Cancer Rcscarch Fund. Food, nutrition and the prevention of
xted their among men in that group, as compared lvith the risk cancer: a global perspective. LVashiigton, D.C.: American Institute for
ables. An- in the control subjects, though these differences were Cancer Rcscarch, 1997.
12. Potter JD. Colorectal cancer: molecu!es and populations J Nat1 Can-
:ntion was not statistically significant.*j cer Inst 1999;91:916-32.
more than In summary, our study provided no evidence that 13. Martinez XlE, JVillctt 1VC. Calcium, vitamin D, and colorcctal cancer:
Af fiber or a diet lovv in fat and high in fiber, fruits, and vegeta- a rsricw ofthe cpidcmiologic cvidencc. Cancer Epidcmiol Biomarkcrs Prev
reduce the bles reduces the risk of recurrent colorectal adeno- 14. Giovannucci E, Stampfcr h!J, Colditz GA, et a!. hlultivitamin use,
I may not mas. Nevertheless, \ve cannot definitively conclude folatc, and colon cancer in women in the h’urscs’ Health Study. Ann Intern
that a change in diet is ineffective in reducing the hlcd 1996;129:517-24.
ts. The 20 15. FuchsCS, Gio~xnnnucciEL, Colditz GA, et al Dictarr fiber and the risk
f red and risk of colorectal cancer. Nor should \ve overlook the ofcolorecral cancerand adcnomain uomcn. X Engl J Mcd 1999;340:169-76.
.ervention abundant data indicating that a diet low in saturated 16. Scharzkin A, Freedman IS, Dawscy SM, Lanza E. Interpreting pre-
cursor studies. xvhat polyp trials tell us about large-bowel cancer J Sat1
he risk of fats and rich in fruits, vegetables, and \l.hole grains has Cancer Inst 1994;86:1053-7.
e true for a favorable influence on the risk of chronic disease 17. BusscyHJ, DcCossc JJ, Dcschner EE, et al. A randomized trial of
:ooked at and mortality.js-37 ascorbic acid in polyporis coli. Cancer 1982;50:1434-9.
18. DeCossc JJ, hliller HH, Lcsscr ML. EtTect ofwheat fiber and vitamins
oncentra- C and E on rectal polyps in patients with familial adenomatous polyp&s.
APPENDIX J Iiatl Cancer Inst 1989;81:1290-7.
19. McKcown-Eysscn G, Holloway C, Jazmaji V, Bright-See E, Dion I’,
K was 61 Other mcmbcrs of the Polyp Prc\ention Trial Srcdy Croup were as fol- Bruce WR. A randomized trial of vitamins C and E in the prevention of
-al events lows: hirior?nl Cnnrcr InflirtirlLrc- R. Ballard-Barbash,C. Clifford, J.Tangrca; recurrence of colorectal polyps. Cancer Rcs 1988;48:4701-5.
.S:,rrr (Inivcrsi;y of?++v YwE (If Bufi!o - D. Hayes, X.J Pctrclli, hi. Bcd- 20. Ror.cuxi L, Di Dor.ato I’, Carati L, et a!. Antioxidant vitamins or
~llular, or dome, I;. Kroldart, S. Rauth, L. Wodarski; Ednard Hints,)., Hsspi:a!, Vrr- IJCN~OSC for the prevention of the recurrence of colorcstal adenomas. Dis
;e in diet rmtukfiirrMcdicnl Ccntrr- I? hlurphy, E.C. BotC, L. Brandt-Whitting- Colon Rectum 1993,36:227-34.
.rely short ton, N. Hxoon, N. Kazi, M.A. hioorc, S.B. Orluff, PV.J.Ottosen, hi. 21. Greenberg ER, Baron JA, Tosteson TD, et al. A clinical trial of antioxi-
I’accl, R.L. Rothschild, hl. Ryan, J.hi. Sullivan, A. Vcrma; K&r Forrndn- dant vitamins to prevent colorectal adenoma.N Engl J hIcd 1934;331:141-7.
-light also rititt fixnrclt Inrrirm - J.V. Sclby, G. Friedman, ht Lawson, G. Taff, D. 22. Baron J.4>Beach hf. hfandcl JS, et al. Calcium supplements for the
A longer Snow, hi. B&a), hl. Schoenbcrgcr, I;. Sampel, T. Giboney, hi. Rand& prcvcntion of colorectal adcnomas.N Engl J &Icd 1999;340:101-7.
6fcrr,orin!Sloan-~~nrri,1~ Cavrtr Comr - S. Winawer, A. Bloch, J. hlaycr,
might al- R. AIorsc, L. Latkany, D. D’Amato, A Schaffcr, L. Cohen; Uniwrrig d
23. hlcK;eown-Eyssen GE, Bright-See E, Bruce WR, ct al. A randomized
trial of a low fat high fibre diet in the recurrence of colorectal polyps.
to reveal PirtsburJb - R.R. Schadc, L. Kuller, B. Gahagan, A. Caggiula, T. Coync, J Clin Epidemiol 1994;47:525-36. [Erratum, J Clin Epidcmiol 1995;48:i.]
ierc lverc C. l.ucas, S. Pappxr, G. I.andis, I. Dyjak, R Robinson, I.. Seaxh, D. 24. hlacl .enn.w R, hlacrx F, Bain C, et al. Rantiomired trial of intake of
Hanson; C’riiwrriry of Utah - K. Visiofsky, J. Benson, J, X&on, R. fat, fiber, and beta carotene to prevent colorcctal adenomas: the Australian
Jementa- O’Donncl, hl. Briley, 1~.McDivitt, K. Heinrich; \V. Samovitz; WukcForrrr Polyp Prevention I’rojcct. J SatI Cancer Inst 1995;87:1760-6.
r, the av- Giverriry Bnptin M~iicai Crnrer- E. Paskett, S. Quandt, C. DcGraffin- 25. Schatzkin A, Lanza E, Freedman L.S,et al. The Polyp Prevention Trial.
interven- rcid, K. Bradham, L. Kent, hi. Self, D. Boylcs, D. 1Vcst,L. hlartin, N. Tay I. Rational<, design, recruitment, and baseline participant characteristics.
