Utilization of Cytokeratins 7 and 20 Does Not
Differentiate between Barrett’s Esophagus and Gastric
Cardiac Intestinal Metaplasia
I. A. Mohammed, M.D., C. J. Streutker, M.D., M.Sc., F.R.C.P.C., R. H. Riddell, M.D., F.R.C.P.C.
Department of Pathology and Molecular Medicine (IAM, CJS, RHR), McMaster University, Hamilton,
Ontario, Canada; Department of Laboratory Medicine (CJS), St. Michael’s Hospital, Toronto, Ontario,
Canada; and Department of Pathology and Laboratory Medicine (RHR), Mt. Sinai Hospital, Toronto,
Long segment Barrett’s esophagus (LSBE) is a rec-
ognized risk factor for the development of esopha-
geal dysplasia and carcinoma. However, the risk of
dysplasia arising within intestinal metaplasia below
a normal-appearing Z-line (i.e., in native cardiac
mucosa) is unknown. Regular endoscopic surveil-
lance is required in patients with LSBE and is fre-
quently performed in short segment BE (SSBE), but
the need for surveillance in cardiac intestinal meta-
plasia (CIM) is unknown. Unfortunately IM arising
in SSBE and immediately below a normal Z-line can
be indistinguishable histologically on H&E stains.
Previous reports suggest that the appearance of su-
perficial CK20 immunohistochemical staining ac-
companied by intermediate and deep CK7 positivity
is characteristic of BE, whereas CIM specimens
show superficial and deep CK20 positivity and weak
to absent CK7 staining. We hypothesized that
CK7/20 immunostaining of metaplastic biopsies
from the esophagus and stomach would allow com-
plete differentiation of these two entities when cor-
related with the endoscopic appearance. We under-
took an evaluation of gastric and esophageal
specimens to determine whether these characteris-
tics were valid. Cases of both BE (long and short
segment) and CIM, as well as cases of gastric cardiac
biopsies lacking IM, were evaluated for CK7 and
appearance. We observed that, although the “Bar-
rett’s” pattern of CK7/20 was maintained for many
cases of BE, the sensitivity and specificity were only
moderate (65% and 56%, respectively). The pattern
of staining for the CIM was variable, i.e., some cases
showed a CK7/20 Barrett’s pattern despite a normal
appearance at endoscopy. The differences between
this and previous studies may be due to inaccurate
visualization of SSBE on endoscopy, the develop-
ment of very early SSBE cases, inter-observer vari-
ability, fixation differences, or antibody differences.
Whatever the cause of the differences, if results be-
tween laboratories are not comparable, CK7/20 im-
munostaining cannot be used to differentiate reli-
ably between IM present in biopsy specimens taken
from above versus below the Z-line. However, fur-
ther studies should be performed to determine
whether the presence or absence of a Barrett’s pat-
tern of CK7/20 immunostaining could predict pro-
gression to dysplasia or carcinoma.
KEY WORDS: Barrett’s esophagus, Cardiac intesti-
nal metaplasia, Cytokeratin.
Mod Pathol 2002;15(6):611–616
Barrett’s esophagus (BE) is defined as the presence
of specialized columnar epithelium with intestinal
goblet cells, replacing the normal squamous epithe-
lium (1–6). BE is generally subdivided into long
segment BE (LSBE), where columnar mucosa ex-
tends 3 cm or more above the gastroesophageal
junction (GEJ), and short segment BE (SSBE), where
the specialized columnar epithelium is restricted to
less than 3 cm above the GEJ. These definitions
were, in the past, based on endoscopic evaluation.
However, it has been recognized that histological
identification of intestinal metaplasia is of particu-
lar importance, as it is this form of mucosa that is
premalignant in nature and in which dysplasia and
adenocarcinomas arise (7). The risk of progression
from intestinal metaplasia to dysplasia and carci-
noma is well recognized in LSBE, though it is now
thought to be lower than previously estimated, at
approximately 0.5% annually (8). Patients with
Copyright © 2002 by The United States and Canadian Academy of
VOL. 15, NO. 6, P. 611, 2002 Printed in the U.S.A.
