Virchows Arch (2003) 442:271–277
DOI 10.1007/s00428-002-0752-4
ORIGINAL ARTICLE
A. Alkushi · J. Irving · F. Hsu · B. Dupuis · C. L. Liu ·
M. Rijn · C. B. Gilks
Immunoprofile of cervical and endometrial adenocarcinomas
using a tissue microarray
Received: 23 May 2002 / Accepted: 29 October 2002 / Published online: 12 February 2003
Springer-Verlag 2003
Abstract Adenocarcinomas of the uterine cervix show a
wide range of morphological features, and can be
confused with endometrial adenocarcinoma in biopsy or
curetting specimens. The objective of this study was to
use tissue microarray technology to evaluate the immunoprofile of a large set of uterine adenocarcinomas
with an extended panel of antibodies, comparing the
profile of primary cervical and endometrial adenocarcinomas. A tissue microarray was constructed using
paraffin-embedded, formalin-fixed tissues from 141 hysterectomy specimens. Duplicate 0.6-mm cores were
obtained from 57 cervical adenocarcinomas (16 in situ
and 41 invasive) and 84 endometrial adenocarcinomas.
Tissue array sections were immunostained with 21
commercially available antibodies [B72.3, CD 99, carcinoembryonic antigen (CEA), c-kit, pancytokeratin, CK 5/
6, CK 7, CK8/18, CK19, CK 20, CK 22, EMA, estrogen
receptor (ER), KP-1, melan-A, p53, PLAP, S-100,
synaptophysin, TTF-1, and vimentin] utilizing the avidin-biotin (ABC) technique. Hierarchical clustering analysis of the tumors was done based on the immunostaining
results. Only ER (P<0.001), CEA (P=0.04), vimentin
(P<0.001), and CK 8/18 (P=0.002) showed a significantly
different frequency of positivity in endometrial relative to
cervical adenocarcinomas. ER, vimentin, and CK 8/18
were more likely to be expressed in endometrial adenocarcinomas, while cervical adenocarcinomas more freA. Alkushi · J. Irving · F. Hsu · B. Dupuis · C. B. Gilks ())
Department of Pathology and Genetic Pathology Evaluation Center,
Department of Pathology,
1st floor JPPN, Vancouver General Hospital,
855 W. 12th Avenue, Vancouver, BC V5Z 1M9, Canada
e-mail: [email protected]
Tel.: +604-875-5555 ext. 63305
Fax: +604-875-5707
C. L. Liu
Department of Biochemistry, Stanford University,
Stanford, CA
M. Rijn
Department of Pathology,
Stanford University, Stanford, CA
quently expressed CEA. We were able to identify
immunoprofiles that were highly specific for endocervical
adenocarcinoma (ER, vimentin, CK 8/18, CEA+) or
endometrial adenocarcinoma (ER+, vimentin+, CK 8/18+,
CEA), but most tumors showed an intermediate, nonspecific immunophenotype. Hierarchical clustering analysis was useful in the interpretation of these intermediate
immunophenotypes. Papillary serous adenocarcinoma of
the endometrium was less likely to express vimentin
(P=0.002) than endometrioid carcinoma of the endometrium.
Keywords Endometrial carcinoma · Cervical
adenocarcinoma · Immunohistochemistry · Tissue
microarray
Introduction
Adenocarcinoma of the uterine cervix and endometrium
have long been recognized as distinct entities, with
different etiologies, behavior and treatments. There is,
however, considerable overlap in the morphological
features of adenocarcinoma arising at these two sites.
This morphological overlap can make their histological
differentiation a diagnostic problem in biopsy or curetting
specimens, especially with a small biopsy, or when
adenocarcinoma is present in both components of a
fractional dilation and curettage specimen [4, 17, 23].
This distinction has preoperative clinical significance as
the choice of definitive therapy may rest on the site of
origin of the tumor, and it is not uncommon for the preoperative clinical, hysteroscopic, radiologic and pathologic examination to fail in identifying the primary site
with certainty.