lor, E. Dickcnson, I? Kuhn, J. Harmon, I. Richardson, H. Lee, E. Marceau; Cancer Epidemiol Biomarkcrs Prev 1996;5:375-83.
to those Wdrcr Rrrd Amy Aicdicnl Ccnrer -- D.J. Mat&i, R.K.H. Wang, C. 26. Lanza E, Schatzkin A, Ballard-Barbash R, et al. The Polyp Prevention
er rccur- Chcney, E. Rucda-Pedraza, V. Jones-hliskovsky, A. Greaser, E. Stoute, S. Tria!. II. Dietary intervention and baseline participant d~etaq characteris-
Hancock, S. Chandler, hi. Burman, E. Crutchfield, C. Slivka, L. Johnson;
n group. lkrtz and hTurririon Cmrdicnrin,? Ccnrer (TVc’cs:nr) - hi. Hasson, C. Da.-
tics. Cancer Epidemiol Biomarkcrs Prcv 1996;5:385-92. [Erratum, CUKCC
Epidemiol Biomarkers Prer 1996;S:SS4.]
IIts. Sub- ton, B. Brewer, C. Sharbaugh, B. O’Brien, N. Odaka, K;. Umbcl, J. PinsAT, 27. Block G, Hutman ASI, Dresser Chl, Carroll XID, Gannon J, Gardner
ho com- H. I’ricc, I? Clark; Cctlrra! Parhohyirrr- K. Lcwin (University of Califor- L. A data-basedapproach to diet questionnaire design and testing. fun J Ep-
nia, Los Angclcs), H. Appelman (University of hiichigan); Laboraroricr- idemiol 1986;124:453-69.
with re- P.S.Bachorik, K. Lovejoy (Johns Hopkins Univcrsi~), A. So\rell (Ccntcrs 28. hIarcs-Perlman JA, Klein BE, Klein R, Rittcr LL, Fisher MR,
n results for Discasc Control and I’rcrsntion); Dnra and .S@cryMmirwi~~g Cmnrrrirrcc Frcudcnhcim JL. A diet history questionnaire ranks nutrient intakes in
: covari- -- E.R. Greenberg (Norris Cotton Cancer Center and Dartmouth Medical middle-aged and older men and women similarly to multiple food records.
School), E. Feldman (Augusta, Ga.), C. Gana (Cornell University), R. Sum- J Nutr 1993;123:489-501.
.ve could mers (University of Iowa); 5. We&d (University of Minnesota), D. De&lets 29. Friedman Lhl, Furberg CD, DcXicts DL. Fundamentals of clinical tri-
the sub- (University of !Yisconsin). als. Littleton, htass : PSG Publishing, 1951.
unaware 30. KJcinbaum DG, Kuppcr LL, hl orgenstern H. Epidemiologic research:
suspect REFEREXCES principlcc and quantitative methods. Belmont, C&f.: Lifetime Learning,
iligentiy, 1. Putter JD, Slatxry ML, Bostick k\l, Gapscur S,\I. Colon cancer: a re- 31. Hill MJ, hlorson BC, BusseyHJ. Actiology ofadenoma-carcinoma sc-
i among vie\\ of the epidemiology. Fpidcmiol Ret 1993;15:499-545. qucnce in large bovcl. Lancet 1978;1:245-7.
2. Tominaga 5, Aoki I;, Fujimoto I, Kurihara hl, cds. Cancer mortaliF 32. Keys A, Anderson JT, Grands F Serum cholesterol response to changes
control and morbidiv statistics: Japan and the world, 1994. Boca Raton, Fla.: CRC in the dxt. I. Iodine value of dietary fa: versus 2S-P. htetabolism 1965;
age and rrcrs, 1994. 14~747-58.
--up co- 3. Ji B-T, Devcsa SS, Chow iv-H, Jin F, Gao Y-T. Colorectal cancer ix- 33. !VakabayashiK;, Sagao Al, Esumi H, Sugimura T. Food-derived mu-
dence trends by subsite in urban Shanghai, 1972.1994. Cancer Epidemiol ta,zensand carcinogens. Cancer Rer 1992;52:Suppl:2092s~209ss.
stimates Biomarkcrs Prcv 1998;7:661-6 34. Little RJA, Rubin DE. Statistical ana!ysissmithmissing data. SCW~OCork:
4. hichlichael AJ, Gilts GG. Cancer in migrants to Austraha: extending the John ~Vllcy, 19Si.
descriptive epidemiological data. Cancer Res 1986;45:751-6. 35. Appcl LJ, hioorc TJ, Obarzanck E, et al. A clinical trial of the efkt~
among 5. Zhao Ll’, Kushi LH, Klein RD, Prentice RL. Quantitative relicw ofstud- of dietary patterns on blood pressure. IX Engl J hlsd 1997;336 1117-24.
g those ies of dietary fa: and rat colon carcinoma. Nutr Cancer 1991;15:169-77. 36. Liu S, Stampfer AIJ, IIu FB, et a!. 11X&-grain consumption and risk
‘een sex 6. KritchcvskT D Protective role ofwheat bran fiber: preclinieal data. Am ofcoronary heart disease:results from the Furscs’ Health Study. Am 7 Clin
J hlcd 1999;106:Suppl 1.4:285-31s. Kutc 1999;70:412-9.
adjust- 7. Xagengast FM, Grubbcn blJ,a, van &1unster II? Role of bile acids in 37. Dietary Guidclincs Advisory Committee. Report of the Dietary Guide-
)ase line colorcctal carcinogenesis. Eur J Cancer 1?95;31.4.1067-70. lines Advisors Commitrce on the dietary widelines for Americans, 1995,
f recur- 8. Lupton JR, Turner XD. Potential protective mechanismsofwhcat bran to the Secrctrj of IIcal:h and Human .&>.ices and the Sscrctav of Agri-
fiber. Am J E.ied 1999.106:Suppl 1.4:24S-27s. culture. LVahington, DC.: Department ofAgriculturc, Agriculturd Research
Never- 9. hlckeon-n-Eysscn G. Epidemiology of colorectal cancer rcvicitcd: arc Service, 3995.