Date of acceptance: February 18, 2002.
Address reprint requests to: Dr. R. H. Riddell, Department of Pathology
and Laboratory Medicine, Suite 600, Mt. Sinai Hospital, 600 University
Avenue, Toronto, Ontario, Canada; e-mail: firstname.lastname@example.org; fax:
LSBE are therefore regularly screened to evaluate
any endoscopic or histological progression that
would indicate the need for more aggressive ther-
apy. The specific incidence of adenocarcinoma in
SSBE is not well quantified, though Rudolph et al.
(9) have shown that the risk for SSBE is essentially
the same as for LSBE. Surveillance endoscopy in
patients with SSBE, though still controversial, is
frequently performed (10).
The gastric cardia is defined as the most proximal
portion of the stomach, immediately distal to the
normal GEJ (Z-line). Cardiac-type mucosa is de-
fined as mucous glands that lack endocrine cells
but may contain occasional parietal cells within the
pits. It is thought to extend endoscopically distally
for up to 2 cm from the GEJ in healthy individuals
(11), although it is much shorter histologically.
However, an autopsy study has noted that the pres-
ence and extent of cardiac mucosa increases with
age (12). The finding of cardiac-type mucosa in the
lower esophagus, which does not meet the defini-
tion of BE, has been correlated with the presence
and severity of gastroesophageal reflux disease
(GERD) (13). These studies suggest that cardiac-
type mucosa may not be a normal finding in either
the stomach or esophagus. Cardiac-type mucosa
may itself be metaplastic and, therefore, may be a
precursor to cardiac intestinal metaplasia (CIM).
CIM is defined as the presence of intestinal meta-
plasia within cardiac-type mucosa below an endo-
scopically normal Z-line. The etiology of CIM may
be different from BE. The main risk factor for BE is
GERD (13–15). Although the presence of Helicobac-
ter pylori infection is linked to CIM (16–20), in the
absence of H. pylori infection CIM may be related
to GERD (21). CIM may therefore have several eti-
ologies. Similar to SSBE, the risk of development of
dysplasia and adenocarcinoma in CIM is unknown.
However, clinical studies suggest that the tendency
for developing dysplasia or carcinoma in patients
with intestinal metaplasia of the cardia is consider-
ably lower than in BE (16, 22). Consequently, pa-
tients with CIM are not usually enrolled in screen-
ing programs. Considering the marked rise in the
incidence of gastric cardiac adenocarcinoma in
conjunction with esophageal adenocarcinoma, this
may change in the future (23, 24).
These differences in treatment protocols and risk
of malignancy between LSBE, SSBE, and CIM indi-
cate the need for accurate diagnosis of the different
entities. However, differentiating these entities can
be very difficult. Endoscopically, an irregular Z-line
can be a normal finding, and determination of
where the esophagus ends and the cardia begins
can be unreliable. Furthermore, histological exam-
ination of biopsy specimens is frequently of little
assistance because the appearance of intestinal
metaplasia is identical in both locations. The pres-
ence of squamous-lined ducts below the surface
can help to identify the tissue as esophagus; how-
ever, these are rarely identified. Staining for acid
mucins and specifically sulfomucins has been sug-
gested as a means of identifying esophageal IM
(25–27); however, this has not been found to be
reliable (19, 28). Immunohistochemical staining of
different cytokeratin subunits, particularly cytoker-
atins 7 and 20, has been proposed as a method that
will differentiate them (28, 29).