Tissue microarrays allow the assessment of hundreds
of tissue cores on a single slide [13]. In this technique,
small tissue cores are retrieved from selected regions of
archival tissue blocks, and hundreds of such samples are
precisely arrayed in a new paraffin block. These tissue
samples can then be analyzed using immunohistochem-
272
the donor block was cored twice with a 0.6-mm diameter needle
and transferred to a recipient paraffin block.
istry, fluorescence in situ hybridization (FISH) or RNA in
situ hybridization. We utilized a tissue microarray with
samples from 141 cases to evaluate the immuoprofile of
uterine adenocarcinomas of cervical and endometrial
origin with a panel of 21 different antibodies. The data
generated was submitted to hierarchical clustering analysis [8], to identify groups of tumors with related
immunoprofiles.
Immunohistochemistry
The avidin-biotin (ABC) method was used for immunostaining.
The recipient block was sectioned at 4 mm, and transferred to
silanized glass slides. Each of these unstained sections was
deparaffinized with xylene, and rehydrated through a series of
graded alcohols. A section from the array was then stained with
haematoxylin and eosin, to assess adequacy. Further sections were
stained with the panel of antibodies listed in Table 2 using a
Ventana (Tuscon AZ) automated immunohistochemical stainer
according to the manufacturer’s guidelines. CEA(p) refers to the
polyclonal anti-carcinoembryonic antigen (CEA) antibody, while
CEA(m) refers to the monoclonal anti-CEA antibody. Antigen
retrieval was done as indicated in Table 2. Microwave antigen
retrieval consisted of placing the slides in 10 mM citrate buffer
(pH 6.0) in a pressure cooker (Nordic Ware), and microwaving on
high power until the buffer had boiled under pressure for 4 min. At
this point, microwaving was stopped and the slides were incubated
in the pressure cooker for a further 20 min, removed and rinsed.
Proteinase antigen retrieval consisted of a 4-min incubation in
protease-1 solution (Ventanna) according the supplier’s recommended protocol. We did not use biotin blocking for this study as
we have determined that with the antigen retrieval protocols used,
in our experience, we do not see endogenous biotin reactivity (seen
as granular cytoplasmic staining) in uterine tumors, in contrast to
our experience with thyroid, renal and hepatocellular neoplasms.
Two pathologists (AA, CBG) scored the immunostaining using
a two-headed microscope. Tumor cell cytoplasmic or nuclear
staining (as appropriate, according to the antigen being detected)
was scored as technically unsatisfactory/uninterpretable, negative
(<5% cells positive), weak positive (5–50% of cells showing weak
or moderately intense positivity), or strong positive (>50% of cells
positive or 5–50% of cells showing intense immunoreactivity).
Scoring results then were simplified into either negative or positive
(combining weak and strong positive results) and unsatisfactory/
uninterpretable results were eliminated from further consideration.
Score results for the duplicate cores were consolidated into one
score/case where positive staining would always supercede either a
negative or uninterpretable result, and a negative result would
supercede an uninterpretable result.
Materials and methods
Study material
The study material consisted of slides and selected tissue blocks
from 141 hysterectomy specimens retrieved from the archives of
the Department of Pathology and Laboratory Medicine, Vancouver
General Hospital (Table 1). These specimens were accessioned
between 1984 and 2000. Fifty-seven were cases of cervical
adenocarcinoma (16 in situ and 41 invasive tumors), and 84 were
endometrial carcinomas. Cervical carcinoma cases were centered in
the endocervical canal, while the endometrial carcinoma cases were
in the uterine fundus. No cases with tumor centered in the lower
uterine segment, or tumor extensively involving both the fundus
and endocervical canal, such that determination of the primary site
of the tumor was problematic, were included in this study. A wide
range of morphological types of cervical adenocarcinoma were
included (endocervical, villoglandular, endometrioid, clear cell,
papillary serous, adenosquamous, glassy cell, adenoid cystic, and
poorly differentiated type). Fifty-four endometrial carcinomas were
of endometrioid type, five with focal mucinous differentiation. Of
these, twenty-five were low-grade (FIGO grade 1) endometrial
carcinomas of endometrioid type, and twenty-nine were high-grade
(FIGO grade 3) endometrial adenocarcinomas of endometrioid
type. The rest of the endometrial carcinomas were papillary serous
carcinoma (n=30). Institutional ethics approval was obtained for
this retrospective immunohistochemical study.