Volume 342 Number 16 * 7155


by Charles 8. Simone,M.D.,Nicole1. Simone,andCharles B. Simone,II

astrointe<rinaI CatlC2lj 2rt: lb.2 SXOlld
leajing caux ol deathamongall cancer
becauset1-xpeople generallycar lessfiber ar
ani.malfat.Thehighestcolowtal cancerlatesal
,patiencs.the dcarh-rate for cancerof the-colon2nd in WesrcrnEurope 2nd English-speakingco rries.
rectum has remined tirtually the samesir.ce1933, The lo\v:st rates are found in Africa and A ) but
which means thcrc has been ~suI~I+,* no prgglrS5 tha[ is chx,$ng rapidly for Asianswho have: ipted
in the treatment of thesecancers.‘Thenc’;vcolor~ta! a Wstcrnizcil diet.
‘ cancercasesfar 200’3 wcrc appxGrrxXcl~~130.200. In cnuntrics \+4tll2 high inciclcnceof CC xral
?hr estimted number of de&E fromcolor~taal cancel cancer,mostof the cancer5an locatedin the le :olon
is 56,600. A p;r;on ha; a one-in-18chanceof dsvel- andwrum, wherszjin mumties with a low in’ WLCC,
oping color~cral cxxer’ o’;cr his or her lil’e;im< 1nos1 ol hc cancers;I172ii\ the right colon. Cart )gCnS
Major Mcrences in death r~!csfrom colossal I)CCVRIC ptqrcssively moreconccntratcdat th nd of
canceroccur in different pars of the world, andepi- the gastrointestinalttaci (left colon and RC~L t. Dr.
demiologicalstudies show that dienry factorsaccount L)enisBurkitl sharesthe lollowing analogy this
for the different incidence rates.’The mart indusrri- distribution of cancer:“%ile a man pnxeeC IOWl
alizeda country, the higher the rateof colorectalc?r~rs: a pathcaq-ing a leakypot of warerconraining abler
of dye that is graduallydissolving,the waterbecomesmore is unlikely to do harm and may have the potenti; for
deeply colored becausethe volume will be progressively reducing cancerratesin North America.”
reducedand the dyemoreconcentrated,andmore dyetill Becausethe evidencefrom epidemiologicaland IO-
be progressivelydissolved.” ratory studies was sufficiently consistentthat high- cr,
low-fat diets could lower cancerrisk, other U.S.age] es,
ROLEOF FIBER organizations, and other governments issued In im
Fiber is a complex carbohydrateconsistingof a poly- dietary guidelines in the mid 1980s.Theseinclude .he
sacchatideand a lignin substancechatprovidesthe strut- United Sta~csNational CancerInstitute,NationalInsr tcs
lure of a plant cell. lt is unrlig,.stedresiduethat teachesthe of Health, United States Department of Agricu re,
end of the small intestine.The threegroupsof dicraryfiber Department of Health and Human Services,Ame :an
typesarevegetablefibers,which a:ehighly fermentable and Cancer Society, Australia, Canada, the Joint Furi: ‘an
have a low undigestedcontent; bran, which i; less fer- Organization for Cooperation in Cancer Prewr )n,
mentable;and purified hbers,such a cellulose,which are Norway, Sweden. and Japan. They all independ .1ly
muchlessfetmentab!candhavea high undigesredcontent. agreedthat to reduce cznctr risk, peopleshould inc 35-e
Dietary fiber act5 as a “glue” for certainchemicals.For their consumption of green. yellcw, and cmcifecrocs:g
il\s\nncc. unconj\~r,qrrd hilt acid<. which the hod,: pro- crablts. cir~s fruits, and whole-grain cereal products nd
duces.canbe absorbedto hberin the colon and passedOUL reduce Iheir intake of fatsto about 3U percent.
in the stool without intestinalbacteriaformingcarcinogens In 1984, the United StaresNational CancerIns .Ke
from those bile rcids. In addition, some fiber binds to recommendedan intake of 25-35 gramsof fiber da t0
cholesterol, lipids, nitrogen, and certain minerals, and decreasethe risk of cancer.However,the American lb-
eliminatesthem in the stool. This action lowersthe blood lit consumesonly about 6-15 grams of liber per da)
concentrationof cholesteroland certain otherlipids. During the last 25 years, thousandsof in rim nd
Drs. Higginson and Oettle were the first to repon in anjmgl srudies have been published demonstrating Tat
1960that dietary fiber consumptionwasassociated with a fiber can decreasethe risk of colorectal cancer.: ?SC
low risk of developingcolon cancec’They notedthat the papers havenot beenlncludcd in this rMew. Since1 30.
Bantutribal peoplein SouthAfricahad a low incidenceof hundreds of published papersde?onsrratethat high 3cr
colorectal,cancer. They ali -:eicrcted largepilesof fecesthat intake can reduce the incidence of colunctal cnnc in
wererelaiedto the largeamountof dietaryfber theyace.Dr. humans.Somehave beenincluded in this review.“10
DenisBurkitt continuedtheresearchandconcludedthat the
high-fiberdiet resultedin a rapid transittime for solidmaih-
rial to pass through the gzstminr&~l tract and also Reports from tie United Srates Nitional Cancer Institute con-
increasedthe amountof stool.These;wo variablesareasso- sistent thatdietaryfiber can dscreasc the risk of colorectal ricer::
ciatedwith a decreasedincidenceof colorectalcancer.+ * ‘This valuation clearly su~erts a relationship b
Diet in rural Ahica and in other similar locationspro- cancer and a diet low in fiber:’ From 40 epidemiologi
vides about 25 grams of crude fiber daily. tvhereas reports (Greenwald FI et a~!Diecar/ fiber in tie red
\\Testerndiets provide only 8 to 15 gramsof fiber daily cancer risk/Am Die:Asrcc 87[9];1 178-I 188. I98T).
LVith a more rapid transit time, bilr acidsand other car- l “The analysis of chest studies gives support for a
cinogensproducedby anazrobicbacteriamoveout of the etTecc (against colorccul cancer] associated WI& fiber
gas:rointestinaltract more quickly! Furthermore,sincethe Fmm 23 case-comrol studies, 7 international comlati
volume of fecesis increased,carcinogensthat are pro- within-country correlation studies, 2 cohort studies, a
ducedpassthrough the gut more diluted. Hence,if mo:e trend stud;es Frock. Lanu, Greenwald: Diecary Eber, veg
dietaryftber is eaten,carcinogenspassout of thegut more colon cancer: critic4 review and men-analysis of epid
quickly and there are fewercarcinogensper square inch. studies. /NJ 82:650-66 I, 1990).
l “Based on current knowled@, recommended nuu
lines for reducing the risk of colon cancer
include decrea
In 1982, the &‘:arionalAcademy of Sciencesfound sbmpiion. adequate amounts of fruits. vqeables, and
rhar,accordingto sn-icrepidemiologicalcriteria,therewas zgoidance of ovcr++eight” (Shlke.WinaweC Gretnwald.
“no conclusiveebqdcnce to indicate that dietaryfiberexerts prevention of colorectal cancer. MWHO 68377.385.1
a prorectiw efIect again.iLcolorectal cancerin humans.” e “Both prospectiw and retrospective studies
Nevertheless,the U.S. National Academyol Sciencesdid etable and fruit intake m3y reduce the risk of cancers
cancer of the colon and recrum” (Ziegler RG:
issuedietary guidelinesbecausethe datawere“highly sug-
carorenoids and the risk of cancer. Environ
gesrivethat rcduccd fnt consunlpiion and increasedcon-
Branch, National Cancer Inrtirute, Bethesda.
sumption of cereals, fruit, and vegetablesrepresentthe
current state of knowledge and form the basis of a diet ihat
: :
,; ‘_
.. ,:,
.; :
“. Various organizations and governments around the world l 1997 American Dietetic Association sition:“Rcsults of all
;‘. have issued c~n~en~u~ statements that high fiber coosumprion scud& prwide substantive evidenca
./ can reduce the risk of colorecca! cancer foods is inwrse!y relaud ~0 risk, of
:: .
‘, ! l I999 Wcrld Health Organintion: “The consumption of foods cerr. It is estimated chat the risk of co
:.: j
: rich in polysaccharides (e.g.. diecsry 6ber or non.starch polysac- population could be reduced by about enr if fiber intake
charides) is associated with a decreased risk of colorectal adeno- from food sourtcs,were increased b
ma and colorectat cancer” (Gr] h Pm 3:57-62, 1999). grams per day”UAmDierAssoc 97[IO]
.. Recommenda$on:Vegetables and whole-grain cereals should Recommendation: Promore food I
I. with the Food Guide Pyramid. This reco
: .:; :j
be consumed In high amounts and should be a major’componenr
;.., of the diet of plant foods to achieve adequate fiber in
.. I : I999 Colon Cancer Prevention Pi-ogram Project: “I 3.5 grams and adulu. Include at least 2 to 3 servings
” .‘. of wheat brin per day de:reases the recurrence rats of adeno- of the daily 6 to I I servings of grains,2 to
.-ii matows colon polyps” (Am) Med 106[IA]:43S45S,f999). 3 to 5 servings of vegetables daily, and le
,.‘;j j l 1999 The Seven Countries Study Conclusion (Croatia, r&e a week.
: jI Finland, Greece, Italy, Japan, Netherlands, Serbia, U.S.): “High fiber 1995 Australia:“Reducrion in the

intake was strongly associated with low colorectal momlity, An mas was observed when a low&t diet
increase of IO grams in the daily inuke of fiber was associarad fiber wheat bran supplementation of
with a 33% lower risk of 25.year colorcctal cancer mortalir/” (IN 87:1760-1766,199s).
J Caocef 84:174:179,1999). Recommendation: 25 grams of fibe
Recommendation: locrexse the daily intake of fiber b IO grams. b 1994 United Nations Food and
1998 European Cancer Prwention Consensus Panel: “A diet
l “High fiber intake consisting of vegeta
rich in high-fiber cereal is associated with a reduced risk of col- cecrive against colorectal cancer” (Eurj C
oreccal cancer” (EurJ Cone Prcv 7[suppl2]:51-53. 1996). 517, 1998).

* Decreasedrisk is most con
DAILY CONSUhfPTfOX OF FIBER from vegetables.followedby fibers
According to the Nationa!tk2derny of Sciencs,over sacchar-ides, starches,and fiber foe
‘40% of North American5 are likely to dmlop cancer and at Supplementalwheat bran

lea,sthalf of them will die from it. The cancerincidence gramsper day can decreasethe r
worldwi& is increasing.Themajorityof healthbudgetswill matous colon polyps.
bc spent on treating cancerin mostdcvdopingcountries. 0 SinceNorth Americanstypi
Canceris largely preventable.Fewert’nzn5%of cancer of only 8 to 15 grams of ftbcr per da
casesare linked to generics. plement may be warranted.
Overwhelming evidencesupportsthe statementthat * Currently, 40% of North
“the consumption of fiber mayreducethe risk of colonctal cancerand the incidenceis rising. I evenrivemeasures
cancer” in fact. basedon the volume,credrbility,and reli- arz not instituted, the cost to the Statesand its
ability of the scientific facts, we arc convinced that fiber people will bc enormousin rcrn
cm, not may, but cun reduce rhc risk ofcnlorectal cancer livily, and lives 10%.

