THE AMERICAN JOURNAL OF GASTROENTEROLOGY
6, 2000
© 2000 by Am. Coll. of Gastroenterology
Published by Elsevier Science Inc.
Topographic Patterns of Intestinal
Metaplasia and Gastric Cancer
Mauro Cassaro, M.D., Massimo Rugge, M.D., Oscar Gutierrez, M.D., Gioacchino Leandro, M.D.,
David Y. Graham, M.D., and Robert M. Genta, M.D.
Department of Oncology and Surgical Sciences, University of Padova, Veneto, Italy; Department of
Pathology, Veterans Affairs Medical Center and Baylor College of Medicine, Houston, Texas; Department of
Medicine, Veterans Affairs Medical Center and Baylor College of Medicine, Houston, Texas; Department of
Gastroenterology, Hospital San Juan de Dios, Universidad Nacional, Bogotá, Colombia; Department of
Gastroenterology, Castellana Grotte Hospital, Bari, Italy; and Department of Molecular Virology, Baylor
College of Medicine, Houston, Texas
OBJECTIVE: The role of intestinal metaplasia in gastric on-
cogenesis has been demonstrated by both cross-sectional
and longitudinal studies. This study was designed to deter-
mine whether, in a population at high risk for gastric cancer,
different topographical patterns and phenotypes of intestinal
metaplasia were associated with different degrees of cancer
risk.
METHODS: A total of 68 Colombian patients with gastric
cancer and 67 controls with nonulcer dyspepsia were stud-
ied by an extensive biopsy protocol. Intestinal metaplasia
was assessed semiquantitatively by histology and was char-
acterized histochemically. In both patients and controls, the
Spearman’s correlation test was applied to the test if the
gastric distribution of metaplastic lesions resulted in specific
topographical patterns associated with different risks for
cancer.
RESULTS: Four topographical patterns of intestinalization
emerged: 1) “Focal,” in 14 cancer patients and 16 controls;
2) “Antrum-predominant,” in seven cancer patients and six
controls; 3) “Magenstraße” (involving the lesser curvature
from cardia to pylorus) in 25 cancer patients and four
controls. This pattern was associated with higher cancer risk
(OR ⫽ 5.7; 95% CI: 1.3–26) than were the two less exten-
sive patterns; and 4) “Diffuse,” involving essentially the
entire gastric mucosa with the exception of the fundus, was
unique to 13 cancer patients. The OR for cancer was 12.2;
95% CI: 2.0 –72.9. Incomplete-type metaplasia significantly
correlated with the extent of total metaplasia and was also
associated with greater cancer risk.
CONCLUSIONS: In a population with high risk for gastric
cancer, the extension of intestinal metaplasia correlates
with the extent of its “incomplete” phenotype and is
significantly associated with increased cancer risk. Both
the extent and location of intestinal metaplasia along the
lesser curvature (from the cardia to the prepyloric zones)
identify patients with the highest cancer risk. (Am J
Vol. 95, No.
ISSN 0002-9270/00/$20.00
PII S0002-9270(00)00866-2
Gastroenterol 2000;95:1431–1438. © 2000 by Am. Coll.
of Gastroenterology)
INTRODUCTION
Although its incidence has been declining, gastric cancer
remains the second highest cause of cancer deaths in the
world (1). Gastric adenocarcinoma is the end stage of a
multifactorial stepwise process in which environmental fac-
tors are believed to play a major causative role (2– 4).
Helicobacter pylori, (H. pylori) the most important cause of
chronic gastritis worldwide, triggers phenotypic and genetic
alterations of the gastric mucosa that have been linked both
epidemiologically and biologically to the development of
cancer (5–7). Based on these lines of evidence, the Interna-
tional Agency for Research on Cancer declared H. pylori as
a class I carcinogen in 1994 (8).
Glandular-type gastric adenocarcinoma develops after de-
cades of H. pylori infection. Therefore, a long period of
incubation must be needed to establish the phenotypical and
genotypical alterations that result in neoplasia (9). During
this latency period the gastric mucosa undergoes progres-
sive phenotypical changes, from chronic active gastritis to
gastric atrophy, intestinal metaplasia, and gastric dysplasia,
which have been defined collectively as gastric precancer-
ous lesions. Each of these morphological alterations is as-
sociated with an increased risk for gastric cancer (10 –13).