Construction of tissue microarray
All hematoxylin and eosin (H&E) stained slides for these cases
were reviewed, a slide with representative tumor was selected from
each case, and an area of tumor was circled on the slide. The
corresponding formalin-fixed, paraffin-embedded blocks were
retrieved from the hospital archives and the area corresponding to
the selected area on the slide was circled on the block with a felt
marker for tissue microarray construction. All tissues were fixed in
10% neutral buffered formalin. Using a tissue microarrayer
(Beecher Instruments, Silver Spring, MD), the area of interest in
Table 1 Cases used in the tissue microarray
Statistical and clustering analysis
Fisher’s exact and Chi-Square tests were used to analyze 22 tables
generated to compare the immunohistochemical scoring results of
cervical and endometrial adenocarcinomas. Calculations were done
using SPSS, version 9.0 software. Correction for making multiple
comparisons was made, and significance was called at P<0.05, after
correction.
Cervical adenocarcinoma
Endometrial carcinoma
Type
No. of cases
Type/grade
No. of cases
In situ
Endocervical type
Villoglandular
Endometrioid
Clear cell carcinoma
Papillary serous
Intestinal type (mucinous)
Adenosqumous carcinoma
Adenosqumous – glassy cell
Adenoid cystic
Poorly differentiated
Total
16
11
2
4
4
5
1
3
4
1
6
57
Grade-I endometrioid
Grade-III endometrioid
Papillary serous carcinoma
25
29
30
Total
84
273
Table 2 Antibodies used in this
study. N no antigen retrieval, M
microwave, E trypsin digestion
Antigen
Clone
Supplier
Dilution
Antigen retrieval
B72.3
CD-68
CD 99
CEA(m)
CEA(p)
c-kit
Pankeratin
CK 5/6
CK 7
CK 8/18
CK19
CK 20
CK 22
EMA
ER
Melan-A
p53
S-100
Synaptophysin
TTF-1
Vimentin
BRST-3
KP-1
O13
TF 3H8–1
Polyclonal
Polyclonal
Polyclonal
D5/16B4
OV-TL 12/30
CAM 5.2
RCK108
Ks20.8
cocktail
E29
6F11
Dako
DO-7
Polyclonal
Polyclonal
8G7G3/1
V9
Signet
Dako
ID Labs
Ventana
Dako
Santa Cruz
Dako
Zymed
Dako
Becton/Dickinson
Dako
Dako
Biomeda
Dako
Novocastra
A103
Dako
Dr. A Marksa
Dako
Dako
Biogenix
1:10,000
1:400
1:20
1:5
1:10,000
1:100
1:4000
1:100
1:200
1:50
1:500
1:500
1:10
1:200
1:50
1:50
1:400
1:1000
1:200
1:100
1:20,000
N
N
M
N
N
N
E
M
E
E
E
E
E
N
M
M
M
N
M
M
M
a
Dr. A. Marks, Banting and Best Dept. of Medical Research, University of Toronto
Table 3 Immunohistochemical
findings
Pankeratin
CK 8/18
CK7
CK20
CK19
CK22
CK5/6
EMA
CEA(m)
CEA(p)
ER
Vimentin
Synaptophysin
c-kit
Melan-A
CD 99
S-100
B72.3
TTF-1
CD 68
Cervical adenocarcinoma
Endometrial carcinoma
In situ
Invasive
Endometrioid
PSCE
100%
63%
94%
2%
83%
97%
33%
95%
30%
80%
11%
10%
0%
13%
0%
47%
6%
43%
0%
3%
94%
94%
96%
0%
88%
100%
34%
98%
17%
61%
73%
92%
21%
19%
0%
62%
0%
60%
0%
0%
100%
91%
100%
0%
94%
100%
8%
100%
5%
25%
36%
52%
13%
12%
0%
41%
0%
61%
6%
0%
92%
66%
92%
0%
88%
100%
25%
69%
33%
78%
33%
0%
0%
0%
0%
50%
0%
17%
0%
0%
(11/12)
(8/12)
(12/13)
(0/15)
(7/8)
(7/7)
(3/12)
(9/13)
(5/15)
(7/9)
(4/12)
(0/13)
(0/11)
(0/12)
(0/16)
(4/8)
(0/7)
(2/12)
(0/8)
(0/8)
(37/37)
(24/38)
(34/36)
(1/39)
(29/35)
(34/35)
(13/39)
(35/37)
(11/37)
(28/35)
(4/37)
(4/39)
(0/39)
(5/38)
(0/38)
(16/34)
(2/36)
(16/37)
(0/36)
(1/35)
(47/50)
(48/51)
(51/53)
(0/52)
(42/48)
(49/49)
(18/53)
(52/53)
(9/52)
(30/49)
(37/51)
(48/53)
(11/52)
(10/53)
(0/52)
(31/50)
(0/49)
(32/53)
(0/49)
(0/50)
P value*
(23/23)
(20/22)
(24/24)