In our study, we evaluated CK7 and CK20 pattern
in patients with BE (both long and short segments),
CIM, antral intestinal metaplasia (AIM), and normal/
inflamed cardiac biopsies to determine whether the
staining pattern of these two antibodies would assist
MATERIALS AND METHODS
Endoscopic biopsy specimens were identified by
searching the information system for specific diag-
nostic categories. These diagnoses were based on
standard H&E staining. Diagnostic categories in-
cluded esophageal mucosa with intestinal metapla-
sia (SSBE and LSBE), cardiac mucosa with intestinal
metaplasia (CIM), antral mucosa with intestinal
metaplasia (AIM), and normal/inflamed cardiac
mucosa. In this last category, the cases of normal
and inflamed cardia were combined as representa-
tives of mucosa lacking intestinal metaplasia. In all
cases, the endoscopy report was consulted to de-
termine the state of the Z-line, the presence or
absence of an endoscopically-evident hiatus her-
nia, and to confirm the location of the biopsies. In
cases where biopsies were labeled as “GEJ,” these
reports were used to determine whether there was
any indication of SSBE. In total, 49 cases of BE, 13
cases of CIM, 7 cases of AIM, and 11 cases of nor-
mal/inflamed cardia were retrieved for the this
study. In all cases, the biopsy specimens were
formalin-fixed for approximately 24 hours. They
then were processed and paraffin-embedded. Sec-
tions of 3-?m thickness were cut. H&E-stained sec-
tions were examined. Alcian blue/periodic acid-
Schiff staining was utilized to confirm the presence
of intestinal-type goblet cells.
Antibodies to CK7 and CK20 were utilized for this
study as follows. The OV-TL clone (DAKO Diagnos-
tics, Mississauga, Ontario, Canada) was utilized for
CK7. For CK20, the Ks 20.8 clone was used. For both
antibodies, slides were digested with Pronase fol-
lowed by antibody at a dilution of 1:50 for 32 min-
utes. Detection was performed with the Zymed goat
anti-mouse streptavidin-biotin system.
Evaluation of the pattern of CK7/20 immunohis-
tochemical staining was performed simultaneously
by three pathologists (IM, CJS, RHR), two of whom
were blinded to the diagnosis. The Barrett’s CK7/20
pattern was considered present if CK20 staining
was seen in surface epithelium with or without
superficial and deep glands and CK7 staining was
present in both superficial and deep glands in areas
of IM. Statistical analysis of results was performed
using the Fisher’s exact test.
Review of patient history where available re-
vealed that the majority of the patients with BE
were male (77%), which was comparable with the
CIM patients (62%) (Table 1). Few patients with BE
had associated H. pylori infection (3 patients ? 6%),
while most had a history of GERD and/or hiatus
hernia (69% and 35%). The presence of hiatus her-
nia may be underestimated due to our reliance on
endoscopic reports for this diagnosis. Patients with
CIM had similar GERD and hiatus hernia rates (92%
and 31%), and three patients were diagnosed as
having concurrent Helicobacter infection (25%).
A total of 49 biopsies from patients clinically di-
agnosed with BE were evaluated for immunohisto-
chemical staining pattern (Table 2). A Barrett’s
CK7/20 staining pattern was identified in 32 of the
48 slides (65%) (Fig. 1). When subdivided into LSBE
versus SSBE, 10 of 22 LSBE (54%) and 22 of 27 SSBE
(81%) had a Barrett’s pattern. The remainder of the
cases had a variety of immunohistochemical stain-
ing patterns with the two antibodies, which did not
conform to the staining pattern as previously estab-
lished (Fig. 2).
In contrast, of the CIM biopsies, 9 of the 13 cases
(69%) showed a non-Barrett’s CK7/20 pattern (Fig.
3), while the remaining four biopsies demonstrated
the characteristic BE CK7/20 pattern (Fig. 4). A sim-
ilar trend was noted in the biopsies of normal or
inflamed cardiac mucosa—in these cases where
histological evidence of intestinal metaplasia was
lacking, 55% had a Barrett’s pattern of staining.
Antral intestinal metaplasia also was capable of
demonstrating a BE pattern of staining, though at a
lower incidence (29%).
In evaluating the utility of this test for diagnosis
of BE versus CIM, the sensitivity was 65%, specific-
ity was 69%, positive predictive value was 89%, and
negative predictive value was 35%. By Fisher’s exact
test, there was no significant difference in the pres-
ence of non-BE patterns in CIM as opposed to BE
patterns in all BE cases (2-tailed P ? .0547). There
was a significant difference when cases of CIM were
compared only with SSBE (2-tailed P ? .0054).