The Houston-updated Sydney system for the classifica-
tion of gastritis emphasized both the multifocal distribution
of atrophic gastritis associated with H. pylori infection
(hence the inclusion of the term “multifocal atrophic gas-
tritis”) and the frequent association of glandular loss with
intestinalization of the gastric mucosa (14). The Houston
consensus document also depicted the characteristic topo-
graphical pattern of both atrophic and metaplastic lesions,
reiterating the early location of the atrophic-metaplastic
changes at the incisura angularis and their progressive
spread throughout the antral and oxyntic mucosa. Conse-
1432 Cassaro et al. AJG – Vol. 95, No. 6, 2000

quently, routine biopsy sampling of the mucosa of the an-

gulus was recommended (14, 15).
Intestinal metaplasia (IM) has been categorized in differ-
ent subtypes according to histological and histochemical
characteristics. The most widely used classification was
originally proposed by Jass and Felipe (16). It is based on
the similarities of the metaplastic epithelium with the small
intestine (complete, or “small intestinal”) or with the colonic
epithelium (incomplete, or “colonic”), and the content of
sulfated mucins in the absorptive cells (none in the com-
plete, but abundant in the incomplete type). Data collected
both in cross-sectional descriptive investigations and fol-
low-up studies have suggested a stronger association of
gastric cancer with the incomplete types of IM than with the
complete type (17, 18).
A highly significant epidemiological association between
IM and glandular-type gastric cancer has been demonstrated
in populations in which this neoplasm is considered en-
demic. These populations also have a high prevalence of H.
pylori infection (19). Therefore, such epidemiological set-
tings are particularly suitable for studies addressing the
biological relationships between gastric precancerous le-
sions and carcinoma. Figure 1. Schematic representation of the sites from which muco-
sal biopsies were obtained.
Our study was designed to test the following hypotheses:
1) the extent and phenotypes of IM are related to each other; Biopsy Mapping Protocol
and 2) not only the extent, but also the topographical dis- A standardized biopsy mapping protocol was followed in
tribution of the IM throughout the gastric mucosa may result both groups of patients. Briefly, endoscopic gastric biopsy
in different cancer risk. specimens were obtained using a jumbo forceps from the 12
gastric sites depicted in Figure 1. Five specimens were from
MATERIALS AND METHODS the antrum, six from the corpus and one from the cardia. In
addition to the standard protocol, at least two biopsy sam-
This study, which was approved by the Human Research ples were obtained from the neoplastic lesions. In some
Committee of Baylor College of Medicine, Houston, TX, patients the size of the neoplasm or other factors prevented
and the Universidad Nacional de Colombia, Santa Fe de the endoscopist from obtaining the entire biopsy set. Only
Bogotá, Colombia, was part of a greater collaborative effort patients from whom at least nine biopsy specimens were
designed to address various aspects of the relationship be- available are included in this study.
tween H. pylori, atrophic gastritis, and gastric cancer in the
Colombian population. All participants were recruited at the Histopathology
Bogotá University Hospital and gave informed consent. Biopsy specimens were shipped to our laboratory fixed in
10% buffered formalin. They were processed, embedded in
Cancer Patients paraffin, and cut in sequential 4-␮m sections. Slides from
A total of 68 subjects with histologically documented distal each specimen were stained by using the triple stain includ-
(i.e., noncardiac) gastric adenocarcinoma were studied. On ing hematoxylin and eosin, silver stain, and Alcian blue
the basis of the biopsy samples, cancers were distinguished (22). H. pylori infection was assessed as either present or
as solid or glandular (20, 21). When both solid and glandular absent. A visual analog scale was used to grade IM accord-
features coexisted, the case was categorized according to the ing to the guidelines of the updated Sydney system (14).
most representative phenotype: 54 patients had a glandular- The information gathered from the evaluation of the
type carcinoma and 14 had a solid-type gastric cancer. There mapped biopsy specimens was used to assign each patient to
were 28 men and 40 women in this group; their mean age one of the following categories: 1) Absence of intestinal
was 48.6 ⫾ 14.4 yr (range, 24 –98 yr). metaplasia: no intestinalization detected in any of the gas-
tric biopsy specimens; 2) Focal intestinal metaplasia: in-
Control Patients testinal metaplasia focally present in one to three biopsy
A total of 67 consecutive patients with nonulcer dyspepsia specimens; 3) Multifocal intestinal metaplasia: intestinal
(39 men and 28 women; mean age 42.3 ⫾ 11.5 yr, range metaplasia present in four to six biopsy specimens; or 4)
19 –76 yr) who underwent endoscopic examination at the Extensive intestinal metaplasia: mucosal intestinalization
same institution in which patients were recruited. present in more than six biopsy sites.