Depending on the study, North Americanstyprcally A Cacw Jfir Chi&n, 49(l):&3 1,1999
consumeonly about 8-15 gramsof fiber eachday.Most of
the consensusreports recommend25 to 35 gramsof fiber 139.:.

eachday to prtxect againstcolorectalcancer.UnlessNorth AGa:
Americanshave the time or inclinationto bccomca grating
anirrwl, it would be diKcult to attainthe protectivelevel of
fiber each day without taking a suppiemenr.

lDietary fiber is safe.
9 Hundreds of studies.involving tens of thousandsof
subjects,demonstratethat 25 to 35 gramsof dietary fiber
daily can reduce the risk of colorcctalcancer.
9. Amcriwn GncuSaicry: h’vrri.& a~~dCorrrrr; O~w~l+wnti~n. I+
York: American Canw Socicty, 19S4.
IO. k+nti~l\ Diccctic A%c&&M &&ion Papci: Hr.&$ impI?adons ddi&vy Third Annual Symposium on Dice and fiu
fibcr.]AmI)ircllrroc97(10):1157-1159, October 1997. cmcw, 1986.
11. Ar’cn;in G,Axclson 0, Ericuon-Be&&i AU, Frtdtikuon M, Nihcon E, 36. EWE DF: PhysicochcmicaJ cnviromncnr of Ihe colon
Sjodahl R: Ccrcal f~bcr, calcium, and cr~l~r~~cal c~nccr. Cdncrr IS; 7(suypl2):S7Y-S80,19YS.
63(8):2042.2046, ApiI1992. 37. Fliw J, Bonithon-Kopp C: Cllemoprcycnlion ofcolo
12. Butwc JC, Bcdcw L, Bouchc 0, Cadior G, Calis G,dc C&n L, Cuniuy I(lw~CllnccrRrr151:122-133,1999.
T,Du~&ux’hl, Eti:m!rc PL,EiiuD;Faiw J, Lc~uxJL, h’5iqonP,hIom:; 3s. F~~T~J,~~I~C,~~I~~~~~O,RIL~U,C~
F, Nordlin8cr B, Ollicr JC, Pciffcrr D, l’cllccicr G, kr I’, Rougicr I’, T, O’blorai~r 0: Chcmoyrwntion ofmcw
&&one-Fourmcrmux A, Scitz JF, Tril~~ulccJP, Trinthcc JC, Ychou M, bwv;l, dig: and inhrim rcrulB ofa nndol
CCal ~‘4%~ CM bc done for digc~tivc cmnccrin 1999! Rrwmmtndxiocs ECP Colon Group. Eur] Cumr &cv 6( 2):
ofthe Finch Foundado,l OfDiatnivc Cancer (1st I%;). Camvcn~~~lClit: 39. F&c J, Gil:w A: Prj.muyprcvcnrion of c
Ili0!23(4),SO2-511, Apnl1995. ’ ’ slrpylcnicnratioi~. Eur] Cacsff Pro 7(Suyyl
13. Blrbuc JC, Rcdcnnc L, Houchc 0, Wiot G, C&is G, dc Ctiv\L,&r~oy 40. Fllvrc J: C&n canc~7: rpidcmiuloyy, pldl
T, Du~~ux M, &cnnc 12. Eliac D, l%vrc J, Lcgou JL, M3inp P, Morrrcx COWC, ~rca~n~cr~txrld prcwld~~~. Rrvhr44(4):5334 , Fcbrnacy 1;:
F, Nocdlingcr B, Ollicr JC, Pciffcrc D, l’cllccicr G. br I’. lb~~g,k I’, 1994.
P.&one-Fourmc.qmtw A, Scio. JF, Tribou!cc JI’.Trinchcr ]C,Ychou M,
tc d Whar c:1n LX &-XX fwpsdcw wide dibtivc FNXSI iu IYY9! Guidclinw
ofckc Yrcnch Four&don ofDigcsdvc Cwtirolu~((2nC ysrr). C;urwrw:rriJ
UJ’U BIPI 23(5):4%496, May 1999.
14. Bisw G: Clwnoprwrrdon ofco!oticcal ~JXCI. T~vti83(1 Suppl);SS. 43. hd, h’uniriun, and rbc l+&:iun of Cnwc A
55, Janwy/FcbruzT 1997. Vhhingon, D.C.: kmcriw Inrcicutc br Cuxr Rex
44. Fcquson Lq rr al. Studier on chc role nf rpccific d:ca
rknl apiwr c&m& cwccr. MuPu 3X( 1):173-l 84,
4% Fra~cc~<lG 5, FWXO A, hrpind Xl, G~xQs.\ A, LZLVCCC
on colorccnl caner with cmpharis on irrflencc ofccr
I~~‘v~(SU~~$~):SIY-~~,~I.I~ lYY8.
46. Fruwsclli S, PaTiticl M, L? Vccchia C, Favcro A, T
Pals of diffcrcn: WCS of vcg:c:blcs and fruit in bc pr
orclx wh, KCCU~, an,d brew. E~idrmii~!r(qyO(3):33
47. FrccmAn HJ: blc ofhigh fibr: food.< 11 the pr:vcndn
pIasia. C~n/Ga?v:n& 13(5):379-380, June lY9’3.

131(4):612:624, April 1990.
49. Fuchh CS. Giuvznnucci Et.. Coliic/. GA, HWXCI. DJ, SCXT

50. hay CA, Engswm 1% Ch~~oprcw&~ of
and phumxolo~s ~pproarhcs. Oncd~(Hunri~~~) 13(
rioI 97*100.105. Janw 1999.

cndpinrc Clncn74(9’ScppI ):273+-2738,Xovc&zr 1,19%.
28. ECP Cwwnwr Panel on ccrcti ami cwccf: COAUrWl$$b:cnwvon ccrc-
alr, fibru, and wlorcm.l &I brew cmcerr. Pwcchgs uf dlc Europn
Cancer Prwcniion Conwnrur mccdng. Sanh Mqhcricia, Idy, Occubcr
2-S. 1997. Eur 1Cm~wJ’wv’/(Su9912):51-583. hlsy 1993.

tinccrLdrr62(1):11~21, Ycbruuy 14,lW2.
32. Farncsr DL. . Samnlincr
. RE. RK DJ. vu L+cuwcn B. Guillcn 1. Reid M.
Mutincz ME, Manh.~lJK,~Albcrrr DS: Progisrr rspoic: dlc A&w ph&
III study of rhc cfitc of whclr bran fiber on rw!rrcnLsc o! jdcnomxotu
c&n polyps AmJblid 106(1A):435.4%, Jmulry 251999.

h 5(5):375-383; May.1996.
99. Srhcppah W. Binsham S, Boutron-blult MC, Gcr rd.wn dc Vcrdicr
HK, Wtircndorf
J: WHO consaw srscmcnc on the nlc of nutti:ion It I colorew! canccc.

ccr; critical rwicw;nd mcr

107. crchrunicdircws

I?% -
bad on knowlcd~~ vfmcchtisms.

10s. ZimWist EH, cr nl: Gcntlc and cnvironmcnul

cino~cncsir;. n[q Dir J3(6):365-375,
Before the
Washington, D.C.