Cytokeratins (CK) are subunits of epithelial cell
intermediate filaments, which are genetically deter-
mined. There is some restriction of the types of
cytokeratins expressed in different epithelium. CK7
is a 54 kd polypeptide found in normal epithelium
ranging from lung, cervix, breast, and bile ducts to
bladder transitional cell epithelium; however, it is
essentially absent from gastrointestinal epithelia
and squamous cell epithelium. CK20 expression is
limited to intestinal epithelium and gastric foveolar
cells as well as urothelial umbrella cells and Merkel
cells of the epidermis (30, 31). Aberrant expression
of cytokeratins may be observed in dysplastic and
malignant cells. Previous studies attempted to dif-
ferentiate between BE and CIM using antibodies to
cytokeratins 7 and 20 (28, 29), though the utility of
utilizing immunohistochemistry to differentiate
these entities has been argued (32). Ormsby et al.
demonstrated an apparently unique pattern of im-
munoreactivity, designated the BE CK7/20 pattern,
showing superficial CK20 staining and strong CK7
staining of both superficial and deep glands in
100% of esophageal biopsy specimens from pa-
tients with LSBE (28, 29). In contrast, a Barrett’s
CK7/20 pattern was not observed in those patients
with gastric cardiac biopsies showing evidence of
intestinal metaplasia. For LSBE, the sensitivity in
this study was 97% with a specificity of 100%. Our
findings did not support this conclusion. Cases of
both LSBE and SSBE showed marked variability in
staining patterns, as did the biopsies from patients
with CIM. This resulted in relatively poor sensitivity
and specificity. There was a significant difference (P
TABLE 1. Clinical Characteristics of Patients with Barrett’s Esophagus, Cardiac Intestinal Metaplasia, and Antral
Male gender (n)
Hiatus hernia (n)
H. pylori positive (n)
Cytokeratins 7 and 20 in BE and CIM (I.A. Mohammed et al.) 613
? .0054) when the proportion of CIM with non-BE
patterns was compared with SSBE (by Fisher’s exact
The differences between this study and the pre-
viously reported results may have been due to a
less-strict interpretation of the specific Barrett’s
pattern of CK7/20 immunoreactivity compared
with the original study. However, for this method of
evaluation to be practical in the average surgical
TABLE 2. CK 7/20 Immunostaining Patterns
(n ? 49)
(n ? 22)
(n ? 27)
(n ? 13)
(n ? 7)
Inflamed or Normal
Cardia (n ? 11)
Barrett’s CK 7/20, n (%)32
Non-Barrett’s CK 7/20, n (%)
BE, Barrett’s esophagus; LSBE, long segment Barrett’s esophagus; SSBE, short segment Barrett’s esophagus; CIM, cardiac intestinal metaplasia; AIM,
antral intestinal metaplasia.
FIGURE 1. Barrett’s esophagus with Barrett’s CK7/20 pattern. (A)
H&E, (B) CK7, (C) CK20.
FIGURE 2. Barrett’s esophagus with non-Barrett’s CK7/20 pattern.
(A) H&E, (B) CK7, (C) CK20.
pathology laboratory, it must be applicable despite
reasonable inter-observer variability. The differ-
ences also may be due to inaccuracies in determin-
ing location at the time of endoscopy, or biopsies
may have been taken from areas of very short SSBE.
Differences in fixation methods or in the antibodies
used for immunohistochemistry also could account
for the variability in staining patterns. Regardless of
the cause, the fact that these differences exist indi-
cates that using cytokeratin profiles to distinguish
BE from CIM is not likely to be clinically useful.
Despite the presence of a significant difference be-
tween CIM and SSBE, in individual cases this
method still would not suffice to distinguish the
Though our results demonstrate that this meth-
odology does not strictly divide cases of BE and
CIM, they raise questions regarding the association
of the BE pattern and subsequent risk of carcinoma.