AJG – June, 2000 Topography of Intestinal Metaplasia 1433

Table 1. Extent of Intestinal Metaplasia and Its Relationships With HID-Phenotype*

Controls Cancer Patients
HID-IM HID-IM
Extent of
Phenotype Phenotype
Intestinal Mean
Metaplasia Cases HP⫹ M/F Mean Age C I Cases HP⫹ M/F Age C I
Absence of 41 32 26/15 35 0 0 9 5 4/5 51 0 0
intestinal
metaplasia
Focal 16 13 5/21 42 16 0 14 12 8/6 63 12 2
intestinal
metaplasia
Multifocal 10 9 8/2 50 6 4 32 32 18/14 58 6 26
intestinal
metaplasia
Extensive 0 13 13 8/5 55 1 12
intestinal
metaplasia
* The HID-phenotype of the intestinal metaplasia (HID-IM) is also indicated (HID-IM ⫽ phenotype of intestinal metaplasia according to the high-iron-diamine stain; C ⫽
complete-type intestinal metaplasia; I ⫽ incomplete-type intestinal metaplasia; HP⫹ ⫽ H. pylori-positive patients; M/F ⫽ Male/Female.

Sections with any degree of intestinal metaplasia were RESULTS

stained with Alcian blue at pH 2.5 and high iron diamine
(AB/HID) (16). In accordance with standardized criteria, IM
was distinguished in complete (i.e., small intestinal type)
and incomplete (i.e., colonic type) (23). The interobserver
consistency in the categorization of the intestinal metaplasia
subtypes was calculated as the overall proportion of agree-
ment (the fraction of total pair observations in which the
same result was obtained) and was tested by using the ␬
statistic and Fleiss’s suggestion (24). Tested in a series of 50
cases, the overall agreement was 93% (␬ ⫽ 0.91). By
applying a semiquantitative scoring method to each set of
biopsies, each patient was categorized as having complete
or incomplete IM. Because no established criteria exist to
categorize a subject as having complete or incomplete
type IM, we arbitrarily assigned to the “incomplete”
group any subject in whom ⱖ30% of the total IM present
on the aggregate biopsy specimens was of the incomplete
type.
Statistical Analysis
Data are expressed as absolute and percent frequencies and
as means ⫾ standard deviations. Statistical analyses were
performed by using the ␹2 test for frequencies, Yates cor-
rected, or for linear trend, as appropriate. Analysis of vari-
ance (one-way ANOVA) for means, with Bonferroni’s test
for multiple comparisons. Odds ratios (OR) were calculated
by the Peto formula. A value of p ⬍ 0.05 was considered
significant. For all calculations except the OR, the Bio
Medical Data Processing statistical package (BMDP, Uni-
versity of California, Los Angeles, CA) was used. Spear-
man’s correlation coefficients were used to test whether the
presence of IM in one biopsy site was more likely to be
associated with IM detection in one or more of the other
sampling sites (25).
Prevalence of H. pylori Infection
H. pylori infection was detected histologically in 62 of the
68 patients with cancer (92%) and in 54 of the 67 controls
(80.6%). Among the 54 patients with glandular-type cancer
the prevalence of infection was 96.3%, whereas it was
71.4% in patients with the diffuse type (p ⬍ 0.05). Six
cancer patients, four with solid type and two with glandular
type, were H. pylori-negative.
Prevalence of Intestinal Metaplasia
Overall, IM was present in 26 of the 67 (39%) control
patients and in 59 of the 68 patients with gastric cancer
(87%). Focal IM was detected in 16 controls (24%) and in
14 cancer patients (21%). Of the 68 cancer patients, 32
(47%) had multifocal metaplastic lesions; the same pattern
was detected in 10 control patients (15%). Extensive IM was
never detected in the absence of gastric cancer, whereas it
coexisted with gastric adenocarcinoma in 13 cases (19%).
The difference in the prevalence of IM between cancer and
control patients was highly significant (␹2 ⫽ 39.43; DF ⫽ 3;
p ⬍ 0.0001).
Extent and Phenotypes of Intestinal Metaplasia (Table 1)
Of the entire study population of 135 subjects, 30 had foci
of IM in one to three biopsy sites (focal IM); 28 of these
subjects (93%) had complete IM, whereas two subjects were
classified (as defined above) as having incomplete-type IM.