(0/23)
(17/18)
(20/20)
(2/24)
(25/25)
(1/22)
(5/20)
(8/22)
(12/23)
(3/23)
(3/25)
(0/22)
(9/22)
(0/17)
(14/23)
(1/18)
(0/20)
0.002
0.04
<0.001
<0.001
*P value comparing the immunostaian result in cervical and endometrial adenocarcinomas, P<0.05 is
considered significant. Differences are not statistically significant unless indicated
Hierarchal clustering analysis of our tissue microarray data was
performed using software tools that were originally designed for
analyzing cDNA microarray data [8]. An Excel macro, TMADeconvoluter [14], was designed and written specifically for
processing of raw tissue microarray staining data into a format
compatible for use with the previously developed Cluster software
[8]. The clustered data was then graphically viewed using
TreeView (Cluster and Treeview software are freely available
programs that can be accessed at http://rana.lbl.gov/EisenSoftware.htm).
Results
Table 3 summarizes the immunohistochemical findings.
For most antibodies tested there was no significant
difference in immunostaining between cervical and
endometrial carcinomas; however, for four antigens
[estrogen receptor (ER) (P<0.001), vimentin (P<0.001),
CK 8/18 (P=0.002), CEA (P=0.04 for the polyclonal
antibody, and P not significant for the monoclonal
antibody)], there was a significant difference in the
frequency of immunostaining. The first three antigens
were more frequently expressed in endometrial adeno-
274
Table 4 Cervical adenocarcinoma subtypes
Cervical adenocarcinoma
CEA(p)
ER
Vimentin
CK 8/18
Endometrial adenocarcinoma, endometrioid type
Endocervical type
Other type
100%
10%
8%
63%
70%
11%
11%
60%
(12/12)
(1/10)
(1/12)
(7/11)
(16/23)
(3/27)
(3/27)
(16/27)
61%
73%
92%
94%
(30/49)
(37/51)
(48/52)
(48/51)
Table 5 Cervical versus endometrial adenocarcinoma immunoprofiles
Cervical
adenocarcinoma
Cervical carcinoma pattern (ER, Vimentin, CEA(p)+)
64% (21/33)
Endometrial carcinoma pattern (ER+, Vimentin+, CEA(p))
0% (0/33)
Extended cervical carcinoma pattern (ER, vimentin, CK 8/18, CEA(p)+)
21% (7/33)
Extended endometrial carcinoma pattern (ER+, vimentin+, CK 8/18+, CEA(p)) 0% (0/33)
Endometrioid
endometrial carcinoma
P value
4%
23%
2%
21%
<0.001
0.002
0.005
0.005
(2/47)
(11/47)
(1/47)
(10/47)
Table 6 Endometrioid versus papillary serous endometrial adenocarcinoma
Endometrial carcinoma
Estrogen receptor
Vimentin
P value*
Grade-I endometrioid
Grade-III endometrioid
Papillary serous carcinoma
76% (19/25)
88% (22/25)
69% (18/26)
92% (26/27)
36% (8/22)
52% (12/23)
0.08
0.002
*P value comparing the immunostaining results for endometrioid endometrial carcinoma and papillary serous carcinoma of endometrium
carcinomas than in cervical adenocarcinomas, whereas
CEA showed the opposite pattern.
Table 4 shows the frequency of CK 8/18, ER,
vimentin, and CEA(p) expression in different subtypes of
invasive cervical adenocarcinoma. All endocervical (mucinous) type cervical adenocarcinomas expressed CEA as
detected with the polyclonal antibody (CEA(p)) and 70%
of the other cervical adenocarcinoma subtypes expressed
this protein. CEA was also present in more than half of
the endometrial adenocarcinomas. The staining of endocervical type cervical adenocarcinomas was indistinguishable from other subtypes of cervical adenocarcinoma with
respect to ER, vimentin, and CK 8/18 immunostaining.
