ANATOMIC
PATHOLOGY
Original
R e d u c e d
E - C a d h e r i n
A s s o c i a t e d
U n f a v o r a b l e
Article
W i t h
E x p r e s s i o n
I n v a s i v e n e s s
P r o g n o s i s
i n
B r e a s t
Is
a n d
C a n c e r
SANNA M. SIITONEN, MD,1 JUHA T. KONONEN, MD, 4 HEIKKI J. HELIN, MD,2
IMMO S. RANTALA, PHD, 2 KAIJA A. HOLLI, MD, 3 AND JORMA J. ISOLA, MD4
E-cadherin (E-cad) is a calcium-dependent, epithelial cell adhesion
molecule whose reduced or lost expression has been associated with
tumor dedifferentiation and increased metastatic potential in human
carcinomas. The authors studied immunohistochemically E-cad expression in frozen sections of 362 breast carcinomas using a monoclonal
antibody (HECD-1). The immunohistochemical detection of reduced Ecad expression was confirmed by mRNA in situ hybridization with two
different oligonucleotide probes. The proportion of tumors with reduced
or lost E-cad expression increased significantly from pure intraductal
carcinomas (20%, 4 of 20) through invasive ductal (IDCs; 52%, 124 of
239) to recurrent carcinomas (64%, 18 of 28; chi square test for trend,
P = .004). Invasive lobular carcinomas (ILCs) and IDCs differed from
each other in their E-cad expression. None of the ILCs (n = 55) retained
normal E-cad expression in contrast to 48% (115 of 239) of the IDCs.
In 259 primary IDCs, reduced E-cad expression was associated with
Cadherins are a family of transmembrane glycoproteins
that are involved in homotypic calcium-dependent intercellular adhesion.1 They play an important role during embryonic development and in the maintenance of
adult tissue architecture: inactivation of other adhesion
systems has little effect on cell-cell adhesion as long as
cadherins are functioning.1
Several types of cadherins have been identified. Ecadherin (E-cad) has a major role in the maintenance of
intercellular junctions between epithelial cells in most
organs.2,3 E-cad is predominantly localized to the lateral
cell borders at zonula adherens junctions either in a polarized or nonpolarized manner. 2 It interacts with the
high histologic grade (chi square test for trend, P <.001), negative estrogen receptor status (ER; Fisher's exact test, P = .042), and marginally with axillary node involvement (Fisher's exact test, P = .063). In a
subset of 109 primary IDC patients whose clinical follow-up was available (median follow-up 51 months), reduced E-cad expression was associated with shortened disease-free survival (DFS; Mantel-Cox test, P
= .027). In Cox's multivariate regression analysis, progesterone receptor status (P = .018) and E-cad expression (P = .072) were selected as
independent predictors of DFS.
Our findings provide clinical evidence that loss of normal E-cad expression is an indicator of increased invasiveness and dedifferentiation
in breast carcinoma. E-cad is a potentially important prognostic factor
in primary IDCs. (Key words: Cell-cell adhesion; Malignancy; Metastasis; Breast cancer; Patient outcome) Am J Clin Pathol 1996; 105:394402.
contractile cytoskeleton via associated cytoplasmic proteins called catenins. 4
E-cadherin is of particular interest in cancer biology
because loss of E-cad function facilitates cancer cell invasion.5"7 Reduced E-cad expression has been shown to
correlate with tumor dedifferentiation8"19 and the presence of lymph node metastases 9 "" 1 3 ' 8 1 9 in a wide variety of human malignancies. It has also been associated
with poorer disease outcome in squamous cell carcinoma of head and neck,15 prostate adenocarcinoma,"
and bladder carcinoma. 19
In breast carcinoma, three of the six studies published
so far have reported an inverse relationship between Ecad expression and histologic grade of the tumors. 16 " 18 A
correlation has also been reported between E-cad expresFrom the department of Clinical Chemistry, ^Pathology Unit, sion
and and axillary node involvement, again with variable
3
Department ofRadiotherapy and Oncology, Tampere University Hosfindings.1718'20 Oka and colleagues18 found that reduced
pital, Tampere, Finland; ^Institute of Medical Technology, University
E-cad expression was associated with axillary node inof Tampere, Tampere, Finland.
volvement, whereas Gamallo and associates17 and LipSupported by a grant from the Finnish Cancer Society, Pirkanmaa
ponen and coworkers20 could not confirm this. In short,
Cancer Society and the Pirkanmaa Cultural Foundation.
the role of reduced E-cad expression in clinical breast
Manuscript received August 1, 1995; revision accepted October 18,
1995.
cancer remains unclear. The findings of all studies with
Address reprint requests to Dr. Siitonen: Department of Clinical
sufficient numbers of invasive lobular carcinomas (ILCs)
Chemistry, Tampere University Hospital, P.O. Box 2000, FIN-33521,
agree that these differ substantially from invasive ductal
Tampere, Finland.
394
SIITONEN ET AL.