Among the 42 subjects with multifocal metaplasia (four to
six biopsy sites with IM), 11 (39.4%) had the complete type
and 31 (60.6%) the incomplete type. A similar ratio was
present in the individuals with seven or more metaplastic
sites (extensive IM, 13 cases): 11 patients had complete IM
(84%) and two (16%) had the incomplete type. These data
indicate that the likelihood of detecting incomplete IM is
1434 Cassaro et al.
Table 2. Topographical Patterns of Intestinal Metaplasia and Their Prevalence in Patients and Controls*
Controls
Topographical Mean
Patterns Cases M/F Age
Focal 16 5/11 42
Antrum-predominant 6 6/0 51
Magenstraße 4 2/2 50
Diffuse 0 13
* The HID-phenotype of the intestinal metaplasia (HID-IM) is also indicated. C ⫽ complete-type intestinal metaplasia; I ⫽ incomplete-type intestinal metaplasia.
much lower when the metaplasia is sparse and rare (three or
fewer sites involved) than when it is more extensive (four or
more sites involved) (␹2 ⫽ 15.11; DF ⫽ 2; p ⬍ 0.001).
Incomplete-type metaplastic epithelium was predominantly
located in the antrum (A2, A3, and A4) and, to a lesser
extent, in the distal corpus along the lesser curvature (B1 and
B2). Incomplete metaplasia in the antrum frequently coex-
isted with incomplete metaplasia in the cardiac area; in
contrast, no incomplete metaplasia was found in the cardia
unless concurrently present in the antrum. Sites B4, B5, and
B6 rarely showed small foci of incomplete-type metaplasia.
Topographical Patterns of Intestinal Metaplasia (Table 2)
The Spearman’s test was applied to determine whether the
intestinalized mucosa was distributed in a random fashion or
instead aggregated in discernible topographical patterns.
In approximately equal numbers of cancer and control
patients (14 and 16, respectively) individual foci of IM were
randomly distributed throughout the stomach. In these sub-
jects the areas of the lesser curvature just proximal (B1) and
just distal (A2) to the incisura angularis and an area of the
greater curvature just proximal to the pylorus (A4) were
most commonly affected. We refer to this pattern as “Focal”
(Fig. 2A).
In seven cancer patients and six controls, IM involved
most of the antrum (both the lesser and the greater curva-
ture) from the incisura angularis to the pylorus (Fig. 2B).
None of these subjects showed IM in the cardia (C), nor at
the proximal lesser curvature (B1 and B2), or in the oxyntic
mucosa of the greater curvature (B4, B5, and B6). This
distribution is referred to as the “Antrum-predominant” pat-
tern.
In 25 cancer patients and four controls, IM spread
throughout the lesser curvature from the cardias to the
pylorus, also involving greater curvature of the prepyloric
antrum (Fig. 2C). This pattern involves almost exactly the
region of the stomach known as the Magenstraße (a name
used by German anatomists to define a narrow area along
the entire lesser curvature believed to represent the canal
through which food would pass from the esophagus to the
intestine), generally sparing the remainder of the gastric
mucosa. We named this type of distribution of IM the
“Magenstraße” pattern.
A “Diffuse” pattern (Fig. 2D) involving essentially the
AJG – Vol. 95, No. 6, 2000
Cancer Patients
HID-IM HID-IM
Phenotype Phenotype
Mean
C I Cases M/F Age C I
16 0 14 8/6 63 12 2
5 1 7 5/2 61 1 6
1 3 25 13/12 57 5 20
8/5 55 1 12
entire gastric mucosa, with the exception of the fundic areas
(B5 and B6), was found exclusively in 13 patients with
cancer.
Type of Intestinal Metaplasia and Gastric Cancer
Of the 67 control patients, 26 had intestinal metaplasia: of
these, 22 (85%) had only the complete type, and four (15%)
had the incomplete type. Incomplete-type IM was signifi-
cantly associated with the presence of cancer. Among 59
cancer patients showing metaplastic lesions, the incomplete
phenotype was detected in 41 cases (70%), whereas 18
patients (30%) had mostly complete IM (␹2 ⫽ 7.13; DF ⫽
1; p ⬍ 0.007). The relative extent of incomplete-type IM
correlated with the four topographical patterns of the meta-
plastic lesions, being lowest among subjects with Focal IM
and progressively increasing in the Antral-predominant, the
Magenstraße, and the Diffuse patterns.