As the combination of ER, CEA and vimentin staining
has previously been shown to aid in differentiation
between cervical and endometrial adenocarcinoma [2],
we determined how many tumors showed a typical
cervical adenocarcinoma staining profile (ER, vimentin,
and CEA(p)+) or a typical endometrial adenocarcinoma
staining profile (ER+, vimentin+, and CEA(p)) (Table 5).
For the purposes of this analysis, we excluded cervical
papillary serous carcinomas, which are rare in the cervix.
While most cervical adenocarcinomas were ER, vimentin, and CEA(p)+ (21 of 33, 64%), occasional endometrial
adenocarcinomas also showed this staining pattern (2 of
47, 4%), so that this immunoprofile has a sensitivity and
specificity of 64% and 91%, respectively, for the
diagnosis of cervical adenocarcinoma. No cervical adenocarcinomas showed a staining pattern for these three
antigens typical of endometrial adenocarcinoma, relative
to 11 of 47 (23%) endometrial adenocarcinomas. Thus,
this immunoprofile (ER+, vimentin+, and CEA(p)) has a
sensitivity of 23% and specificity of 100% for the
diagnosis of endometrial adenocarcinoma.
Our analysis identified one novel discriminating
marker in the differential diagnosis between endometrial
and cervical adenocaricnomas, CK 8/18, and this antigen
was then added to the immunoprofile for distinguishing
between endometrial and cervical adenocarcinomas. The
extended cervical-type pattern (ER, vimentin, CK 8/
18, CEA(p)) has a sensitivity of 21% and specificity of
87.5% for the diagnosis of cervical adenocarcinoma. The
endometrial-type pattern (ER+, vimentin+, CK 8/18+,
CEA(p)) has a sensitivity of 21% and specificity 100%
for the diagnosis of endometrial adenocarcinoma. With
either the limited (three antibodies) or extended (four
antibodies) immunoprofiles, a majority of tumors do not
show either a cervical-type or endometrial-type pattern,
but instead have an intermediate immunophenotype, not
characteristic of either primary site.
The result of hierarchical clustering analysis with the
four antibodies showing a difference in staining between
endometrial and endocervical adenocarcinoma is shown
in Fig. 1. This shows that although cases of cervical and
endometrial adenocarcinoma tend to cluster on separate
branches of the dendrogram, there is considerable overlap, with some cases of cervical adenocarcinoma sharing
the same immunophenotype as endometrial adenocarcinomas.
275
One antibody (vimentin) showed a significant difference in frequency of positivity when endometrioid and
papillary serous carcinomas of the endometrium were
compared (Table 6). ER was also expressed less frequently in papillary serous carcinomas than endometrioid
carcinomas, but this difference was not statistically
significant.
Discussion
The preoperative distinction of cervical and endometrial
adenocarcinomas is clinically important as stage-1 “low
risk” endometrial adenocarcinoma can be treated by
simple hysterectomy alone, while the treatment of
cervical adenocarcinoma is generally by radiotherapy
with or without subsequent hysterectomy [23]. We took
advantage of recently described tissue microarray technology [13] to study a large number of cases of cervical
and endometrial adenocarcinoma with multiple antibodies. Although only small amounts of the tumor are
sampled in tissue microarrays, there is now convincing
evidence that immunohistochemical staining of two or
more cores/case gives results that are highly representative of whole sections of tumor [3, 18, 19, 22].
The earliest identified immunophenotypic difference
between cervical and endometrial adenocarcinoma was
the increased frequency and intensity of CEA immunoreactivity in cervical adenocarcinomas [5], an observation
since confirmed in multiple studies [4, 7, 15, 17]. These
studies have given a wide range of results, with 50–100%
of cervical adenocarcinomas reported to be CEA positive,
compared with 0–50% of endometrial adenocarcinoma
studies [4, 5, 7, 15, 17]. Differences in staining techniques, antibodies (in particular monoclonal versus polyclonal anti-CEA antibodies), and case selection
presumably account for the wide range of reported
results. We found that the polyclonal anti-CEA antiserum
was more discriminatory than the monoclonal antibody,
with 35 of 44 (80%) cervical adenocarcinomas versus 35
of 69 (51%) endometrial adenocarcinomas staining positively with CEA(p). Of note, CEA was expressed by all
endocervical (mucinous) type cervical adenocarcinomas
relative to 70% of other types of cervical adenocarcinomas. This reduces the value of CEA immunostaining as a
marker of cervical adenocarcinoma since these other
histological subtypes of cervical adenocarcinoma are
more difficult to differentiate from endometrial carcinoma, and would be the cases where one would be most
likely to resort to immunostaining as a diagnostic aid.