E-Cadherin Expression in Breast Cancer
carcinomas (IDCs) in their E-cad expression: 161720,21
ILCs always seem to have reduced E-cad expression.
We studied E-cad expression in frozen sections of 362
breast carcinomas with a carefully validated immunohistochemical method. The purpose was to examine the relationship between E-cad expression and various pathobiologic tumor features in a larger series of patients
and to evaluate the prognostic value of E-cad expression
in primary invasive ductal breast carcinoma.
MATERIALS AND METHODS
Patients and Tumors
The study sample included 362 breast carcinoma patients presenting between 1986 and 1994 of whom a frozen tumor section was available. The frozen sections
were obtained from the tumor bank of the Institute of
Medical Technology, University of Tampere, to which
the samples were sent from the Tampere University Hospital, Tampere City Hospital, and from Mikkeli, KantaHame and Paijat-Hame Central Hospitals for immunohistochemical estrogen and progesterone receptor and cerbB-2 oncoprotein assays. Vast majority of the cases
(334) were primary carcinomas and 28 were recurrent
tumors. The age of the patients ranged from 32 to 92
years (mean 60 years). Clinical follow-up for more than
15 months was available for 124 consecutive patients
who were treated at the Department of Radiotherapy
and Oncology, Tampere University Hospital. Of these
124 patients, 109 with primary IDC were included in the
survival analyses. Fifty-seven of the 109 patients had
stage I (52%), 40 had stage II (37%), 7 had stage III (6%),
and 5 stage IV (5%) disease at presentation. The median
follow-up period of patients was 51 months (range 17-90
months). Of these 109 IDC cases, 52 received radiation
therapy (40 patients with positive and 12 with negative
lymph nodes) and 33 systemic adjuvant therapy (30 hormonal and 3 cytotoxic therapies). The surgical biopsy
specimens were frozen in liquid nitrogen within 15 minutes of removal. Cryostat sections (5-6 /an) were prepared for intraoperative histopathologic diagnosis, and
additional sections were cut for immunohistochemical
and DNA flow cytometric studies.
Primary tumor size and axillary node involvement
were determined according to the tumor-node-metastasis (TNM) classification. Data on primary tumor size
were available in 286 cases. Forty-five percent of the tumors (128 of 286) were 2 cm or less in diameter, 48%
(136 of 286) were more than 2 cm but less than 5 cm in
diameter, and 7% (20 of 286) were more than 5 cm in
diameter. Data on axillary node involvement were available in 327 cases. The histopathologic diagnosis was
395
made on the basis of the World Health Organization
guidelines.22 For the evaluation of possible differences in
E-cad expression between different histologic types of
breast carcinoma, 55 pure invasive lobular and 20 pure
intraductal carcinomas were included in the study
group. The remaining 259 primary tumors (76%) were
invasive ductal carcinomas, including 20 ductal variants
(6 mucinous, 5 tubular, 5 medullary, 2 cribriform, 1 epidermoid, and 1 apocrine carcinoma). The carcinomas
were grouped into three categories on the basis of the tubular arrangement of cancer cells, nuclear atypia, and
the frequency of mitotic or hyperchromatic nuclear figures according to Bloom and Richardson. 23 The histopathologic assessment was first done by the local pathologist and the sections were then reviewed by one of our
pathologists (HJH).
Immunohistochemical
Staining for E-cadherin
The cryostat sections were fixed in absolute acetone at
room temperature for 10 minutes and air dried. After
washing the slides in phosphate-buffered saline (PBS)
and blocking nonspecific antibody binding with PBS
containing 5% (w/v) bovine serum albumin (Sigma
Chemical, St. Louis, MO) and 5% (w/v) powdered milk,
the immunostaining was done overnight at +4 °C with
the monoclonal mouse anti-E-cad antibody (clone
HECD-1, subclass IgGi 3 ; Zymed Laboratories, San
Francisco, CA) at a dilution 1:800 from the stock. Bound
primary antibody was visualized by the streptavidin-biotin-peroxidase technique (Vectastain ABC Kit, Vector
Laboratories, Burlingame, CA) according to the manufacturer's instructions. Diaminobenzidine (BDH Laboratory Supplies, Poole, England) was used as a chromogen. Peroxidase reaction was intensified according to
Peacock and colleagues.24 Finally, the sections were
counterstained with Harris's hematoxylin. Ductal epithelium in unaffected breast tissue an E-cad positive
breast cancer cell line BT-474 were used as positive controls;25 E-cad negative breast cancer cell line MDA-MB436 was used as a negative control. 25
The staining results were classified semi-quantitatively
into three categories according to the staining intensity
of E-cad positive cells. The intensity of the E-cad staining
of cancer cells was first compared with that on benign
mammary epithelium in the section. If E-cad staining
in tumor cells was uniform and equal to that in normal
epithelium, the result was defined as E-cad positive (++).