Pattern of Intestinal Metaplasia and Cancer Risk
In this cross-sectional study, different patterns of metaplasia
were associated with strikingly different cancer risk. Be-
cause the prevalence of either the Focal or the Antral-
predominant patterns were comparable in both the patients
and the controls, we arbitrarily assigned the likelihood of
these groups having adenocarcinoma as equal to 1. By
adopting such a reference value, the Magenstraße pattern
was associated with a 6-fold risk of adenocarcinoma (OR ⫽
5.7; 95% CI: 1.3–26), whereas the Diffuse pattern showed a
12-fold increase in the risk for gastric cancer (OR ⫽ 12.2;
95% CI: 2.0 –72.9).
Histological Type and Location
of Adenocarcinoma (Table 3)
Table 3 illustrates the prevalence of gastric cancer in all
patients (68 cancer patients and 67 noncancer controls)
according to the pattern of IM (none, and each of the four
topographical patterns of gastric intestinalization). For the
purpose of this study, a cancer was deemed to be located in
the cardia if present exclusively within approximately 2 cm
distal to the gastroesophageal junction. Noncardial carcino-
mas were further divided into distal (restricted to the an-
trum) and proximal (restricted to the corpus). Nine of the 68
(18%) gastric cancers did not coexist with metaplastic le-
sions; all of these cancers had a solid histological pattern,
AJG – June, 2000 Topography of Intestinal Metaplasia 1435

Figure 2. The four patterns of intestinal metaplasia. These patterns are extrapolated from the frequency distribution of metaplastic changes
in the individual biopsy sites and are not intended to depict the precise extent of the intestinal metaplasia; rather, they represent the
prevalent configuration associated with different degrees of cancer risk. A (“Focal”): scattered foci, mostly in the lesser curvature just
proximal and just distal to the incisura angularis; B (“Antrum-predominant”): involving most of the antrum (both the lesser and the greater
curvature) from the incisura angularis to the pylorus; C (“Magenstraße”): spread throughout the lesser curvature from the cardia to the
pylorus, also involving greater curvature of the pre-pyloric antrum; and D (“Diffuse”): involving essentially the entire gastric mucosa, with
the exception of the fundic areas.

Table 3. Cancer Prevalence, Cancer Histotype, and Tumor Location in the 50 Subjects Without Intestinal Metaplasia and in the 85
Patients, Grouped According to the Four Topographical Patterns of Intestinal Metaplasia
No Intestinal Antrum-
Patterns of Intestinal Metaplasia Metaplasia Focal Predominant Magenstraße Diffuse
Cases 50 30 13 29 13
Cancer patients 9 (18%) 14 (46.6%) 7 (53.8%) 25 (86.2%) 13 (100%)
Histological type: glandular/solid 9/9 10/4 5/2 23/2 12/1
Glandular type prevalence (%) 0 71.4 71.4 92 100
Cancer location
Cardia 3 3 1 1 0
Noncardia 6 11 6
Proximal 1 0 0 5 4
Proximal ⫹ distal 1 4 2 5 0
Distal 4 7 4 14 9
1436 Cassaro et al.
and three were topographically restricted to the cardia area.
Glandular adenocarcinoma was most common in patients
with coexisting metaplasia. In all, 71% of the patients with
either Focal or Antrum-predominant IM and 92% of those
with either the Magenstraße or the Diffuse pattern had a
glandular phenotype. One cancer found in each of the two
latter patterns was localized to the cardia.
DISCUSSION
The central hypothesis of this study was that certain topo-
graphical patterns of IM in the stomach, the overall extent of
the intestinalized mucosa, and the subtype of metaplasia
may be associated with different levels of gastric cancer
risk. Our results show that, in this group of Colombian
patients, IM was distributed in four distinct patterns. The
two patterns with less extensive involvement of the gastric
mucosa (Focal and Antrum-predominant) consisted almost
exclusively of complete type IM (intestinal or type I) and
were equally represented in subjects with or without malig-
nancy. This finding is consistent with the observation that
scattered foci of mostly antral IM are commonly present
both in subjects with nonulcer, H. pylori-gastritis and those
with duodenal ulcer (26, 27). These patterns of limited
intestinalization of the gastric mucosa are likely to have
limited, if any, significance as predictors of gastric cancer.