Vimentin has also been consistently demonstrated to
be differentially expressed in endometrial and cervical
adenocarcinomas, being present in most endometrial
Fig. 1 Graphical representation of the hierarchical clustering
analysis. Positive staining is indicated by a red cube, absence of
staining as green, and no available data as gray for each of the
antibodies indicated at the top of the figure. Each case is listed at
the right of the figure
276
adenocarcinomas and in a much smaller percentage of
cervical adenocarcinomas [2, 4, 6, 7, 17]. Our findings are
in agreement with these previous studies, with vimentin
immunoreactivity in only 4 of 52 (8%) cervical adenocarcinomas, including only one tumor of endocervical
type, relative to 60 of 76 (79%) endometrial adenocarcinomas, including 92% of endometrioid endometrial
carcinomas. Thus, vimentin expression is most consistently negative in typical endocervical-type cervical
adenocarcinomas and positive in endometrioid-type endometrial adenocarcinomas, with unusual subtypes more
likely to give an "aberrant" staining pattern. Staining for
ER protein has also been found to be of help in
distinguishing between cervical and endometrial adenocarcinomas, with 70–90% of endometrial adenocarcinomas being ER+, compared with 10–40% of cervical
adenocarcinomas [9, 10, 16, 17]. We found 8 of 49 (16%)
cervical adenocarcinomas to be ER+, compared with 45 of
73 (62%) endometrial carcinomas. Our relatively low
figure for the endometrial adenocarcinomas most probably reflects the inclusion of a relatively high proportion of
papillary serous and high-grade endometrioid carcinomas
in this study, as they are ER+ in only 36% and 69% of
cases, respectively. We identified one additional antibody
that may help in differentiation between cervical and
endometrial adenocarcinoma; cytokeratin 8/18 is more
frequently expressed in endometrial than cervical adenocarcinomas.
McCluggage et al. suggested that a panel of ER, CEA
and vimentin will allow distinction between cervical and
endometrial adenocarcinoma, based on their study of 56
cases [17]. We found that this panel of three antibodies
had a sensitivity and specificity of 64% and 96%,
respectively, for the diagnosis of cervical adenocarcinoma, and a sensitivity of 23% and specificity of 100% for
the diagnosis of endometrial adenocarcinomas. Thus,
most tumors do not conform exactly to a “typical”
immunophenotypic profile of either cervical or endometrial adenocarcinoma, with rare endometrial adenocarcinomas having an immunophenotype characteristic of
cervical adenocarcinoma, i.e., ER, vimentin and CEA+.
When we added CK 8/18 to the profile, the specificities of
endometrial adenocarcinoma staining pattern (CEA(p) /
ER+/vimentin+/CK 8/18+) or cervical adenocarcinoma
staining pattern (CEA(p)+/ER/vimentin/CK 8/18) were
not significantly improved, while there was a drop in
sensitivity. Thus, although CK 8/18 was differentially
expressed in cervical and endometrial adenocarcinomas,
it did not help to define immunoprofiles characteristic of
cervical and endometrial adenocarcinomas. Even with the
more limited panel of CEA, ER and vimentin, most
tumors have an immunophenotype that is intermediate
between the typical patterns of cervical and endometrial
adenocarcinomas.
Although our primary objective was to compare
cervical and endometrial adenocarcinomas, other novel
observations emerged as a result of the broad screening
approach using many antibodies and a large number of
cases. Comparing papillary serous carcinoma to endome-
trioid carcinoma revealed that vimentin is less frequently
expressed in papillary serous adenocarcinoma than in
endometrioid type endometrial adenocarcinoma. These
two types of endometrial carcinoma represent the prototype of types 2 and 1 of endometrial carcinoma,
respectively, and these differences presumably reflect
differences in molecular events during oncogenesis for
these tumors. Halperin et al. found that ER is less
frequently expressed in papillary serous carcinoma than
endometrial adenocarcinoma of endometrioid type [12].