The previous studies note that in the cases where
anomalous staining was found for either BE or CIM
that these cases behaved according to the manner
in which they stained, i.e., BE with a non-BE
CK7/20 pattern acted like CIM. It may therefore be
possible to link the pattern of CK7/20 immunohis-
tochemical staining, irrespective of the endoscopic
location of the biopsy, to the risk of development of
FIGURE 3. Cardiac intestinal metaplasia with non-Barrett’s CK7/20
pattern. (A) H&E, (B) CK7, (C) CK20.
FIGURE 4. Cardiac intestinal metaplasia with Barrett’s CK7/20
pattern. (A) H&E, (B) CK7, (C) CK20.
Cytokeratins 7 and 20 in BE and CIM (I.A. Mohammed et al.)615
carcinoma. However, this has not been extensively Download full-text
studied and would require a long-term prospective
1. Paull A, Trier JS, Dalton MD, Camp RC, Loeb P, Goyal RK.
The histologic spectrum of Barrett’s esophagus. N Engl
J Med 1976;295:476–80.
2. Reid BJ, Weinstein WM, Lewin KJ, Haggitt RC, VanDeventer
G, DenBesten L, et al. Endoscopic biopsy can detect high-
grade dysplasia or early adenocarcinoma in Barrett’s esoph-
agus without grossly recognizable neoplastic lesions. Gastro-
3. Spechler SJ, Robbins AH, Rubins HB, Vincent ME, Heeren T,
Doos WG, et al. Adenocarcinoma and Barrett’s esophagus.
An overrated risk? Gastroenterology 1984;87:927–33.
4. Zwas F, Shields HM, Doos WG, Antonioli DA, Goldman H,
Ransil BJ, et al. Scanning electron microscopy of Barrett’s
epithelium and its correlation with light microscopy and
mucin stains. Gastroenterology 1986;90:1932–41.
5. Hamilton SR, Smith RR. The relationship between columnar
epithelial dysplasia and invasive adenocarcinoma arising in
Barrett’s esophagus. Am J Clin Pathol 1987;87:301–12.
6. Schmidt HG, Riddell RH, Walther B, Skinner DB, Riemann
JF. Dysplasia in Barrett’s esophagus. J Cancer Res Clin Oncol
7. DeMeester SR, Demeester TR. Columnar mucosa and intes-
tinal metaplasia of the esophagus: fifty years of controversy.
Ann Surg 2000;231:303–21.
8. Falk GW. Gastroesophageal reflux disease and Barrett’s
esophagus. Endoscopy 2001;33:109–18.
9. Rudolph RE, Vaughan TL, Storer BE, Haggitt RC, Rabinovitch
PS, Levine DS, et al. Effect of segment length on risk for
neoplastic progression in patients with Barrett esophagus.
Ann Intern Med 2000;132:612–20.
10. Sharma P. Recent advances in Barrett’s esophagus: short-
segment Barrett’s esophagus and cardia intestinal metapla-
sia. Semin Gastrointest Dis 1999;10:93–102.
11. Hayward J. The lower end of the esophagus. Thorax 1961;
12. Chandrasoma PT, Der R, Ma Y, Dalton P, Taira M. Histology
of the gastroesophageal junction: an autopsy study. Am J
Surg Pathol 2000;24:402–09.
13. Chandrasoma PT, Lokuhetty DM, Demeester TR, Bremmer
CG, Peters JH, Oberg S, et al. Definition of histopathologic
changes in gastroesophageal reflux disease. Am J Surg Pathol
14. Carton E, Mulligan ED, Keeling PW, Tanner A, McDonald G,
Reynolds JV. Specialized intestinal metaplasia: analysis of
oesophageal reflux disease. Br J Surg 2000;87:362–73.
15. El-Serag HB, Sonnenberg A, Jamal MM, Kunkel D, Crooks L,
Feddersen RM. Characteristics of intestinal metaplasia in the
gastric cardia. Am J Gastroenterol 1999;94:622–7.
16. Goldblum JR. Inflammation and intestinal metaplasia of the
gastric cardia: Helicobacter pylori, gastroesophageal reflux
disease, or both. Dig Dis 2000;18:14–9.
17. Hoshi T, Sasano H, Kato K, Ohara S, Shimosegawa T, Toyota
T, et al. Cell damage and proliferation in human gastric
mucosa infected by Helicobacter pylori–a comparison be-
fore and after H pylori eradication in non-atrophic gastritis.