In contrast, the two other models of topographical distri-
bution (Magenstraße and Diffuse) were characterized by a
much greater extension, by the involvement of the corpus,
and by a distinct predominance of incomplete type (colonic
or types II and III) metaplasia. The Magenstraße pattern was
associated with a 6-fold cancer risk compared with the Focal
and Antral-predominant patterns; the Diffuse pattern, found
exclusively in cancer patients, conferred a 12-fold risk for
cancer. Both of these high-risk patterns shared the extensive
intestinalization of the mucosa of the lesser gastric curvature
from the prepyloric to the cardia zone. A similar distribution
of IM was described in two previous studies that addressed
the macroscopic and histological evaluation of the IM co-
existing with gastric cancer. Stemmermann and Hayashi
(28) sampled extensively 46 resected stomachs from Japa-
nese cancer patients, and described a predominant intesti-
nalization of the antral-oxyntic border that coexisted with
metaplastic changes involving both the lesser curvature and
“ . . . a well-defined zone of IM in the gastric cardia.” In-
testinal metaplasia was also located predominantly in the
lesser curvature and up to the cardia in the stomach of 40
patients with gastric carcinomas confined to the mucosal
layer reported by Nakahara (29). Other studies conducted in
Japanese subjects also indicated that, in H. pylori gastritis,
the initial atrophic-metaplastic changes occur in the area of
the incisura angularis and expand both proximally and
distally to involve much of the antrum and the distal corpus
(28 –32).
These converging data obtained in Japanese and Colom-
bian subjects suggest that the precursors of gastric cancer,
AJG – Vol. 95, No. 6, 2000
and possibly the carcinogenic pathway, may be analogous in
ethnically diverse groups with similar cancer risk. The con-
sistency of these observations may also have practical as-
pects. First, it lends further support to the recommendation
made in the updated Sydney system that a biopsy specimen
be obtained from the incisura angularis to increase the
detection of intestinalized areas. Second, it suggests that
endoscopists ought to pay special attention to the sampling
of the cardia, a frequently neglected zone. In selected pa-
tients or in high-risk epidemiological contexts, the simulta-
neous evaluation of biopsies from these two areas (cardia
and angulus) might yield useful information for the assess-
ment of cancer risk.
The relationship between the incomplete type of IM,
dysplasia, and gastric carcinoma has become a topic of great
practical interest. Several researchers have stated that the
incomplete types of IM, those that have an appearance
similar to colonic mucosa and contains sulfated mucins,
indicate a greater risk for cancer. Incomplete IM has been
equated with dysplasia in one study (33) and has been
associated with a greater incidence of gastric adenocarci-
noma in a retrospective study conducted in Slovenia (34).
Gastroenterologists often ask pathologists whether a focus
of IM has been subtyped, and they inquire about the pre-
dictive significance of the findings. In an attempt to cast
some light on this controversial issue, we used the high-iron
diamine stain to determined the subtype of each focus of IM
in all patients in this study, and correlated the type of
metaplasia with the total extent of the IM and the cancer
risk.
A significant association was found between the total
extent of IM and its phenotype. Specifically, in our patients
a greater extension of IM was associated with a dispropor-
tionately greater increase in the extent of the incomplete
type. Furthermore, of 30 patients with focal IM (presumably
one of the earlier stages of the intestinalizing process), only
two (6.6%) had the incomplete type. This observation sup-
ports the theory (32) that gastric intestinalization proceeds
from the complete (syalomucin-predominant) type to a less
differentiated (sulfomucin-predominant) phenotype.
In this population, both the extent and the presence of
incomplete-type intestinal metaplasia were strongly related
to cancer risk. The incomplete type was significantly pre-
dominant in cancer patients, whereas complete IM was more
prevalent among the controls, confirming the previously
noted association of complete IM with benign gastric con-
ditions, such as gastric and duodenal ulcer, as well as
chronic gastritis without ulcer (35, 36).