In re: Food Labeling: Health Claims )
And Label Statements; Request for 1 Docket No. 91N-0098
Scientific Data and Information 1
> (Fiber and colorectal cancer)


I, Charles B. Simone, M.MS., M.D., declare under penalty of perjury that the following is
true and correct to the best of my knowledge, information, and belief:

1. I am a medical oncologist, radiation oncologist, and immunologist. A copy of my
curriculum vitae is attached as Exhibit A.
2. I have investigated the field of nutrition and cancer and conducted research on the
association between nutrition and cancer for more than 22 years.
3. Since 1978 I have studied the association between consumption of dietary fiber and
risk of colorectal cancer. I conclude that vegetable and whole grain fiber
supplementation may reduce the risk of colorectal cancer.
4. The results of two recently published trials wrongly concluded that a recommended
high-fiber, low-fat diet does not reduce the incidence of recurrent colorectal
a. In the Polyp Prevention Trial’ 1037 people received 50 hours of nutritional
counseling over four years and were supposed to eat a 20 percent fat diet and
18 grams of fiber/day; and 1042 people were to eat their regular diet. At the
end of 6 months for the intervention group, and again at the end of each year,
all subjects had to complete a four-day food record of the entire period before.
At the end of 4 years, the weights and cholesterol levels did not change
appreciably in either the intervention or control group even though a 20% fat,
high-fiber diet should have lowered weight and cholesterol. The reliance on
self-reporting over such a long period of time introduces a high probability of
inaccuracy that makes the study results unreliable, There is no way to
determine, under the study design, whether subjects actually complied with
consumption restrictions or simply reported compliance yet actually failed to
follow instructions. The study methodology and results suggest that the
subjects may not have adhered instructions and may have written down foods
they knew would comply \vith what they were supposed to be eating in order
to stay in the study. Most of the subjects in this study were male and had an
average age of 62.

’ Schatzkin A, Lanza E, Corle D, et al, and the Polyp Prevention Trial Study Group. Lack of effect of a
low-fat, high-fiber diet on the recurrence of colorectal adenornas. N Eng Jh’ed 2000; 342: 1149-55.
b. The second trial studied 1303 people in a 34 month period.* Some subjects
wcrc asked to consume either a high-fiber supplcmcnt ( 13.5 grams/day) or a
low-fiber supplement (2 grams/day). Measurement of compliance was even
more “challenging.” Compliance with the protocol was evaluated primarily
by counts of returned cereal boxes and fiber bars at each visit and secondarily
through a specialized intake calendar. Weights and choles\erol levels were
not indicated in the paper. .Uy conclusions based upon this study would be
invalid because of a lack of assurance of compliance with trial protocols in the
treatment or tht: control groups. Most of the subjects in this trial wcrc male
and had an average age of 66.
6 In reviewing the results of the trials, NC1 scientists offcrcd the following as an
attempt to explain why the studies’ designs did nol show an effect of diet on polyp

1. Development ofcolureclal cancer takes decades; an intervention UTihrec to
four years may not be long enough.
2 Nutritional factors may intluence criticul molecular, cellular, or tissue-level
events in colorectal cancer formation well before polyps are formed.
3. The recurrent polyps tended to be small. Dietary changes might affect only
the growth of small polyps into large polyps or large polyps into invasive

7. Although the points raised by NC1 arc important, the conclusions of both studies are
invalid because there is little or no evidence of complizmce and thcrcfore no
confirmation of the amount of fiber consumed by either the treatment groups or
cvntrvl gruups. The flaws in rhc study render them outlicrs, not useful in assessing
the weight of scientific evidence concerning the fibcr/colorectal cancer association.
There is little doubt, based on the overall body of publicly available scientific
evidence, that a high-fiber diet reduces the risk of coiorectal cancer.’ Well-designed
and reliable studies that stand for this latter proposition an: described in the European
Cancer Prevention (ECP) Consensus Statement, The Review of Food, N&lion, and
the Prevention of Cancer: A Global Perspective, and the World Health Organization
(Exhibit B).

Charles B.‘Simonc, M.MS.. M.D.

2 Albert5 TX, Martinc~ ME, Roe l3J. et al, and the Phoerlix Colon Cancer Prcvcrltion Physicians’ Network
Lack of cttcct ol‘a high-fiber ccrcal supplcmcnr on rhe recurrence of colorccral aderratnas N Eng J Mu/.
2000; 342: 11 S6-62.
’ NC1 “Trials Show No Effect of Luw-F&. Iligh-Fiber., and High-Fruit eod Vegctnblc DWY on the Growrh
of New Colorectal Polyps in People with a History or Prtttutzarous Polyps.” NCI Press Release, April 19.
2000, www nih govlnewslpr/apr2000/nci-l9.hfm.
’ Simone CB, Simone NL, CB Simonc 11. Consumption of fiber reduces the risk uisolorecralcancer: A
&view. Zn~urnofiona~JInlcgraiive Med. July-August ZUUU

Name: Charles B. Simone, M.MS., M.D

Present Address: 123 Franklin Corner Road
Lawrenceville, NJ 08648

Date and Place of Birth: June 21, 1949 in Trenton, NJ

Marital Status: Married, two children,

Military Service: 1977-1982 Commander, U.S. Navy, Public HealthService

1967-1971 - B.A. (Biological Sciences) - Rutgers University, New Brunswick, NJ

1971-1975 - M.MS. and M.D. - Rutgers Medical College, Piscataway, NJ

Positions Held:

1967-1971 Research Assistant to Ralph J DeFalco, Professor of
Immunology, Rutgers University, New Brunswick

1970-1972 Consultant for criminal investigations requiring
immunological corroborations.

1971-1972 Acting Chairperson of Rutgers University Serological

1974-1975 Research Appointment with Robert A. Good, Ph.D., M.D.,
President and Director of Memorial Sloan Kettering Cancer Hospital, New York City.

1975-I 976 Internship, Department of Medicine, The Cleveland Clinic Foundation,
Cleveland, Ohio.

1976-l 977 First Year Assistant Resident, Department of Medicine,
The Cleveland Clinic Foundation, Cleveland, Ohio.

1977-1979 Clinical Associate, Immunology Branch, National Cancer
Institute, National Institutes of Health, Bethesda, MD.

1978-l 980 Clinical Assistant Professor of Medicine, George
Washington University School of Medicine, Washington, D.C.

1979-l 980 Clinical Associate, Medicine Branch, National Cancer
Institute, National Institutes of Health, Bethesda, MD.

1980-l 982 Investigator, Clinical Pharmacology Oncology Branch,
National Cancer Institute, National Institutes of Health, Bethesda, MD.

1980- present Founder, Director, Simone Protective Cancer Institute,
Lawrenceville, NJ.
1982-1985 Radiation Therapy Department, Hospital of the University
of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104.

1984-1989 Consultant, New Jersey Education Oncology Program.

1984-1987 Advisor, U.S. Postal Service, Preventive Health.

1984-present Speaker for the American Cancer Society.

1985-l 988 Associate Professor, Radiation Therapy and Nuclear
Medicine Department, Thomas Jefferson University Hospital, Philadelphia, PA.

1985-l 988 Chief, Breast Section, Radiation Therapy and Nuclear
Medicine Department, Thomas Jefferson University Hospital, Philadelphia, PA.

1985-l 989 Consultant, lmmunobiochemistt-y for BASF

1985-l 988 Chairman, Publication Review Committee, Thomas Jefferson
University Hospital, Philadelphia, PA.

19861991 Consultant for Hoffmann-LaRoche, Nutley, NJ. “Protector
Nutrient” Program.

1986-1988 Member, Jefferson Hospital Nutrition Committee.

1986-l 988 Speaker for Jefferson Educational Program

1988-l 992 Medical Advisor to NJ Governor - Substance Abuse

1989-l 991 Consultant to Spain

1989-pres Consultant to Cambodia

1989- pres Consultant to Russia

1990-pres Medical Advisor to National Alliance of Breast Cancer Organizations

1991-pres Consultant to Chechen

1993-pres Advisor to US Senators Tom Harkin and Orin Hatch

1993-l 995 Advisor for organization and implementation of Office of Alternative Medicine,
National Institutes of Health

1993-pres Consultant to U.S. Senate, Expert Witness

1993-pres Consultant to U.S. House of Representative, Expert Witness

1994 Consultant to Senate Committee: I wrote the language that assured passage of the Dietary
Supplement Health and Education Act of 1995 that ensured that all Americans have free access
to food supplements.

