Role of Mitomycin in Combination With Fluorouracil
and Radiotherapy, and of Salvage Chemoradiation in the
Definitive Nonsurgical Treatment of Epidermoid
Carcinoma of the Anal Canal: Results of a Phase III
Randomized Intergroup Study
By Marshall Flam, Madhu John, Thomas F. Paajk, Nicholas Petrelli, Robert Myerson, Scotte Doggett, Jeanne Quivey,
Marvin Rotman, Herbert Kerman, Lawrence Cola, and Kevin Murray
Purpose: Definitive chemoradiation (CR) has replaced
radical surgery as the preferred treatment of epidermoid
carcinoma of the anal canal. To determine the impor-
tance of mitomycin (MMC) in the standard CR regimen
and to assess the role of salvage CR in patients who have
residual tumor following CR, a phase III randomized trial
was undertaken by the Radiation Therapy Oncology
Group (RTOG)/Eastern Cooperative Oncology Group
(ECOG).
Patients andMethods: Between August 1988 and De-
cember 1991, 310 patients were randomized to receive
either radiotherapy (RT) and fluorouracil (5-FU) or radio-
therapy, 5-FU, and MMC. Of 291 assessable patients,
145 received 45 to 50.4 Gy of pelvic RT plus 5-FU at
1,000 mg/m 2/d for 4 days, and 146 received RT, 5-FU,
and MMC (10 mg/m 2 per dose for two doses). Patients
with residual tumor on posttreatment biopsy were
treated with a salvage regimen that consisted of addi-
tional pelvic RT (9 Gy), 5-FU, and cisplatin (100 mg/m 2 ).
E
I
PIDERMOID CARCINOMAS of the anal region
comprise only 1% to 2% of large bowel cancers
and 3.9% of anorectal carcinomas.' Nevertheless, these
neoplasms have dramatically increased in importance
over the past 20 years as a model for organ preservation
with the use of concomitant radiation (RT) and chemo-
therapy in lieu of radical surgery. 2 Before the application
of chemoradiation (CR) regimens, the 5-year survival rate
of patients with anal epidermoid malignancies managed
by radical surgery ranged from 40% to 60% and local
relapse was common.34
The combined modality regimen initially described by
Nigro et a 5 in 1974 was a preoperative regimen that
subsequently evolved into a definitive treatment program
by 1983.67 More than 500 patients treated with CR using
fluorouracil (5-FU) infusion and mitomycin (MMC) have
been reported over the past 18 years.6 '12 Complete and
sustained tumor eradication has been reported in 75% to
90% of patients reported in series in which both modal-
ities are used concomitantly. 7 °8' '0"1' 3 ' 20 All reported regi-
mens use variations of both 5-FU infusion and MMC
chemotherapy as initially described by Nigro et al. 5 How-
ever, there are no randomized data regarding the relative
importance of each component used in this combined
modality regimen. Some investigators 22 have questioned
the necessity of MMC in this regimen, while Cummings
Journal of Clinical Oncology, Vol 14, No 9 (September), 1996: pp 2527-2539
Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083
Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
Results: Posttreatment biopsies were positive in 15%
of patients in the 5-FU arm versus 7.7% in the MMC arm
(P = .135). At 4 years, colostomy rates were lower (9%
v 22%; P = .002), colostomy-free survival higher (71% v
59%; P = .014), and disease-free survival higher (73%
v 51 %; P = .0003) in the MMC arm. A significant differ-
ence in overall survival has not been observed at 4 years.
Toxicity was greater in the MMC arm (23% v 7% grade
4 and 5 toxicity; P < .001). Of 24 assessable patients
who underwent salvage CR, 12 (50%) were rendered
disease-free.
Conclusion: Despite greater toxicity, the use of MMC
in a definitive CR regimen for anal cancer is justified,
particularly in patients with large primary tumors. Sal-
vage CR should be attempted in patients with residual
disease following definitive CR before resorting to radi-
cal surgery.
J Clin Oncol 14:2527-2539. © 1996 by American So-
ciety of Clinical Oncology.
et al,23 in a series of nonrandomized trials, reported im-
proved local control with MMC-containing regimens
when compared with regimens in which 5-FU and RT
were used without MMC.
In view of the high success rate of this combined mo-
dality program and the increased toxicity attributed to
From the University of California, San Francisco,Fresno; Radia-
tion Oncology Center, Sacramento; University of California, San
Francisco, CA: Statistical Section, Radiation Therapy Oncology
Group Headquarters;Fox Chase Cancer Center, Philadelphia, PA;
Roswell Park Cancer Institute, Buffalo; State University of New
York Health Science Center, Brooklyn, NY: Washington University
Medical School, St Louis. MO: Halifax Hospital Medical Center,
Daytona Beach, FL; and Wisconsin Medical Complex, Milwaukee,
WI.
Submitted August 3, 1995: accepted April 1, 1996.
Supported by grants no. CA-32115 and CA-21661 from the Na-
tional Cancer Institute, Bethesda, MD.
Presented in part at the Thirty-FirstAnnual Meeting of the Ameri-
can Society of Clinical Oncology, Los Angeles, CA, May 23, 1995.
Address reprint requests to Marshall S. Flam, MD. Hematology-
Oncology Medical Group, Cancer Center at Saint Agnes, 7130 N
Millbrook Ave, Suite 100, Fresno, CA 93720; Email MSLEVEN@
SAMC.com.
© 1996 by American Society of Clinical Oncology.
0732-183X/96/1409-001 7$3.00/0
2527
2528 FLAM ET AL