In light of these observations, does histochemical pheno-
typing of metaplasia increase the accuracy of the assessment
of cancer risk? We do not believe that there is an unequiv-
ocal answer to this question. When a research-type biopsy
protocol such as ours is followed, the finding of IM in four
or more of 12 biopsy specimens can be taken as evidence of
extensive metaplastic changes. In such a case, the determi-
nation of the IM subtype may be redundant, because a
AJG – June, 2000
patient with extensive metaplasia should be viewed as being
at risk for cancer, irrespective of the IM subtype. On the
other hand, when only a few biopsy specimens are available
for examination (as happens in the “real world” of managed
care), the finding of significant foci of incomplete metapla-
sia may be used as a surrogate to indicate that extensive
metaplastic changes have taken place and, therefore, that the
patient has an increased cancer risk. It must be kept in mind,
however, that these speculations reflect our experience with
a group of Colombian subjects. Before guidelines can be
formulated, detailed mapping studies must be performed in
different populations to test the broader significance of our
findings.
The majority of adenocarcinomas in our patients were of
the glandular phenotype, the type of gastric carcinoma most
commonly found in high-risk populations. Virtually all of
these tumors were associated with extensive intestinaliza-
tion of the gastric mucosa. All nine cases of gastric cancers
without coexisting metaplasia had a solid pattern. These
findings are consistent with the generally accepted theory
that glandular-type carcinomas result from the neoplastic
transformation of previously intestinalized glands (20, 37).
Intestinal metaplasia was either absent or focal in six of the
eight patients with cancer restricted to the region of the
cardia, indicating that at least a portion of the adenocarci-
nomas of the gastroesophageal junction follow carcinogenic
pathways different from those of the most common endem-
ic-type gastric adenocarcinomas.
ACKNOWLEDGMENTS
This study was supported by the Department of Veterans
Affairs, Washington, DC, and by the Cittadella Association
against Cancer (“CCC”), Cittadella-Padova, Italy.
Reprint requests and correspondence: Robert M. Genta, M.D.,
Department of Pathology—113, Veterans Affairs Medical Center,
2002 Holcombe Boulevard, Houston, TX 77030.
Received Mar. 18, 1999; accepted Jan. 26, 2000.
REFERENCES
1. Neugut AI, Hayek M, Howe G. Epidemiology of gastric
cancer. Semin Oncol 1996;23:281–91.
2. Correa P, Cuello C, Duque E, et al. Gastric cancer in Colom-
bia. III. Natural history of precursor lesions. J Natl Cancer Inst
1976;57:1027–35.
3. Correa P, Cuello C, Haenszel W. The pathogenesis of gastric
carcinoma—Epidemiologic pathology of precursor lesions.
Leber Magen Darm 1976;6:72–9.
4. Correa P. Human gastric carcinogenesis: A multistep and
multifactorial process—First American Cancer Society Award
Lecture on Cancer Epidemiology and Prevention. Cancer Res
1992;52:6735– 40.
5. Nomura A, Stemmermann GN. Helicobacter pylori and gastric
cancer. J Gastroenterol Hepatol 1993;8:294 –303.
6. Rugge M, Leandro G, Farinati F, et al. Gastric epithelial
Topography of Intestinal Metaplasia 1437
dysplasia. How clinicopathologic background relates to man-
agement. Cancer 1995;76:376 – 82.
7. Rugge M, Cassaro M, Leandro G, et al. Helicobacter pylori in
promotion of gastric carcinogenesis. Dig Dis Sci 1996;41:
950 –5.
8. Infection with Helicobacter pylori. IARC Monogr Eval Car-
cinog Risks Hum 1994;61:177–240.
9. Blaser MJ, Chyou PH, Nomura A. Age at establishment of
Helicobacter pylori infection and gastric carcinoma, gastric
ulcer, and duodenal ulcer risk. Cancer Res 1995;55:562–5.
10. Correa P, Haenszel W, Cuello C, et al. Gastric precancerous
process in a high risk population: Cross-sectional studies.
Cancer Res 1990;50:4731– 6.
11. Ihamaki T, Sipponen P, Varis K, et al. Characteristics of
gastric mucosa which precede occurrence of gastric
malignancy: Results of long-term follow-up of three family
samples. Scand J Gastroenterol 1991;186(suppl):16 –23.
12. Kokkola A, Haapiainen R, Laxen F, et al. Risk of gastric
carcinoma in patients with mucosal dysplasia associated with
atrophic gastritis: A follow up study. J Clin Pathol 1996;49:
979 – 84.
13. Dixon MF. Histological responses to Helicobacter pylori
infection: Gastritis, atrophy and preneoplasia. Baillieres Clin
Gastroenterol 1995;9:467– 86.
14. Dixon MF, Genta RM, Yardley JH, Correa P. Classification
and grading of gastritis. The updated Sydney System. Inter-
national Workshop on the Histopathology of Gastritis, Hous-
ton 1994. Am J Surg Pathol 1996;20:1161– 81.