1995 Consultant to New Jersey Medical Board, Expert Witness

1996pres Advisor for organization and implementation of Departmentof Alternative Medicine,
College of Physicians and Surgeons of Columbia University, New York

1997 Judge for Mrs. America Pageant

1997 Participant in National Cancer Institute and Office of Alternative Medicine POEMS

1998 Consultant to U.S. House of Representative, Expert Witness

1998 Consultant to U.S. Senate, Expert Witness

1998 Judge for Mrs. America Pageant

1998-pres Organizing new health care.system for Chechen.

1998-pres Editor, Women’s Health Alternative Medicine Report ’


Diplomate of the National Board of Medical Examiners 1976

American Board of Internal Medicine, Eligible 1978

Medical Oncology Subspecialty Board, Eligible 1980

Allergy and Immunology Board, Eligible 1980

Radiation Oncology Subspecialty Board, Eligible 1983


Honorable Mention Award - SAMA Research Forum April 1973
Visiting Professor in Rheumatology, Cleveland Clinic 1979
Visiting Professor in Clinical Immunolgy, University of Hawaii 1979
Elected into New York Academy of Sciences 1983
Elected into American College of Immunologists 1983
American Academy of Sciences 1984
Elected, Who’s Who in Frontiers of Science and Technology 1984
Elected, Contemporary Authors 1984
Author Citation Award, Nineteenth Annual New Jersey Writers
Conference, NJ Institute of Technology, 1986
Visiting Professor in Immunology and Oncology, Cleveland Clinic 1987and 1989

Invited/Special Lectures:

Lecture to Radiation Therapeutic Oncology Group 1983
Keynote Speaker - 18th Annual Congress, AACIA 1984
Keynote Speaker - Annual Cancer Symposium, University of Louisville 1985
Speaker - New Jersey State Justice Department 1985
Speaker - United States Arsenal, Picattiny, NJ 1985
Keynote Speaker - New Jersey Superintendents’ and Principals’
Keynote Speaker - New York Open Center 1986
Keynote Speaker - New Jersey Superintendents’ and PrincipalsConvention 1986
Keynote Speaker - New Jersey State Kiwanis 1985
Speaker for Jefferson Outreach Program
Keynote Speaker - New Jersey Superintendents’ and Principals’ Convention 1989

Please see Media Events for a More Complete Listing of all SpeakingEngagements.


New York Academy of Sciences

American College of Immunologists

American Academy of Sciences

Contemporary Authors

Military Service:

1977-1982 Commander, U.S. Public Health Service, Navy

Licensed to Practice Medicine:

New Jersey

1. Katz, P., C.B. Simone, P.A. Henkart, and A.S. Fauci. 1980
Mechanisms of antibody-dependent cellular cytotoxicity. The use of effector cells from chronic
granulomatous disease patients as investigative probes. J Clin Invest. 65:55-63.

2. Simone, C.B., and P.A. Henkart. 1980.
Permeability changes induced in erythrocyte ghost targets by antibody-dependent cytotoxic effector
cells: evidence for membrane pores. J Immunol. 124:954-963.

3. Dourmashkin, R.R., P.Deteix, C.B. Simone, P.A. Henkart. 1980.
Electron microscopic demonstration of lesions in target cell membranes associated with antibody-
dependent cellular cytotoxicity. Clin Exp Immunol. 42:554-560.

4. Dourmaskin, R.R., P.Deteix, C.B. Simone, P.A. Henkart. 1980.
Electron microscopic evidence of membrane assoicated pores with antibody- dependent cellular
cytotoxicity. J Immunol. 124:1518.

5. Simone, C.B., and P.A. Henkart. 1982.
Inhibition of marker entry into complement treated resealed erythrocyte ghosts by anti-C5.
J Immunol. 128:1168-1175

6. Myers, C., L. Gianni, C.B. Simone, R. Klecker, and R. Greene.
1982. Oxidative destruction of erythrocyte ghost membranes catalysed by the doxorubicin-iron
complex. Biochemistry. 21(8): 1707-I 713.

7. Simone, C.B. 1982. Directed effector cells selectively lyse human
tumors. Nature. 297:234-236.

8. Simone, Charles B. 1983. Cancer and Nutrition, A Ten Point Plan to
Reduce Your Risk of Getting Cancer. McGraw-Hill Book Company. 265 pages.

9. Kligerman, M.M., Glover, D.J., Turrisi, A.T., Norfleet, A.L., Yuhas,
J.M., Coia, L.R., Simone, C.B., Glick, J.H., and Goodman, R.L. 1984.
Toxicity of WR-2721 administered in single and multiple doses. Int J
Radiation Oncology Biology and Physics. IO:1 773-l 776.

10. Simone, C.B. 1985. What is your cancer risk? Immunol. Allergy
Pratt. 7(10): 47-51.

Il. Simone, C.B., and M. Mohiuddin. 1987. Radiation Therapy and
Cancer: A Practical Guide for Referring Physicians. Network for
Continuing Medical Education. No. 497.

12. Simone, C.B., and C.Mansfield. 1987. Perioperative lr-192 implant
at time of lumpectomy for breast cancer. Radiology. Nov Supplement,

13. Simone, C.B. Cancer and Nutrition. In: Therapeutic Radiology.
Editor: Mansfield, CM. 1989. Elsevier Science Piublishing Co. Inc.

14. Simone, C.B. 1992. Cancer and Nutrition, A Ten Point Plan to
Reduce Your Risk of Getting Cancer. Revision. Avery Publishing. 338 pages.

15. Simone, Charles B. 1993. How to reduce America’s runaway health
care costs. Federal Register. Subcommittee of the Committee on
Appropriations United States Senate, Senator Tom Harkin.
16. Simone, Charles B. 1993. Treatment of advanced cancers using
shark cartilage. Federal Register. Subcommittee of the Committee on
Appropriations United States Senate.

17. Simone, Charles B. 1994. Cancer and Nutrition, A Ten Point Plan to
Reduce Your Risk of Getting Cancer. B. Jain Publishers, Pvt, Ltd. New
Delhi, India. 338 Pages.

18. Mansfield, C.M., Komarnicky, L.T., et al. 1994. Perioperative
implantation of iridium-192 as the boost technique for stage I and II breast cancer: results of a 1 O-
year study of 655 patients. Radiology. 192(l): 33-36.

19. Simone, Charles B. 1994. The Scientific Standard Already Exists
for Food Supplement Health Claims. Federal Register. I wrote the language
that assured passage of the Dietary Supplement Health and Education Act
(DSHEA) of 1995 that ensured that all Americans have free, access to food

20. Mansfield, C.M., Komarnicky, L.T., et al. 1994. The role of
radiotherapy in the treatment of breast cancer: results of
perioperative implantation. Cancer Detect Prev. 18(6): 493-499.

21, Simone, Charles B. 1994. Krebs und Ernahrung, Ein Sehn-Punkte Plan zur Verringerung des
Krebsrisikos. Quintessenz Verlag-GmbH, Berlin, Germany 432 Pages,

22. Simone, Charles B. 1994. Rehydration formula. US Patent Issued.

23. Mansfield, C., Schwartz, G. et al. 1995. Ten-year results in 1070 patients with Stages I and II
breast cancer treated by conservative surgery and radiation therapy. Cancer. 75(9): 2328-2336.

24. Simone, Charles B. 1995. Breast Health. Avery Publishing. 410 pages,

25. Simone, Charles B. 1995. Rehydration formula. Canadian Patent Issued.

26. Simone, Charles B. 1995. Shark Cartilage and Cancer. Paradigm Press. Monograph.

27. Simone, Charles B. 1996. Rehydration formula. European Patents Issued.

28. Simone, Charles 8. 1996. How to augment oncology care. In: Alternative Medicine Definitive
Guide for Cancer. Future Medicine Publishing, Inc. Tiburon, CA.