Although we observed a difference in the frequency of ER
immunoreactivity between these two types of tumors, the
difference was not statistically significant (P=0.08). This
result does highlight one aspect of studies such as this
one, using multiple immunohistochemical markers. If we
had studied only ER, the difference we observed with 37
of 51 (73%) endometrioid adenocarcinomas and 8 of 22
(36%) papillary serous carcinomas staining positively for
ER would have been highly significant (P=0.004), but this
effect is lost in the present study when correction is made
for multiple comparisons.
Application of hierarchical clustering analysis allowed
us to graphically demonstrate the relatedness of cervical
and endometrial adenocarcinoma immunoprofiles. Hierarchical clustering analysis is a method of dealing with
complex sets of data and offers a possible systematic
approach to complex immunophenotypes. Clustering
analysis has been successfully applied to gene-expression
data, based on the expression of thousands of genes, and
has been remarkably successful in its ability to group
tumors according to their primary site based on gene
expression profiles [1, 11, 20, 21]. The dynamic range of
immunostaining scores is significantly narrower than that
obtained for mRNA levels in gene microarray experiments, however, and, as a result, the clustering of tumors
based on immunoreactivity can be expected to be less
well defined than is seen in gene expression array studies.
When individual tumors are stained with multiple antibodies, a complex immunophenotype is generated that in
typical cases is readily interpretable but, as shown
previously, is more often than not atypical for at least
some of the markers used. For example, what is the
significance of a tumor in a D&C specimen with the
immunoprofile of CEA(p)/ER/vimentin+/CK 8/18+? By
reviewing the graphical display of the clustering analysis
(Fig. 1) we can see that this immunophenotype was seen
in ten tumors, including six endometrioid-type endometrial adenocarcinomas, three papillary serous adenocarcinomas of endometrium and one cervical adenocarcinoma.
It thus is not a specific immunoprofile but is more
suggestive of an endometrial rather than endocervical
primary site. This application of clustering analysis
allows a more objective interpretation of immunoprofiles
based on staining with multiple antibodies and holds great
promise for the immunohistochemical subclassification of
tumors according to primary site; however, there is
clearly a need for better, more tissue-specific, antigen
targets.
277
References
1. Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS,
Rosenwald A, Boldrick JC, Sabet H, Tran T, Yu X, Powell
JI, Yang L, Marti GE, Moore T, Hudson J, Lu L, Lewis DB,
Tibshirani R, Sherlock G, Chan WC, Greiner TC, Weisenburger DD, Armitage JO, Warnke R, Staudt LM, et al (2000)
Distinct types of diffuse large B-cell lymphoma identified by
gene expression profiling. Nature 403:503–511
2. Azumi N, Battifora H (1987) The distribution of vimentin and
keratin in epithelial and non-epithelial neoplasms. A comprehensive immunohistochemical study on formalin- and alcoholfixed tumors. Am J Clin Pathol 88:286–296
3. Camp RL, Charette LA, Rimm DL (2000) Validation of tissue
microarray technology in breast carcinoma. Lab Invest
80:1943–1949
4. Castrillon DH, Lee KR, Nucci MR (2002) Distinction between
endometrial and endocervical adenocarcinoma: an immunohistochemical study. Int J Gynecol Pathol 21:4–10
5. Cohen C, Shulman G, Budgeon LR (1982) Endocervical and
endometrial adenocarcinoma: an immunoperoxidase and histochemical study. Am J Surg Pathol 6:151–157
6. Dabbs DJ, Geisinger KR, Norris HT (1986) Intermediate
filaments in endometrial and endocervical carcinomas. The
diagnostic utility of vimentin patterns. Am J Surg Pathol
10:568–576
7. Dabbs DJ, Sturtz K, Zaino RJ (1996) The immunohistochemical discrimination of endometrioid adenocarcinomas. Hum
Pathol 27:172–177
8. Eisen MB, Spellman PT, Brown PO, Botstein D (1998) Cluster
analysis and display of genome-wide expression patterns. Proc
Natl Acad Sci U S A 95:14863–14868
9. Fujiwara H, Tortolero-Luna G, Mitchell MF, Koulos JP, Wright
TC Jr. (1997) Adenocarcinoma of the cervix. Expression and
clinical significance of estrogen and progesterone receptors.