15. Rugge M, Cassaro M, Farinati F, et al. Helicobacter pylori and
atrophic gastritis: Importance of the cagA status. J Natl Cancer
Inst 1996;88:762–3.
16. Jass JR, Filipe MI. Sulphomucins and precancerous lesions of
the human stomach. Histopathology 1980;4:271–9.
17. Filipe MI, Potet F, Bogomoletz WV, et al. Incomplete sul-
phomucin-secreting intestinal metaplasia for gastric cancer.
Preliminary data from a prospective study from three centres.
Gut 1985;26:1319 –26.
18. Filipe MI, Munoz N, Matko I, et al. Intestinal metaplasia types
and the risk of gastric cancer: A cohort study in Slovenia. Int
J Cancer 1994;57:324 –9.
19. Correa P, Haenszel W, Cuello C, et al. Gastric precancerous
process in a high risk population: Cohort follow-up. Cancer
Res 1990;50:4737– 40.
20. Arslan PC, Rugge M. Gastric cancer: Problems in histological
diagnosis. Histopathology 1982;6:391– 8.
21. Solcia E, Fiocca R, Luinetti O, et al. Intestinal and diffuse
gastric cancers arise in a different background of Helicobacter
pylori gastritis through different gene involvement. Am J Surg
Pathol 1996;20(suppl 1):S8 –22.
22. Genta RM, Robason GO, Graham DY. Simultaneous visual-
ization of Helicobacter pylori and gastric morphology: A new
stain. Hum Pathol 1994;25:221– 6.
23. Filipe MI, Jass JR. Intestinal metaplasia subtypes and cancer
risk. In: Filipe MI, Jass JR, eds. Gastric carcinoma. Edinburgh:
Churchill Livingstone, 1986:97–115.
24. Fleiss JI. Statistical methods for rates and proportions. 2nd ed.
New York: Wiley, 1981:212–36.
25. Armitage P, Barry G. Statistical methods in medical research.
Oxford: Blackwell Scientific Publications, 1987:329 –32.
26. Genta RM, Lew GM, Graham DY. Changes in the gastric
mucosa following eradication of Helicobacter pylori. Mod
Pathol 1993;6:281–9.
27. Wyatt JI. Histopathology of gastroduodenal inflammation:
The impact of Helicobacter pylori. Histopathology 1995;26:
1–15.
28. Stemmermann GN, Hayashi T. Intestinal metaplasia of the
1438 Cassaro et al.
gastric mucosa: A gross and microscopic study of its distri-
bution in various disease states. J Natl Cancer Inst 1968;41:
627–34.
29. Nakahara K. Special features of intestinal metaplasia and its
relation to early gastric carcinoma in man: Observation by a
method in which leucine aminopeptidase activity is used.
J Natl Cancer Inst 1978;61:693–701.
30. Kimura K. Chronological transition of the fundic-pyloric bor-
der determined by stepwise biopsy of the lesser and greater
curvatures of the stomach. Gastroenterology 1972;63:584 –92.
31. Kimura K, Satoh K, Ido K, et al. Gastritis in the Japanese
stomach. Scand J Gastroenterol 1996;214(suppl):17–20.
32. Stemmermann GN. Intestinal metaplasia of the stomach. A
status report. Cancer 1994;74:556 – 64.
33. Tosi P, Filipe MI, Luzi P, et al. Gastric intestinal metaplasia
AJG – Vol. 95, No. 6, 2000
type III cases are classified as low-grade dysplasia on the basis
of morphometry. J Pathol 1993;169:73– 8.
34. Filipe MI, Munoz N, Matko I, et al. Intestinal metaplasia types
and the risk of gastric cancer: A cohort study in Slovenia. Int
J Cancer 1994;57:324 –9.
35. Soren A. Joint inflammations as complications in intestinal
inflammations. Z Rheumatol 1981;40:1–5.
36. Utriainen M, Hanninen ML. Detection of Helicobacter-like
bacteria in porcine gastric biopsy samples by amplification of
16S rRNA, ureB, vacA and cagA genes by PCR. Vet Res
Commun 1998;22:373– 83.
37. Driessen A, Ectors N, Creemers J, et al. Intestinal metapla-
sia in gastric malignancies: A comparison between carci-
noma and limphoma. Eur J Gastreonterol Hepatol 1998;10:
595– 600.