Hum Pathol 1999;30:1412–7.
18. Jedrychowski W, Popiela T, Drews M, Gabryelewicz A, Mar-
licz K, Misiunia P, et al. Effect of Helicobacter pylori infec-
tion, smoking and dietary habits on the occurrence of an-
trum intestinal metaplasia. Clinico-epidemiological study in
Poland. Pol J Pathol 1999;50:289–95.
19. Ormsby AH, Kilgore SP, Goldblum JR, Richter JE, Rice TW,
Gramlich TL. The location and frequency of intestinal meta-
plasia at the esophagogastric junction in 223 consecutive
autopsies: implications for patient treatment and preventive
strategies in Barrett’s esophagus. Mod Pathol 2000;13:614–
20. Smith VC, Genta RM. Role of Helicobacter pylori gastritis in
gastric atrophy, intestinal metaplasia, and gastric neoplasia.
Microsc Res Tech 2000;48:313–20.
21. Goldstein NS, Karim R. Gastric cardia inflammation and
intestinal metaplasia: association with reflux esophagitis and
Helicobacter pylori. Mod Path 1999;12:1017–24.
22. Sharma P, Weston AP, Morales T, Topalovski M, Mayo MS,
Sampliner RE. Relative risk of dysplasia for patients with
intestinal metaplasia in the distal oesophagus and in the
gastric cardia. Gut 2000;46:9–13.
23. Devesa SS, Fraumeni JF. The rising incidence of gastric car-
dia cancer. J Natl Cancer Inst 1999;91:747–9.
24. Sampliner RE. Adenocarcinoma of the esophagus and gas-
tric cardia: is there progress in the face of increasing cancer
incidence? Ann Intern Med 1999;130:67–9.
25. Byrne JP, Bhatnagar S, Hamid B, Armstrong GR, Attwood SE.
Comparative study of intestinal metaplasia and mucin stain-
ing at the cardia and esophagogastric junction in 225 symp-
tomatic patients presenting for diagnostic open-access gas-
troscopy. Am J Gastroenterol. 1999;94:98–103.
26. Chen YY, Wang HH, Antonioli DA, Spechler SJ, Zeroogian
JM, Goyal R, et al. Significance of acid-mucin-positive non-
goblet columnar cells in the distal esophagus and gastro-
esophageal junction. Hum Pathol 1999;30:1488–95.
27. Sirigu F, Capeccioni S, Dessi A. Mucin histochemistry in
Barrett’s esophagus. Riv Eur Sci Med Farmacol 1990;12:229–
28. Ormsby AH, Vaezi MF, Richter JE, Goldblum JR, Rice TW,
Falk GW, et al. Cytokeratin immunoreactivity patterns in the
diagnosis of short-segment Barrett’s esophagus. Gastroen-
29. Ormsby AH, Goldblum JR, Rice TW, Richter JE, Falk GW,
Vaezi MF, et al. Cytokeratin subsets can reliably distinguish
Barrett’s esophagus from intestinal metaplasia of the stom-
ach. Hum Pathol 1999;30:288–94.
30. Moll R, Lowe A, Laufer J, Franke WW. Cytokeratin 20 in
human carcinomas. A new histodiagnostic marker detected
by monoclonal antibodies. Am J Pathol 1992;140:427–47.
31. Ramaekers F, van Niekerk C, Poels L, Schaafsma E, Huijs-
mans A, Robben H, et al. Use of monoclonal antibodies to
keratin 7 in the differential diagnosis of adenocarcinomas.
Am J Pathol 1990;136:641–55.
32. Glickman JN, Wang H, Das KM, Goyal RK, Spechler SJ, An-
tonioli D, et al. Phenotype of Barrett’s esophagus and intes-
tinal metaplasia of the distal esophagus and gastroesopha-
geal junction: an immunohistochemical study of cytokeratins 7
and 20, Das-1 and 45 MI. Am J Surg Pathol 2001;25:87–94.
616 Modern Pathology