MMC, a randomized study was initiated by the Radiation
Therapy Oncology Group (RTOG) and Eastern Coopera-
tive Oncology Group (ECOG) to determine whether
MMC is a necessary component of this efficacious com-
bined modality treatment. A previous phase I/II study
(RTOG 83-14) suggested that there were sufficient pa-
tients available to launch a phase III trial.' 6
Historically, abdominoperineal resection has been used
for salvage attempts in the majority of the 10% to 20%
of patients who had either gross or microscopic residual
disease following combined CR treatments.7 81 ' 3 16,2 1 How-
ever, it appears that surgical salvage is often ineffective
in preventing subsequent local recurrence and death from
distant metastases.8' 2 In 1987, Flam et al" described an
effective CR salvage regimen using 5-FU infusion, cis-
platin, and additional RT. The randomized study we re-
port here used salvage CR in all patients with residual
disease following definitive CR, to assess the efficacy of
nonsurgical salvage therapy. Only if patients demon-
strated residual tumor on biopsy following salvage CR
were they subjected to abdominoperineal resection.
PATIENTS AND METHODS
Before activation of the randomized trial, a pilot study was under-
taken to determine the tolerable dose of MMC that could be adminis-
tered twice during RT. Based on 19 patients treated in a phase II
component of the study, the tolerable dose was determined to be 10
mg/m'2.
The randomized portion of the study began accrual in August
1988 and closed to new patient entry in December 1991. A total of
310 patients were randomized from 104 institutions that contributed
to the study through the RTOG or ECOG. Eligibility criteria included
any epidermoid malignancy of the anal canal in which the primary
tumor was measurable and presenting with any tumor or nodal stage.
A Karnofsky performance status 2 60 was required.
Before entry, patients were required to undergo a staging work-
up that consisted of a chest x-ray, computed tomographic (CT) scan
of the abdomen and pelvis, and a bipedal lymphangiogram if the CT
scan was positive for pelvic lymph nodes. Laboratory investigations
consisted of complete blood cell counts, and blood chemistries, in-
cluding liver and renal function tests. Tumor size and extent was
indicated on an anatomic diagram. A fine-needle aspiration biopsy
of any enlarged inguinal or pelvic lymph nodes was required to
stage the patients properly before entry. If the fine-needle aspiration
biopsy was negative, an excisional biopsy of one inguinal lymph
node was required. Determination of creatinine clearance was re-
quired before salvage therapy.
All patients gave informed written consent before they were ran-
domized. Patients were stratified before randomization by nodal sta-
tus (NO v NI), histology (keratinizing squamous nonkeratinizing
squamous), and primary tumor size (< 5 cm v 2 5 cm). The random-
ization scheme derived by Zelen 24 was used to achieve balance in the
treatment assignments among the institutions with three stratification
variables.
The study schema is demonstrated in Fig 1. The RT dose and
fields. as well as the 5-FU dose and timing, were identical in the
two treatment arms, which differed only in the absence or adminis-

Chemotherapy

Cycle
#1 Cycle #2 Cycle#3 |
Cheno Chemo Chemo
AP Resection

Radiation Therapy

cy x BxlC)
45 Gy/25 f5 wks. 9.0 G f

b,
Fig 1. Treatment schema of
phase Ill randomized study of 5-
No treatment I No treatment FU + RT v 5-FU, MMC, and RT
Da- l I I I I I I in carcinoma of the anal canal-
llr
0 1 4 28 32 35 63-77 60-74 105-126 RTOG 87-04/ECOG 1289.

Radiation Therapy: 1.8 Gy daily 5 times per week for 5 weeks.

Reduced field at 30.60 Gy
For + inguinal nodes and/or local residual disease after 45 Gy proceed to 50.40 Gy with boost fields.

Chemotherapy: ARM 1: 5-FU 1000 mg/m/24 hrs. continuous IV for 96 hours, Day 1 and Day 2B of RT
Cycles 1 &2 ARM 2: 5-FU as above +Milo C: 10 mg/ma IV bolus on day 1 of each S-FU course (max 20 mgs per cycle)

Surgery: a) Biopsy of palpable inguinal lymph nodes prior to treatment.

b) Full-thickness biopsy to tbe performed 4-6 weeks after completion of radiation and chemotherapy.
c) Repeat full-thickness biopsy for patients reactedwith salvage combined therapy 21- 28 days after completion
of salvage therapy.

Salvage Therapy: Chemo (Cycle 3): 5-FU as above. Starting within 2 weeks of biopsy. CDDP 100 mg/m' IV over 6 hrs. on day
2 of RT (if creatinine < 50 substitute mito C).
Radiation: 9.00 Gy boost RTwith electrontphoton beam on perineum/inguinal nodes/both.

Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
MITOMYCIN IN ANAL CANCER 2529
tration of two doses of MMC at 10 mg/m2 during week 1 and week entire inguinal canals (Fig 2B). Part or all of the shaded inguinal
5 of CR. Four to 6 weeks following completion of CR, patients region would then be boosted with an electron beam or bolused
were subjected to a full-thickness biopsy of the primary tumor site photon beam, depending on dose delivery at 3 cm at various points
(postinduction biopsy). If the biopsy was negative, no further treat- in the shaded zones. The inferior border in all cases included the
ment was offered. However, if the biopsy was positive, patients were anal sphincter and the entire perineum.
then treated by a salvage regimen that consisted of an additional RT Dose. At the rate of 1.8 Gy/d, five times per week, the pelvis
boost of five fractions of 1.8 Gy to the area of residual tumor, received 45 Gy in 5 weeks. Fields were reduced at 30.6-Gy and 36-
concomitant with 5-FU infusion and cisplatin. Six weeks following Gy levels. Subsequent RT up to 45 Gy was delivered through a
completion of the salvage treatment, repeat full-thickness biopsies direct perineal field or through reduced 10- x 10-cm AP-PA fields.
of the primary tumor site were obtained (postsalvage biopsy). If these If the primary tumor was still palpable at this point, an additional
biopsies were negative, no further treatment was offered. However, if 5.4 Gy would be delivered in three fractions through either technique
the biopsy was positive, patients were subjected to abdominoperineal described.
resection. With NO disease, the dose in the medial inguinal nodes was 45
The RTOG adopted the 1978 International Union Against Cancer Gy at 3-cm depth. With Nl disease, both inguinal chains would be
25' 26
UICC staging system in this study so that it would have compara- boosted with an electron-beam field or a bolused photon field, so
tive data with trials it had initiated as far back as 1982. Table 1 that a dose of 50.4 Gy was delivered at a depth of 3 cm. The
provides a comparison with the current American Joint Committee distribution of RT doses to the primary tumor site is listed in Ta-
27
on Cancer (AJCC)/UICC staging system. Patients who presented ble 2.
with distant metastases (Ml) were ineligible. Only data for primary Treatment variations. Treatment interruption up to 10 days was
tumor sizes less than 5 cm or 2 5 cm were collected; thus, it was allowed if severe local skin reaction developed. There was a major
not possible to restage the patient population according to the AJCC/ compliance problem with RT fields early in the study. The protocol
UICC staging system. called for three fractions to be given to a second, reduced volume
at the 30.6-Gy level. However, instead of reducing the field to a
Treatment Schedule much smaller volume for the final boost, some patients received five
additional fractions to the second volume. This deviation occurred
RT. All patients were treated with megavoltage therapy units.
in 30% of patients assigned to 5-FU alone and in 40% of patients
An electron or photon beam was used to deliver a final boost to the
inguinal nodes and/or perineum at the discretion of the investigator. assigned to 5-FU and MMC.
Salvage treatment. On histologic confirmation of residual pri-
Treatmentfields. RT treatments were delivered through anterior-
mary disease after 45.6 Gy (or 50.4 Gy to the perineum in some
posterior (AP-PA) fields (Fig 2A). Patients were preferably treated
cases as described earlier), an additional 9 Gy would be delivered
in a prone position with a full bladder to reduce bowel toxicity.
through a reduced 10- X 10-cm pelvic field or a direct perineal field
Patients were treated in a supine position when inguinal nodes were
involved. Lower corner blocks were not used. with the technique described earlier.
With palpable residual inguinal adenopathy at the 45-Gy or 50.4-
The superior borders of the AP-PA fields extended to the IA-L5
Gy level, an additional 9 Gy would be directed to the area of residual
interspace for the initial 30.6 Gy; they would drop to the bottom of
disease. Histologic confirmation was not mandated here due to in-
the sacroiliac joints for an additional 5.4 Gy. With NO disease, the
consistencies in the past RTOG experience with such biopsies.
lateral borders of the fields extended to 1 cm lateral to the bony
pelvis to include the medial inguinal nodes. With Nl disease, the
lateral borders of the anterior field would be flared to include the Chemotherapy
Chemotherapy was delivered concomitantly with radiation. If se-
Table 1. Comparison of RTOG and AJCC/UICC Staging Systems vere local skin reactions developed, both RT and chemotherapy were
interrupted.
Tumor
Stage RTOG AJCC/UICC During the initial and definitive phase of CR, two cycles of 5-FU
(arm 1) or two cycles of 5-FU plus MMC were initiated on days 1
T1 < 1/3 circumference or length 5 2 cm greatest dimension 2
and 29. The first cycle of 5-FU was administered at 1,000 mg/m /d
of anal canal + no as a continuous infusion for 96 hours starting on day 1. The daily
infiltration of external
dose of 5-FU did not exceed 2,000 mg. 5-FU was repeated starting
sphincter muscle on day 29.
T2 > /3 circumference or length > 2 cm but 5 5 cm greatest MMC was administered at 10 mg/m2 by intravenous bolus on the
of anal canal or infiltration dimension first day of both 5-FU infusions in arm 2 patients only. The total
of external sphincter muscle dose was limited to 20 mg during each cycle. If the nadir WBC
T3 Extension to rectum or skin > 5 cm greatest dimension count was less than 2,400/gtL but more than 1,000/pL or nadir plate-
T4 Extension to adjacent Extension to adjacent let count more than 50,000/IL and less than 85,000/gL, the second
structures* structures' 2
dose of MMC was reduced to 7.5 mg/m . If the nadir WBC count
NO No regional lymph node No regional lymph node was less than 1,000/pL or the nadir platelet count less than 50,000/
involvement involvement 2
/L, the second dose of MMC was reduced to 5 mg/m . If on day
N1 Any regional lymph node Metastases to perirectal 28, the WBC count was less than 2,400/yL and/or the platelet count
involvement lymph nodest less than 85,000/gL, cycle 2 of both chemotherapy and RT was
*Bladder, urethra, or vagina. delayed 1 week.
tN2, unilateral iliac and/or inguinal lymph node metastases. N3, bilat- In the majority of patients in either treatment arm, the number of
eral iliac and/or inguinal lymph node metastases. cycles of 5-FU and MMC were correctly delivered. Two patients

Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
2530 FLAM ET AL

Fig 2. RT fields RTOG-87-04. (A) Two-field (AP-PA) technique. Superior border dropped to bottom of scaroiliac joints at 30.6 Gy and lower
corner blocks were not used. At 36.0 Gy, 10- x 10-cm fields with unchanged inferior borders were used. (B) When inguinal nodes were
involved, anterior field flared to include both inguinal regions. Shaded area was boosted by electrons or photons to bring nodal dose at 3-cm
depth to 50.4 Gy.

assigned to receive 5-FU alone did receive MMC. Table 3 lists the than 4 cm and had left an area of thickening that was more than
number of cycles of chemotherapy given in each treatment arm. 25% of the anal circumference, a biopsy of the central portion of
Salvage chemotherapy consisted of the same dose and schedule the thickening that would include the deepest area of the scar was
2
of 5-FU infusion. Cisplatin was administered at 100 mg/m over 4 acceptable and advised to avoid complications of biopsy, including
to 6 hours on day 2 of the 5-FU infusion. If the creatinine clearance rectovaginal fistula, long-standing fissures, or incontinence. The re-
was less than 50 mL/min at the time of initiation of salvage therapy, sponse of initially metastatic lymph nodes was gauged by palpation
2
MMC at 10 mg/m by intravenous bolus was used in lieu of cisplatin by the investigating physician. If palpation proved equivocal, the
if the WBC count was more than 2,400/tpL and the platelet count investigator could use a fine-needle biopsy at his/her discretion.
more than 85,000/pL.
Study End Points
PostinductionBiopsy The end points for the study were incidence of negative postinduc-
The primary tumor response was determined by a full-thickness tion biopsy, incidence of positive salvage biopsy, local regional con-
biopsy 4 to 6 weeks following completion of CR (postinduction trol, time to colostomy, colostomy-free survival, disease-free sur-
biopsy). Biopsy was delayed until the WBC count was more than vival, overall survival, and toxicity rates. Patients with a positive
2,400/tL and the platelet count more than 100,000/pL. If patients postinduction biopsy were considered to have local failure on the day
were found to have residual tumor and were treated with the salvage of the biopsy and patients with a negative biopsy were considered to
CR program, an additional full-thickness postsalvage biopsy was have local failure on the day a local recurrence was first reported.
required 4 to 6 weeks following completion of salvage therapy. With
anterior lesions in women and in lesions that were initially greater
Table 3. Actual No. of Chemotherapy Courses Delivered
No. of Courses
Table 2. Distribution of RT Doses to Primary Tumor During
2
Induction Therapy No.
Arm Drug Assessable 0 1 No. %
41.76- 47.26- 52.26-
Total < 41.75 Gy 47.25 Gy 52.25 Gy 57.25 Gy RT + 5-FU 5-FU 144 0 8 136 94
RT +5-FU 145 7 95 42 1 RT + 5-FU/MMC 5-FU 146 0 8 138 95
RT + 5-FU MMC 144 142 2 0 0
RT+ 5-FU/MMC 146 6 101 39 0
RT + 5-FU/MMC MMC 146 0 9 137 94
Combined 291 13 196 81 1
NOTE. No information on 1 patient in the RT + 5-FU group.

Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
MITOMYCIN IN ANAL CANCER
Time to local failure was measured from postinduction biopsy to
the occurrence of the first failure or to the date of the last follow-
up evaluation. Patients who had a colostomy, abdominoperineal re-
section, or exenteration for any reason were considered treatment
failures on the day of surgery. Time to colostomy was measured
from the day of treatment initiation to date of such surgery or last
follow-up evaluation if there was no surgery. Colostomy-free sur-
vival was measured from the day of treatment initiation to colostomy,
death, or last follow-up evaluation if the patient was alive with no
surgery. Disease-free survival was measured from study entry to the
occurrence of locoregional failure as stated earlier, distant metasta-
ses, a second primary tumor, or death from any cause. If none of
these events occurred, disease-free survival was measured to the
date of last follow-up evaluation. Overall survival was measured
from study entry until death from any cause or last follow-up evalua-
tion.
Toxicity
Chemotherapy toxicity was graded according to National Cancer
Institute (NCI) toxicity criteria and RT toxicity was graded according
to the RTOG toxicity criteria.2" A toxicity was considered acute if
it occurred within 90 days from the start of treatment. If a toxicity
occurred after 90 days, it was considered late.
Statistical Methods
Patient and treatment characteristics were compared by the X2 test
for discrete data and by the Wilcoxon test for continuous data.2 9
All statistical comparisons were made with two-tailed tests. The
cumulative-incidence approach was used to estimate time to local
failure and time to colostomy because it specifically adjusts for
the other competing risks of failure.30 For example, the cumulative
incidence for local failure adjusts for patients who die without it.
Colostomy-free survival, disease-free survival, and overall survival
were estimated according to the Kaplan-Meier method.3 1
Differences in the initial biopsy rates between the two treatment
arms were tested by the logistic model.32 Differences in the local
failure rates, time to colostomy rates, colostomy-free survival rates,
disease-free survival, and overall survival between the two treatment
arms were tested by the Cox proportional model.33 The three stratifi-
cation variables (nodal status, histology, and primary tumor size)
were considered as fixed covariates and the impact of adding the
treatment assignment variable to each was evaluated. The central
pathology diagnosis was used for all multivariate analyses. Patients
with keratinizing squamous cell tumors were compared against all
others, except for nine patients without central pathology review.
The derived significance value associated with the treatment variable
was tested for overall treatment effect with the addition of MMC.
As histology proved to be nonsignificant, it was dropped from all
models for the reported significance values. The difference in time
to colostomy rates among RTOG tumor-stages was tested by Gray's
test.3 4
Quality Assurance
Central pathology review of all biopsy material was undertaken.
Tumors were classified as keratinizing squamous cell carcinomas or
nonkeratinizing squamous cell carcinomas.' 3 Additional categories
were recorded in a small percentage of patients.
A central review of the RT and the chemotherapy given was
performed. Individual treatment parameters, such as total dose, field
borders, and elapsed treatment days, were reviewed and scored rela-
Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083
Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
2531
tive to the protocol prescription. For drug delivery, the chemotherapy
flow sheets were reviewed and drug dosing and delivery was scored
relative to the protocol prescription. For each treatment modality,
the delivery was scored as per protocol, minor variations acceptable,
major variations acceptable, and major variations unacceptable. The
institutional audit program mandated by the NCI was in place during
the entire study. Every RTOG and ECOG institution is site-visited
at least every 3 years. Randomly selected records were compared
with the original records at each institution. There was a formal
Data Monitoring Committee in place to oversee the trial's progress.
Two interim analyses were planned and the final analysis will be
performed after every patient has been monitored for 2 years. The
study protocol was approved by the National Institutes of Health
and the review boards of the RTOG and all of the participating
institutions.
Review of RT was completed for all but five of 291 analyzable
patients. Compliance to RT prescription was assessed by the percent-
age of patients scored as per protocol or with minor deviations. The
compliance rate was 87% for patients assigned to the RT plus 5-FU
arm and 90% for patients assigned to the RT/5-FU/MMC arm.
A review of drug records by the study chair was completed for
all but seven of 291 analyzable patients. Compliance to the chemo-
therapy was assessed by the percentage of patients scored as per
protocol or with minor deviations. Compliance was 95% for patients
assigned to the RT plus 5-FU arm and 90% for patients assigned to
the RT/5-FU/MMC arm.
RESULTS
Between August 1988 and December 1991, 310 pa-
tients were randomized. Nineteen patients were excluded
from all subsequent analysis. The three most common
reasons for exclusion were inadequate data (n = 7), no
measurable disease (n = 4), and metastatic disease (n =
4). Of the remaining 291 assessable patients, 145 were
assigned to RT and 5-FU and 146 to RT/5-FU/MMC.
The analysis was performed on all of the data received
at the RTOG headquarters and processed as of October
24, 1994. Of 291 cases available for all analyses, 217
patients (75%) were last reported alive. Since the protocol
closed in December 1991, every patient has been moni-
tored for at least 2 years; however, the 2-year information
is incomplete for 24 patients (8.2%). The median follow-
up duration for all analyzable patients is 3.01 years, with
a range of 0.071 to 6.17 years. If only living patients are
considered, the median follow-up time is 3.52 years, with
a range of 0.17 to 6.17 years.
Pretreatment characteristics of patients randomized to
both groups are listed in Table 4. There was no significant
imbalance for age, sex, performance status, distribution
of histologic types, and tumor size between the two treat-
ment arms. No significant imbalance of RTOG tumor or
nodal stage was observed.
Biopsy Results
A total of 262 of 291 randomized patients were sub-
jected to postinduction biopsy. Of 29 patients who did
2532 FLAM ET AL

Table 4. Pretreatment Characteristics Table 6. Biopsy Results by Stratifying Variables