Cancer 79:505–512
10. Ghandour FA, Attanoos R, Nahar K, Gee JW, Bigrigg A, Ismail
SM (1994) Immunocytochemical localization of oestrogen and
progesterone receptors in primary adenocarcinoma of the
cervix. Histopathology 24:49–55
11. Golub T R, Slonim D K, Tamayo P, Huard C, Gaasenbeek M,
Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri MA,
Bloomfield CD, Lander ES (1999) Molecular classification of
cancer: class discovery and class prediction by gene expression
monitoring. Science 286:531–537
12. Halperin R, Zehavi S, Habler L, Hadas E, Bukowsky I,
Schneider D (2001) Comparative immunohistochemical study
of endometrioid and serous papillary carcinoma of the endometrium. Eur J Gynaecol Oncol 22:122–126
13. Kononen J, Bubendorf L, Kallioniemi A, Barlund M, Schraml
P, Leighton S, Torhorst J, Mihatsch MJ, Sauter G, Kallioniemi
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
OP (1998) Tissue microarrays for high-throughput molecular
profiling of tumor specimens. Nat Med 4:844–847
Liu CL, Prapong W, Natkuman Y, Alizadeh A, Montgomery K,
Gilks CB, van de Rijn M (2002) Software tools for highthroughput analysis and archiving of immunohistochemistry
staining data obtained with tissue microarrays. Am J Pathol
161:1557–1565
Maes G, Fleuren G, Bara J, Nap M (1988) The distribution of
mucins, carcinoembryonic antigen, and mucus-associated antigens in endocervical and endometrial adenocarcinomas. Int J
Gynecol Pathol 7:112–122
Masood S, Rhatigan RM, Wilkinson EW, Barwick KW, Wilson
WJ (1993) Expression and prognostic significance of estrogen
and progesterone receptors in adenocarcinoma of the uterine
cervix. An immunocytochemical study. Cancer 72:511–518
McCluggage WG, Sumathi VP, McBride HA, Patterson A
(2002) A panel of immunohistochemical stains, including
carcinoembryonic antigen, vimentin, and estrogen receptor,
aids the distinction between primary endometrial and endocervical adenocarcinomas. Int J Gynecol Pathol 21:11–15
Nocito A, Bubendorf NA, Maria Tinner E, Suess K, Wagner U,
Forster T, Kononen J, Fijan A, Bruderer J, Schmid U,
Ackermann D, Maurer R, Alund G, Knonagel H, Rist M,
Anabitarte M, Hering F, Hardmeier T, Schoenberger AJ, Flury
R, Jager P, Luc Fehr J, Schraml P, Moch H, Mihatsch MJ,
Gasser T, Sauter G (2001) Microarrays of bladder cancer tissue
are highly representative of proliferative index and histological
grade. J Pathol 194:349–357
Parker R, Cupples J, Lesack D, Grant D, Huntsman D, Gilks
CB (2002) Assessment of inter observer variation in the
immunohistochemical determination of estrogen receptor status
using a breast cancer tissue microarray. Am J Clin Pathol
117:723–728
Perou CM, Sorlie T, Eisen MB, van de Rijn M, Jeffrey SS, Rees
CA, Pollack JR, Ross DT, Johnsen H, Akslen LA, Fluge O,
Pergamenschikov A, Williams C, Zhu SX, Lonning PE,
Borresen-Dale AL, Brown PO, Botstein D (2000) Molecular
portraits of human breast tumours. Nature 406:747–752
Sorlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H,
Hastie T, Eisen MB, van De Rijn M, Jeffrey SS, Thorsen T,
Quist H, Matese JC, Brown PO, Botstein D, Eystein Lonning P,
Borreson-Dale AL (2001) Gene expression patterns of breast
carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci U S A 98:10869–10874
Torhorst J, Bucher C, Kononen J, Haas P, Zuber M, Kochli OR,
Mross F, Dietrich H, Moch H, Mihatsch M, Kallioniemi OP,
Sauter G (2001) Tissue microarrays for rapid linking of
molecular changes to clinical endpoints. Am J Pathol
159:2249–2256
Zaino RJ (2002) The fruits of our labors: Distinguishing
endometrial from endocervical adenocarcinoma. Int J Gynecol
Pathol 21:1–3