29. Simone CB, Simone NL, Simone CB II. Oncology care augmented with nutritional and lifestyle
modification. J Ortho Mol Med 1997; 12(4):197- 206.

30. Simone CB, Simone NL, Simone CB II. Folic acid does not interfere with methotrexate. Lancet.
1997; 350:1556.

31. Simone, CB. Chinese language and distribution of Cancer and Nutrition, A Ten Point Plan to
Reduce Your Risk of Getting Cancer. Redwoods Publishing Company through Big Apple Tuttle-Mori
Agency. 1997.

32. Simone, CB. Cancer. In The Complete Book of Alternative Nutrition. 1997. Rodale Press.
Emmaus, PA.

33. Simone CB, Simone NL, Simone CB II. Oncology Care can be Augmented with Nutritional and
Lifestyle Modification. In: Prasad KN and Cole WC, ed. Cancer and Nutrition. 1998. Amsterdam,
Netherlands. IOS Press

34. Simone CB. Nutritional and Lifestyle Modification in Oncology Care. In: Torosian M. Integrated
Cancer Management: Surgery, Medical Oncology, and Radiation Oncology. 1998. Marcel Dekkar,
New York.

35. Simone CB. Food and Drug Administration Reform before the Government Reform and
Oversight Committee, Chariman Dan Burton. Federal Register, February 4, 1998.

36. Simone CB. Misinformation given to patients regarding food supplement with chemotherapy and
radiation therapy. Food and Drug Administration Reform before the Government Reform and
Oversight Committee, Federal Register, February 1998.

37. Simone CB, Benjamin SA, Traub M. Beyond the conventional: Cancer Treatment Part 3. Patient
Care, The Practical Journal for Primary Care Physicians. 1998; 53-68.

38. Simone CB, Simone NL, Simone CB II. Shark cartilage and cancer. Lancet. 1998; 351: 1440

39. Charles B. Simone, Nicole L Simone, Charles B. Simone, Il. Nutritional and Lifestyle Modification
to Augment Oncology Care. J American College of Nutrition. October 1998; 17(5): 496-497.

40. Charles B. Simone, Nicole L Simone, Charles B. Simone, II. Oncology
Care is Augmented with Lifestyle Modification. Fourth Annual International Congress on Alternative
and Complementary Therapies. October 1-4, 1998. Arlington, VA.

41. Simone CB, Simone NL, Simone CB II. Do we always need to tell patients the truth? Lancet.
1998; 352: 1787.

42. Simone CB, Simone NL, Simone CB II. Nutrients and cancer treatment. International J
Integrative Med. 1999.1(1):22-27.

43. Simone CB, Simone NL, Simone CB II. Empowering patients by telling the truth. Women’s Health
Alternative Medicine Report. 1999; 1 (7): 6-8.

44. Simone CB. Breast Cancer; Cancer. In: Rodale Press Editors. Nature’s Medicine. 1999.
Emmaus, PA.

45. Simone CB. Foreword; and Are Antioxidants Compatible with Chemotherapy and Radiation
Therapy. In: Drs Eberhard and Phyllis Kronhausen eds. Formula for Life. 1999. Quill William
Morrow Press, New York, NY.

46. Simone, CB. Japanese language and distribution of Cancer and Nutrition, A Ten Point Plan to
Reduce Your Risk of Getting Cancer. lmamura Publishing Company, Tokyo, Japan. 1999.

47. Simone CB, Simone NL, Simone CB II. Conjunctivitis, rosacea, and Helicobacter pylori. Archives
Opthalmology. 1999; in press.

48. Simone CB, Simone NL, Simone CB II. Regression of mucosa-associated lymphoid-tissue
(MALT) lymphoma of the colon after eradication of Helicobacter pylori. Manuscript submitted.
49. Simone CB, Simone NL, Simone CB II. Kombucha tea can cause liver dysfunction. Manuscript

For other articles, papers, or chapters written by me for the lay
audience, please refer to Print Media.
~~ropem~ournnlofCancerPrevenrion 1990, 6, 512-514

Consensus meeting on cereals, fibre and
colorectal and breast cancers.

ECP consensus panel on cereals and cancer -
This Consensus Meeting was held in Santa Margherita, Italy 2-5 October and was attended by
17 experts in the field of diet and cancer; A further 5 who could not attend the discussions were
‘corresponding participants’ and gave their views by post and telephone. The agreed consensus
statement was as follows:

l A diet rich in high-fibre cereal is associated with a reduced risk of colorectal cancer.
o There is suggestive evidence that cereal fibre protects against breast cancer.
o There is good reason to examine the relationship between cereal fibre intake and risk of cancer at other sites.


All plant foods contain plant cell walls containing B vitamins and protective agents. At other times of
dietary fibre (DF) and a range of other agentswhich day cereals are usually eaten as breads, pasta, rice,
are suspected to be protective or anticarcinogenic pastries etc. These are usually made from low extrac-
(eg vitamins, antioxidants, tannins, polyphenolics, tion cereals which contain lower levels of DF and
flavonoids etc). In general vegetables contain rela- other protective agents and anticarcinogens; whole-
tively modest amounts of DF but are rich in a wide meal breads and products are richer in both.
array of protective agents and anticarcinogens, the Different cereals contain different amounts of DF
amounts and classes of which vary between and anticarcinogens (rice has least and wheat and rye
vegetable type. Whole grain cereals are relatively have most of both). Further, rice, which is most com-
rich in DF and also contain protective agents such monly eaten in Europe in the southern countries, is
as phytate and a range of anticarcinogens. However almost always eaten in polished and refined form and
these latter are partially removed with the husk so contains even less DF and anticarcinogen than
during milling. Fruits contain the least DF bit usual. The cereals which are most often consumed
contain an array of anticarcinogens which differ in unrefined and high extraction form are wheat
from those in cereals and vegetables. and ‘ye, but rye is rarely consumed in the southern
Current hypotheses suggest that fruit and vegeta- countries.
bles protect against cancer at a wide range of sites The postulated mechanisms of action indicate that
mainly through the action of their anticarcinogens. the pr0tectiL.e action will be greater in the unrefined
In contrast cereals have been assumed in the past cereal than in that in which the husk has been
to act mainly through the action of DF. removed by milling, In most epidemiological studies
In this Consensus Statement ‘cereal fibre’ will the cereals are primarily low extraction products and
imply cereal retaining a high proportion of its husk so are low in DF and other protective agents. A
(and the accompanying anticarcinogens) intact. major conclusion was that, in future, questionnaires
In Europe cereals may be consumed as breakfast should be framed to distinguish between low extrac-
cereals which are often rich in DF and also rich in tion and high extraction cereals.