RT+5-FU + Positive Negative
RT+5-FU MMC Total No. Not -
Variable No. Done No. % No. %
(N = 145) (N = 146)
Characteristic No. % No. % Size (cm)
< 5cm 117 16 11 7 150 93
Age, years
< 60 75 52 62 42
5 cm 114 13 17 17 84 83 P.02
2 60 70 48 84 58
Nodal stage
Median 59 62.5
NO 240 24 22 10 194 90
Range 26-86 29-85
N1 51 5 6 13 40 87
Sex
Histology'
Male 57 39 44 30
Keratinizing 75 4 9 13 62 87
Female 88 61 102 70
Nonkeratinizing 207 24 17 9 166 91
KPS
100-90 112 77 118 81 *All except 9 cases underwent central pathology review.
80 28 19 18 12
70 4 3 7 5
60 1 1 3 2 difference was not statistically significant (P = .135 by
Histology' logistic model).
Keratinizing squamous 45 31 30 20
The only stratifying variable that significantly influ-
Nonkeratinizing
Basaloid squamous 91 63 111 76
enced the presence of residual disease was primary tumor
Basaloid small cell 1 1 0 0 size (Table 6). Patients with primary tumors less than 5
Other 3 2 1 1 cm achieved a 93% negative biopsy rate, as compared
Unknown 5 3 4 3 with 83% for patients with bigger primary tumors (P =
Tumor size (cm)
.02). The favorable impact of MMC on negative biopsy
<5 88 61 89 61
5 57 39 57 39
rates was seen in patients irrespective of the size of their
Tumor stage primary tumors (Table 7).
T1 22 15 22 15
T2 50 35 61 42 Colostomy Rates
T3 61 42 48 33
Of 145 patients assigned to receive RT plus 5-FU, 32
T4 12 8 15 10
Nodal stage
underwent a colostomy, abdominoperineal resection, or
NO 119 82 121 83 more extensive surgical procedure. Of 146 patients as-
N1 25 17 25 17 signed to receive MMC in addition, 13 underwent similar
NX 1 1 0 0 surgical procedures. All of these surgical interventions
Abbreviation: KPS, Karnofsky performance status. will be designated as colostomies in subsequent discus-
*Central review. sion and tables. If a colostomy was performed within the
first 7 months of CR, it was considered immediate, since
not undergo a posttreatment biopsy, 12 of 145 (8.3%) these were associated with both a positive postinduction
were treated with 5-FU and 17 of 146 (11.6%) with 5-FU biopsy and a positive salvage biopsy. For the 5-FU-alone
plus MMC. Eighty-one percent of the biopsies performed arm, 8.2% of patients underwent immediate colostomy
were full-thickness biopsies. The negative biopsy rate for as compared with 3.4% for the 5-FU/MMC arm. The time
the 5-FU arm was 86%. The negative biopsy rate for the to colostomy is shown in Fig 3 by assigned treatment
5-FU plus MMC arm was 92.2% (Table 5). This 6.2% arm. At 4 years, 9% of the 5-FU plus MMC group versus

Table 5. Posttreatment Biopsy Results

Table 7. Biopsy Results by Size of Primary Tumor and Treatment
Positive Negative
Total No. Not
Arm No. Done No. % No. % P Positive Negative
Size Total No. % No. %
5-FU 145 12' 18 14 115 86
.135 < 5 cm
5-FU + MMC 146 17t 10 8 119 92 5-FU 81 8 10 73 90
5-FU + MMC 80 3 4 77 96
*Reason biopsy not done: 2, death; 2, toxicity; 1, refusal; 6, doctor
5 cm
preference; 1, other.
5-FU 52 10 19 42 81
tReason biopsy not done: 4, death; 2, toxicity; 4, refusal; 7, doctor
5-FU + MMC 49 7 14 42 86
preference.

Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
MITOMYCIN IN ANAL CANCER 2533

100 - ASSIGNED TREATMENT COLOSTOMY/TOTAL

RT + 5FU (32/145)
RT + 5FU/MITO (13/146) P= .002 (GRAY'S TEST)

75 -

o
50 -
Fig 3. Time to colostomy-
RTOG 87-04/ECOG 1289. 0

q
25- I

-- --- - - - - - - - - - - - - - - ---

n-
U -I
I

0 1 2 3 4
YEARS FROM START OF TREATMENT

23% of the 5-FU group underwent colostomies (P = .002,
Cox model) (Table 8). Of 13 patients in the 5-FU plus
MMC group who underwent colostomies, 11 had residual
disease; only two patients had the surgical procedure for
treatment complications. Of 32 patients in the 5-FU arm
who underwent colostomies, 29 had residual disease, two
had treatment complications in addition to residual dis-
ease, and one underwent colostomy for unknown reasons.
After 2 years, the number of patients who had a colostomy
was extremely low, including three of 90 patients at risk
in the 5-FU-alone group and two of 106 patients at risk
in the 5-FU/MMC group. Thus, the 2-year colostomy
rate probably represents a relatively accurate long-term
projection of the final colostomy rates.
The impact of MMC on colostomy rate reduction was
most significant in RTOG T3/T4 primary tumors (P =
.019) (Fig 4) and did not achieve statistical significance
for RTOG T1/T2 primary tumors (P = .141) (Fig 5). In
patients assigned to the MMC arm, there was no differ-
ence found in the colostomy rates between the smaller
and the larger primary tumors.
Table 8. Frequency of Patients With Colostomy by Treatment Program
% No. Colostomy'/ % No. Colostomyt/
Treatment Colostomy' Total Colostomyt Total
RT + 5-FU 8.2 12/145 22 32/145
RT + 5-FU +
MMC 3.4 5/146 9 13/146
NOTE. Colostomy designates colostomy, AP resection, and more exten-
sive resection.
*Within 7 months of the start of treatment.
tActuarial estimate at 4 years.
The colostomy rate for patients with RTOG stage NO
was 13.3%, as compared with 28% for patients with posi-
tive nodes (RTOG Nl). Thus, N1 patients are at greater
risk of colostomy failure than NO patients. Among pa-
tients with RTOG stage NO disease, nine of 121 patients
who received 5-FU/MMC, compared with 23 of 119 pa-
tients who received 5-FU, underwent colostomy (P =
.009, Gray's test). Among patients with RTOG stage N1
disease, five of 25 patients who received 5-FU/MMC,
compared with nine of 25 patients who received 5-FU,
underwent colostomy. Thus, a benefit of MMC in the
reduction of colostomy rates was observed in NO patients.
While a trend in favor of MMC in N1 patients was ob-
served, the sample size of this subgroup is too small to
achieve statistical significance.
The treatment difference was significant when colos-
tomy-free survival was evaluated (Fig 6). At 4 years, an
estimated 71% of patients assigned to receive MMC were
alive without a colostomy, as compared with 59% of
patients not assigned to MMC (P = .014, Cox model).
Disease-Free Survival
The two treatment groups were evaluated with respect
to disease-free survival, where failure was defined as the
first disease progression. Patients were censored from
analysis if and when death occurred without disease pro-
gression. The disease-free survival rate was significantly
better for MMC patients (P = .0003, Cox model). At 4
years, an estimated 73% of MMC patients were alive
without disease, as compared with 51% of the group that
received only 5-FU (Fig 7).

Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
2534 FLAM ET AL

100 - ASSIGNED TREATMENT COLOSTOMY/TOTAL

RT + 5FU (21/73)
_-- RT + 5FU/MITO (8/63) P = .019(GRAY'S TEST)

75-

IP

50-
0 Fig 4. Time to colostomy for
RTOG T3 and T4-RTOG 87-04/
u ECOG 1289.

-
J -

25-

-___ _ _ _ ____ -I
- - - - - - - - - -- - - - - -
0 ___
0 1 2 3 4
YEARS FROM START OF TREATMENT

Overall Survival Toxicity

The overall survival is demonstrated in Fig 8. At this
analysis, there was no statistically significant difference
(P = .31, Cox model), although there was a trend in favor
of the MMC arm that emerged after 18 months. The
distribution of the causes of death by arm is listed in
Table 9. There were fewer reported deaths due to tumor
in the MMC arm.
Table 10 lists the toxicities observed in each treatment
arm. Nonhematologic toxicities of the gastrointestinal
tract, skin, and mucous membranes were similar in both
arms. A significantly greater degree of neutropenia and
associated infection was identified in the MMC arm. A
greater degree of thrombocytopenia was also identified
in the MMC arm, but no significant bleeding complica-

100 ED TREATMENT COLOSTOMY/TOTAL

RT + 5FU (11/72)
-- - - RT + 5FU/MITO (6/83) P = .141(GRAY'S TEST)

75

O
Io

50
Fig 5. Time to colostomy for
RTOG TI and T2-RTOG 87-04/ t
ECOG 1289.

25

0 1 2 3 4
YEARS FROM START OF TREATMENT

Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
MITOMYCIN IN ANAL CANCER 2535
100

-'-- 4

, __ _.- 1*% - N4 - * -4- *

75

III

.J
1 50
4 FAILURES/TOTAL
Fig 6. Colostomy free sur- RT + 5FU (56/145)
vival-RTOG 87-04/ECOG 1289. RT + 5FU/MITO (37/146) P = .014 (COX MODEL)

25

115 93 71
125 108 I
0
0 1 2 3 4
YEARS FROM RANDOMIZATION

tions were observed. The distribution of late toxicities
was not significantly different in the two treatment arms.
Overall, 7% (10 of 145) of patients randomized to the 5-
FU arm experienced grade IV toxicity and one (0.7%)
grade V (fatal) toxicity. In contrast, 23% (34 of 146) of
the patients randomized to 5-FU plus MMC experienced
grade IV toxicity and 2.7% (four patients) experienced
fatal toxicities (P - .001). All of the patients who experi-
enced toxic deaths died of neutropenic sepsis. Two toxic
deaths in the MMC group were related to a lack of dose
reduction per protocol of the second cycle of MMC.
Results of Salvage Treatment
Of 28 patients with positive biopsies, three did not
receive salvage treatment because of patient refusal (n =
1) or clinician's preference (n = 2) (Fig 9). Of 25 patients
who received salvage treatment, 15 were treated initially
with 5-FU alone and 10 with 5-FU plus MMC. All 25
patients received 5-FU and cisplatin with RT as salvage
treatment. Of 25 analyzable patients, 22 were subjected
to postsalvage biopsies and 12 of the 22 or 55% demon-
strated negative biopsies. No striking difference was ob-
served in the outcome of salvage treatment by initial treat-

100

/A--//f# - - ''. , ,
75 . w-, ens##
.M

Lu
4> 50
Fig 7. Disease-free sur-
FAILED/TOTAL vival-RTOG 87-04/ECOG 1289.
RT + 5FU (62/133)
RT + 5FUIMITO (31/129) P = .0003 (COX MODEL)
25

0
0 1 2 3 4
YEARS FROM RANDOMIZATION

Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
2536 FLAM ET AL

100-

-+~ ~'~
rmNIIY- M*,IftH 1 M

75.

50- DEAD/TOTAL
RT + 5FU 42/145) Fig 8. Overall survival-
RT + 5FU/MITO (32/146) P = .31 (COX MODEL) RTOG 87-04/ECOG 1289.

25

NUMBER AT RISK
145 135 103 71 32
146 134 114 89 40
n. I
I

0 1 2 3 4
YEARS FROM RANDOMIZATION

ment arm. Toxicity observed with the salvage regimen treatment of epidermoid malignancies of the anal canal.
was limited and no deaths from salvage treatment were Table 11 lists the 4-year efficacy end points for this study.
observed. Initial posttreatment biopsies were positive in 15% of
Follow-up evaluation of the 12 patients who underwent patients who received RT plus 5-FU and in 7.7% of pa-
successful salvage CR (negative biopsy postsalvage) tients who received RT/5-FU/MMC. The improvement
shows that four remain free of disease at 4 years, four in negative biopsy rates occurs mainly in patients with
had a subsequent abdominoperineal resection (and are primary tumors > 5 cm. This difference was achieved
free of disease), and four have died (two with recurrent with a significantly higher incidence of acute hematologic
disease, one of unknown cause with recurrent disease, toxicity in the MMC group.
and one of unrelated cause free of disease). In the patients Although MMC-treated patients did not experience a
who failed to respond to salvage treatment (positive post- statistically significant increase in overall survival, a sta-
salvage biopsy), nine of 10 underwent abdominoperineal tistically significant reduction in colostomy rates and an
resection. Seven patients have died (six of progressive increase in colostomy-free survival and disease-free sur-
disease and one of unrelated causes) and three remain vival was observed at 4 years, thereby justifying the inclu-
free of disease with colostomy. In summary, additional sion of MMC in this definitive CR regimen. The major
CR was able to salvage, without colostomy, one third benefit of MMC in reducing colostomy rates and improv-
(seven of 22) of assessable patients who failed to respond ing colostomy-free survival occurred in patients with
to initial standard CR. Overall, 50% (11 of 22) of these
patients were alive without disease at 4 years.
Table 10. Observed Grade 4 and 5 Toxicities
DISCUSSION
%Toxicity per Treatment
This randomized phase III study attempts to define the Type of Toxicity 5-FU 5-FU + MMC P
role of MMC and salvage CR in the definitive nonsurgical
Acute*-total 7 20 < .001
Acute hematologic 3 18 < .001
Table 9. Causes of Death by Treatment Arm Acute nonhematologic 4 7 .63
Latet 1 5 .26
5-FU 5-FU +MMC
Maximum (acute or late)
Total dead 42 32 Grade 4 7 (10/145) 23 (34/146)
Progressive disease 24 15 < .001
Treatment complications 1 4 Grade 5 0.7 (1/145) 3 (4/146)
Unrelated/other 13 9
*Acute toxicity, occurs within 90 days from start of treatment.
Unknown 4 4
tLate toxicity, occurs after 90 days.

Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
MITOMYCIN IN ANAL CANCER 2537

Fig 9. Outcome of salvage

CR-RTOG 87-04/ECOG 1289.

larger primary tumors (RTOG T3/T4) as compared with the early timing of postsalvage biopsies. The effective-
smaller primary tumors (RTOG T1/T2). ness of immediate salvage CR in patient who fail to
This study demonstrates a possible salvage CR benefit achieve complete remission with definitive CR may be
using additional CR consisting of cisplatin, 5-FU, and a difficult to ascertain, as a number of investigators do not
9.0-Gy pelvic radiation boost in patients with positive undertake elective biopsy of clinical residual abnormali-
biopsies following initial CR to a dose of 45 to 50.4 Gy. ties and do not initiate immediate CR, but monitor these
Twelve of 22 assessable patients with residual disease patients without treatment.3 7'3 8
following initial CR demonstrated negative postsalvage While additional follow-up evaluation will be required
therapy biopsies. A third of these 12 patients remained to determine any potential long-term survival impact of
disease free without colostomy at 4 years. MMC and nonsurgical salvage therapy, and to ascertain
Salvage abdominoperineal resection has proven to be the incidence of late toxicities, the following conclusions
ineffective in the majority patients with residual tumor regarding the future treatment of these patients can be
following CR.8 36' In our patients with residual disease drawn.
following salvage CR, surgery resulted in one third (three (1) MMC in combination with 5-FU and RT favorably
of nine) of these patients being disease-free at 4 years. affects postinduction biopsy rates, colostomy rates, colos-
Thus, it appears that if both MMC and salvage CR are tomy-free survival, and disease-free survival in all pa-
used, less than 10% of patients require abdominoperineal tients with epidermoid malignancies of the anal canal,
resection and colostomy at 4 years. Nevertheless, the is- regardless of the histologic subtype, nodal status, and size
sue of the efficacy of nonsurgical salvage therapy remains of the primary tumor, and should be used in all patients,
uncertain due to the relatively small number of cases and except possibly those with very small primary tumors. 39
(2) Based on the significant incidence of neutropenia
Table 11. Summary of Efficacy End Points: RTOG 8704/ECOG 1289
and the significant grade 4 and 5 toxicities encountered
in the MMC arm, the MMC should not be used routinely
Estimated
4-year
rates(%) P in patients who are immunosuppressed or who are human
End Point 5-FU 5-FU + MMC Univariate Multivariate immunodeficiency virus (HIV)-positive, due to the inabil-
Positive postinduction biopsy 15 8 .098 .135
ity of these patients to tolerate neutropenia.
Local failure rate 34 16 .0007 .0008 (3) Patients with positive inguinal or pelvic nodes
Colostomy rate 22 9 .0025 .002 should not be excluded from future trials, as both node-
Colostomy-free survival 59 71 .019 .014 positive and node-negative patients appear to benefit sig-
Disease-free survival 51 73 .0002 .0003 nificantly from nonsurgical definitive CR with MMC in
Overall survival 67 76 .18 .31
combination with 5-FU and RT.

Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
2538
(4) For patients treated with definitive nonsurgical CR
who demonstrate residual disease on early posttreatment
biopsies at 4 to 6 weeks after CR, abdominoperineal re-
section should not be immediately undertaken. This sub-
group of approximately 10% to 15% of patients should
either be treated with early salvage CR using 5-FU infu-
sion, cisplatin, and pelvic radiation boost, or should be
monitored closely without treatment for an additional 4
to 6 weeks to see if they achieve a complete response
REFERENCES
1. Stearns MW. Jr, Urmacher C, Sternberg SS. et al: Cancer of
the anal canal. Curr Probl Cancer 4:1-44, 1980
2. Flam MS, John MD: Anal canal, in John MJ. Flam MS, Legha
SS, et al (eds): Chemoradiation: An Integrated Approach to Cancer
Treatment. Philadelphia, PA, Lea & Febiger, 1993, pp 337-346
3. Beahrs OH: Management of cancer of the anus. Am J Roent-
genol 133:790-795, 1979
4. Quan SHQ: Epidermoid carcinoma of the ano-rectum. NY
State J Med 77:2056-2057, 1977
5. Nigro ND, Vaitkevicius VK, Considine B Jr: Combined ther-
apy for cancer of the anal canal: A preliminary report. Dis Colon
Rectum 17:354-356, 1974
6. Flam MS, John MJ, Lovalvo LJ, et al: Definitive nonsurgical
therapy of epithelial malignancies of the anal canal: A report of 12
cases. Cancer 51:1378-1387, 1983
7. Cummings B, Keane T, Gillian T, et al: Results and toxicity
of the treatment of anal canal carcinoma by radiation therapy or
radiation therapy and chemotherapy. Cancer 54:2062-2068, 1984
8. Leichman L, Nigro N, Vaitkevicius VK, et al: Cancer of the
anal canal: Model for preoperative adjuvant combined modality
treatment. Am J Med 78:211-216, 1985
9. Michaelson RA, Magill GB, Ivan SH, et al: Preoperative che-
motherapy and radiation therapy in the management of anal epider-
moid cancer. Cancer 51:390-395, 1983
10. Sischy B: The use of radiation therapy combined with chemo-
therapy in the management of squamous cell carcinoma of the anus
and marginally resectable adenocarcinoma of the rectum. Int J Radiat
Oncol Biol Phys 11:1587-1593, 1985
11. Flam MS, John MJ, Mowry PA, et al: Definitive combined
modality therapy of carcinoma of the anus: A report of 30 cases
including results of salvage therapy in patients with residual disease.
Dis Colon Rectum 30:495-502, 1987
12. Secco GB, Sertoli MR, Scarpati D, et al: Preoperative chemo-
therapy and radiotherapy in the management of epidermoid carci-
noma of the anal canal. Tumori 73:151-155, 1987
13. Meeker WR Jr, Sickie-Santanello BJ, Philpotti G, et al: Com-
bined chemotherapy, radiation and surgery for epithelial cancer of
the anal canal. Cancer 57:525-529, 1986
14. Pipard G, Peytremann R, Marti MC: Conservative multidiscipli-
nary treatment of locally advanced epidermoid and cloacogenic cancer
of the anal canal. Proc Am Soc Clin Oncol 5:268, 1986 (abstr)
15. Tiver KW, Langlands O: Synchronous chemotherapy and ra-
diotherapy for carcinoma of the anal canal: Alternative to abdomino-
perineal resection. Aust NZ J Surg 54:101-108, 1984
16. Sischy B, Doggett RLS, Krall JM, et al: Definitive irradiation
and chemotherapy for radiosensitization in management of anal car-
cinoma: Interim report on RTOG study no. 8314. J Natl Cancer Inst
81:850-856, 1989
Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083
Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
FLAM ET AL
with additional follow-up time. Only those patients with
persistent and progressive local disease following salvage
CR on longer follow-up evaluation of at least 12 weeks
and those who recur locally following successful CR
should be subjected to abdominoperineal resection.
ACKNOWLEDGMENT
We thank Nancy Connor for data management and Elizabeth Mar-
tin, RTT, for dosimetry.
17. Schlienger M, Krzisch C, Pene F, et al: Epidermoid carcinoma
of the anal canal: Treatment results and prognostic variables in a
series of 242 cases. Int J Radiat Oncol Biol Phys 17:1141-1151,
1989
18. Habr-Gama A, DaSilva E, Sousa A, et al: Epidermoid carci-
noma of the anal canal. Results of treatment by combined chemother-
apy and radiation therapy. Dis Colon Rectum 32:773-777, 1989
19. Knecht BH: Combined chemotherapy and radiotherapy for
carcinomas of the anus. Am J Surg 159:518-521, 1990
20. Cho CC, Taylor CW II, Padmanabhan A, et al: Squamous-
cell carcinoma of the anal canal: Management with combined che-
moradiation therapy. Dis Colon Rectum 34:675-678, 1991
21. Zelnick RS, Haas PA, Ajlouni M, et al: Results of abdomino-
perineal resections for failures after combination chemotherapy and
radiation therapy for anal canal cancers. Dis Colon Rectum 35:574-
578, 1992
22. Byfield JE, Barone RM, Sharp TR, et al: Conservative man-
agement without alkylating agents of squamous cell anal cancer
using cyclical 5-FU alone and x-ray therapy. Cancer Treat Rep
67:709-712, 1985
23. Cummings BJ, Keane TJ, O'Sullivan B, et al: Epidermoid
anal cancer: Treatment by radiation alone or by radiation and 5-
fluorouracil with and without mitomycin-C. Int J Radiat Oncol Biol
Phys 21:1115-1125, 1991
24. Zelen M: The randomization and stratification of patients to
clinical trials. J Chron Dis 27:365-275, 1974
25. Harmer MH (ed): TNM Classification of Malignant Tumors
(ed 3). Geneva, Switzerland, International Union Against Cancer,
1978, pp 77-81
26. Sischy B, Krall J, Pazdur B, et al: Evaluation of RTOG 83-
4 phase II study of the treatment of squamous cell basaloid carcinoma
of the anal canal by radiation therapy and chemotherapy for sensiti-
zation followed by biopsy + surgery. Proc Am Soc Clin Oncol 7:93,
1988 (abstr)
27. Beahrs OH, Henson DE, Hutter RVP, et al (eds): Manual for
Staging of Cancer (ed 3). Philadelphia, PA, Lippincott, 1988, pp 81-83
28. John MJ: Grading of chemoradiation toxicity, in John MJ,
Flam MS, Legha SS. et al (eds): Chemoradiation: an Integrated
Approach to Cancer Treatment. Philadelphia, PA, Lea & Febiger,
1993, pp 601-606
29. Wilcoxon F: Individual comparison by ranking methods. Bio-
metrics 1:80-83, 1945
30. Kalbfleish JD, Prentice RL: The Statistical Analysis of Failure
Time Data. New York, NY, Wiley, 1980
31. Kaplan EL, Meier P: Nonparametric estimation from incom-
plete observations. J Am Stat Assoc 53:457-481, 1958
32. Cox DR: Analysis of Binary Data. London, United Kingdom,
Methuen, 1970
MITOMYCIN IN ANAL CANCER
33. Cox DR: Regression models and life tables. J R Stat Soc B
34:187-229, 1972
34. Gray RJ: A class of K-sample tests for comparing the cu-
mulative incidence of a competing risk. Ann Stat 16:1141-1154,
1988
35. Boman BM, Moertel CG, O'Connell MJ, et al: Carcinoma of
the anal canal. A clinical and pathological study of 188 cases. Cancer
54:114-125, 1984
36. Zelnick RS, Haas PA, Ajlouni M, et al: Results of abdomino-
perineal resections for failures after combination chemotherapy and
2539
radiation therapy for anal canal cancers. Dis Colon Rectum 35:564-
578, 1992
37. Cummings BJ: Anal cancer. Int J Radiat Oncol Biol Phys
19:1309-1315, 1990
38. Papillion J, Montbarbon JF: Epidermoid carcinoma of the
anal canal. A series of 276 cases. Dis Colon Rectum 30:324-333,
1987
39. Doggett SW, Green JP, Cantril ST: The efficacy of radiation
therapy alone for limited squamous cell carcinoma of the anal canal.
Int J Radiat Oncol Biol Phys 15:1069-1072, 1988

Downloaded from ascopubs.org by 80.82.77.83 on February 27, 2017 from 080.082.077.083

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.