512 Europsan Journal of Cancer Prcvenfion. Vol 0.199s 0 1997 Rapid Scicnx Publishers
Consemris meeting

Colorectal cancer Apart from-fermentable cell wall polysaccharide
and starch. cereal foods also contain phenolic sub-
stances and phytate which ma}’ be important intralu-
A cfiet rich itz high fibre cereal is nssocinted with n
minal antioxidants. Faecal material containing trace
reduced risk of colorecta/ cnncer.
quantities of free iron has been shown to be a source
In support of this we cite the review of 5s previous of free radicals which can probably enhance the pro-
studies of diet and colon cancer; cereal fibre was mea- duction of carcinogens, or damage crypt cell DNA
sured in only 19 studies. Of these, 16 reported an directly. Phytate can chelate iron and hence suppress
inverse association betiveen cereal fibre and colon intraluminal free radical production. Moreover phe-
cancer risk and the other 3 showed no relation. In nolic substances such as flavonols and tannins are
addition a revieiv of FAO data showed that there is effective antioxidants and may quench free radical
an inverse relation between the risk of colorectal and mediated chain reactions in the gut lumen, Phytates
of breast cancer and cereal and vegetable disappear- and other low molecular \veight,species associated
ance, no relation with fruit and starchy root intake, with plant cell walls may also act as anticarcinogens
and a positive correlation with total energy intake. by upregulating epithelial cell differentiation, sup-
The data are consistent with those from the Italian pressing mitosis or stimulating apoptosis and thereby
study where, in the context of the Italian diet, high deleting potentially cancerous cells from the mucosa.
consumption of pasta was shown to be a major con- An exciting area which is receiving more and more
tributor to high total energy intake and was partly or attention recently is the interaction between environ-
largely related to the risk for cancers of the colon and ment and genes in the colon. Mutations in several
breast. This suggested that the real association was genes control& 0 celi division, apoptosis and DNA
with total energy intake. This consensusreaffirms and repair have been implicated in tumour development.
extends that reached by the Colon Group at the Already some of this work has implications for the
VW0 Consensus Conference in Stuttgart in 1996 effect of dietary fibre on colon cancer development.
(European Journal of Cnncer Prevention 6 404-407), Perhaps the most interesting to date is the observa-
and with the COLqA recommendations in the UK. tion that the short chain fatty acids (SCFA) acetate,
A variety of mechanisms has been proposed for propionate and particularly butyrate can induce
the protective effect of cereal fibre. Burkitt popu- apoptosis in colonic cells in culture; this gives a plau-
larised the idea that a diet high in fibre-rich foods sible hypothesis for the protective action of fer-
could influence the course of colorectal carcinogen- mentable fibre in colorectal carcinogenesis. In
esis. He proposed that it nas fermentation of the addition there is evidence that gut factors, including
fibre itself that gave the protection through (a) bile acids and SCFA can (I) interact with mutated
increased faccal ireight, (b) increased frequency of APC gene, (ii) modulate expression of the ~53
defecation, (c) decreased transit time, and (d) dilu- tumour suppressor gene, and (iii) modulate expres-
tion of the colonic contents. The evidence is strongest sion of transcription factors important in control of
for (a) and (d). In addition he proposed that fibre cell division. We now need to characterise the inter-
metabolism influenced microbial grotvth in the action of fibre components in the lumen with the
colon - an area about which we know very little. above gut factors to ascertain whether fibre is indi-
More recently, mechanisms involving the metabolic rectly affecting gene expression in the colon. For
consequences of fibre metabolism have been pro- example, colonic fibre may be influencing the control
posed including (c) alteration of energy metabolism. of cell division through effects on formation and sol-
It is no~v generally accepted that energy restriction ubility of secondary bile acids in the colonic lumen.
will inhibit carcinogenesis and a fibre-rich diet may
make a contribution to overall energy management;
(f) influence on bile acid metabolism, a theory that Breast cancer
refuses to go an’ay; (g) production of short-chain
fatty acids (SCFA) Lvhich may inhibit carcinogenesis There is sr1,ogestil.e evidence rhnt cerealfibre protects
through its effect on colonic pH and through the qyinst breast cmlcer.
supply of butyrate. This latter has been shotvn in Although there are many epidemiological studies
vifro to promote apoptosis and cell differentiation, showing a protective effect of cereal fibre, some oth-
both of nhich are central to the carcinogenesis ers sho\v no such effect and there is insufficient evi-
process. Zrr r~iro verification of these actions is still dence to reach a strong conclusion. The LVHO
awaited. Consensus Group on Breast Cancer, in Stuttgart,

Europzan Jourml of Cancer Prevenrion. Sol 0 199s 513
Consetwir rneefitlg

concluded that the epidemiological e\,idence \vassug- mechanism similar to that proposed for ,vegetables
gesti1.eof a protective effect (as did \ve) and svent on and fruits. If such a generalised protection \vas con-
to sugges; that cereal fibre consumption should there- firmed it \vould of course strengthen the recommen-
fore be increased. dation to increase intake of high fibre cereals.
It is generally accepted that high levels of circulat-
ing oestrogens and insulin growth factor (IGF-1, part General Recommendations
of the insulin-resistance syndrome together with Questionnaires need to be directed in future also
abdominal obesity, high plasma insulin levels and
to the study of food groups (eg cereals) rather
other hormonal changes) represent major risks for than nutrients or anutrients (eg dietary fibre),
the development of breast cancer. This is because
since the latter are highly heterogeneous and not
they induce [oestrogens) or are (IGF-1) growth fac- necessarily lvell quantitated.
tors for mammary tumour cells. It has been proposed In view of the data presented in the review by
that a high intake of fibre affects the risk of breast Hill (1997), a pooled analysis of the foods and
cancer through an effect on these factors. Diets low food group rich in cereals and cereal fibre to
in fat and rich in cereal fibre reduce levels of plasma determine the importtince of cereal fibre in the
oestradiol, oestrone and oestrone sulphate. This may case-control and the cohort studies of diet and
be through interfering with their enterohepatic cir- colorectal cancer should be carried out.
culation and so increasing their rate of faecal excre- hlany of the effects of dietary fibre in protecting
tion. Dietary fibre contains phytoestrogens (iso- against colorectal and breast cancers are
flavonoids), which could modulate the activity of concerned with events in the caccum and prox-
endogenous oestrogens. Fibre intakes have also been imal colon. We need to understand much more
shown to be inversely related to total, subcutaneous about the ecology of this important but experi-
and extra-abdominal fat and to lower insulin levels. mentally inaccessible subsite of the large bowel.
These findings reflect the influence of fibre in con-
trolling aspects of the insulin-resistance syndrome.
Other mechanisms have been proposed which Signatories to the consensus statement
include the trapping of carcinogens, the regulation of Dr Michael Hill (Chairman Slough, UK,
cell proliferation, through a direct effect of Chairman
of ECP)
isoflavonoids such as genestein and apigenin on the Dr Attilio Giacosa Genoa, Italy
ceil cycle, or through an activation of the PKC. (Scientific Coordinator
of ECP)
Other sites Dr David Beckley Plymouth, UK
Dr Christine PJ Caygill Slough, UK
There is goocf reflSo?lto examine seriously the Dr Paula Chaws Lisbon, Portugal
rdationship be:wcetl cereal fibre intake ad cnncer Dr David Evans London, UK
41‘other sites. Prof. Jean Faivre Dijon, France
An analysis of the Italian data suggested Dr Fabio Farinati Padua, Italy
that people ivho reported consuming whole grain Prof Silvia Franceschi Aviano, Italy
cereals ivere at a lower risk of cancer at a range of Dr Miquel Gassull Badalona, Spain
o:her sites in addition to the large bowel and breast. Dr Mariette Gerber Montpelier, France
There \<‘ere many potential confounding factors that Dr Ian T Johnson Korwich, UK
could esplain these Italian data, and they need to be Dr David Kritchevsk) Philadelphia, USA
confirmed. Holvever there are good theoretical rea- Prof Carlo La Vecchia hiilan, Italy
sons for suspecting such a general protective effect. Prof Paul hIainguet Brussels, Belgium
Ii the mechanisms proposed to explain the protecti5.e Dr Alain hfaskens Brussels, Belgium
effects against breast cancer are true then \se ~soulcl Dr Robert iV Oiven Heidelberg, Germany
expect them to apply also to other hormone-rslated Dr Joseph Rafter Stockholm, S\veden
cancer sites such as the cndometrium. ovary and Prof Ian I\: RoLvland Irish United Nutr.
prostate. Carcinogen binding in the colon lumen Assoc
might also gii.2 rise to a generalised protection, and DI- David Southgate i\‘orLvich, UK
the presence of isoflavonoids. tannins and other phe- Prof Reinhold Stockbrugger hlaastricht,
nolic compounds in the cereal husk would prol,ide a The Netherlands

511 EuropwlJournal
of CnwzrPrcvcntion. Vol 0 199s