When the staining of cancer cells was clearly recognizable but weaker than that of normal epithelium and/or
the staining pattern was heterogenous (with > 10% negative areas), the result was defined as reduced E-cad expression (+). When E-cad expression was completely
Vol. 105-No. 4
396
ANATOMIC PATHOLOGY
Article
lost, the result was classified as E-cad negative (-). E-cad
staining was evaluated by one investigator (SMS) without prior knowledge of the tumors' biopathologic features.
To assess the reproducibility of the E-cad classifications, a second investigator (JJI) repeated the grading in
50 cases. The K coefficient for interobserver association
between repeated E-cad classifications was 0.783. The K
coefficient for intraobserver association (0.879; n = 105)
indicated good intraobserver reproducibility also.
E-cadherin mRNA In Situ Hybridization
Messenger RNA in situ hybridization was carried out
according to Dagerlind and colleagues.26 Briefly, unfixed, air-dried 5-7 ^m frozen sections were hybridized
with an oligonucleotide probe complementary to the
mRNA sequence encoding human E-cad (nucleotides
1058-1102, sequence derived from Rimm and colleagues27). The probe was labeled at the 3'-end with 33 Plabeled a-dATP (DuPont-NEN Research Products, Boston, MA) using terminal deoxynucleotidyltransferase
(Amersham International, Buckinghamshire, UK). Seventy-five nL of the hybridization mixture described by
Dagerlind and associates26 with 1 X 107 cpm/mL of labeled probe was applied onto each slide. The sections
were hybridized for 18 hours in a humidified chamber at
42 °C. After hybridization, the sections were washed four
times (15 minutes each) at 55 °C in 1 X SSC and while in
a final rinse left to cool to room temperature. The sections were then rinsed in distilled water, dehydrated with
60% and 95% ethanol (30 seconds each) and air dried.
The slides were dipped into Kodak NTB 2 liquid photographic emulsion (Kodak, Rochester, NY) diluted 1:1
with distilled water, exposed for 3 weeks, and developed
according to Dagerlind and colleagues.26 The sections
were then counterstained with Meyer's hematoxylin, dehydrated, and mounted. The autoradiographic signal
was visualized using a Nikon Microphot-FX microscope
using a combination of transmitted and polarized light
epiluminescence. A hybridization signal clearly exceeding the level of background labeling was considered indicative of E-cad mRNA expression.
Controls included the use of another oligonucleotide
probe derived from the E-cad gene sequence (nucleotides
271-315). Hybridization results with these two E-cad
probes were identical. Addition of a 100-fold molar excess of respective unlabeled probe abolished the hybridization signals of both probes. Using Northern blot analysis, the oligonucleotide probe recognized the putative
4.8 kb band corresponding to the human E-cad mRNA.
Finally, the mRNA hybridization was validated with E-
cad expressing and nonexpressing breast cancer cell lines
BT-474 and MDA-MB-436, 25 respectively.
Assays for Other Prognostic Factors
DNA flow cytometry was performed from adjacent
frozen 200 /urn sections as described previously.28 Analysis was carried out using an EPICS C flow cytometer
(Coulter Electronics, Hialeah, FL) for data acquisition
and the MultiCycle program (Phoenix Flow Systems,
San Diego, CA) for data analysis. In DNA aneuploid histograms, the S-phase was analyzed from the aneuploid
clone only. Cell cycle evaluation could be done in 85%
of the tumors. Estrogen (ER) and progesterone (PR) receptors were detected immunohistochemically on cryostat sections as described previously.29 The staining
results were evaluated semi-quantitatively, and a histoscore greater than or equal to 100 was considered positive for both ER and PR. 29 Over-expression of c-erbB2
oncoprotein was verified immunohistochemically from
air-dried, Carnoy-fixed cryostat sections according to
Kallioniemi and colleagues.30 Only strong uniform immunostaining of the cell membranes was scored as positive. 30
Statistical
Methods
Contingency tables were analyzed with Fisher's exact
test and the chi square test for trend. P values were twotailed. Interobserver and intraobserver reproducibility of
the E-cad classifications were compared using the K statistics (BMDP4F 31 ). Univariate survival analysis of disease-free survival was performed using the BMDP1L
program and the Mantel-Cox test.31 Multivariate survival analysis was done using the BMDP2L program and
the Cox regression model. 31
RESULTS
Immunohistochemistry and mRNA In Situ
Hybridization of E-cad Expression
Histologically normal lobular and ductal epithelium
showed uniformly strong E-cadherin immunostaining
on cell-cell boundaries (Figs. 1A and ID). Of the 334 primary tumors examined, 148 (44%) were E-cadherin positive, 129 (39%) had reduced E-cad expression and 56
(17%) were negative. Examples of unchanged (comparable to normal ductal epithelium), reduced and negative
E-cad immunostainings are shown in Figure 1.
Messenger RNA in situ hybridization was performed
in eight cases. Three of these were immunohistochemically evaluated as E-cad positive, three showed reduced
E-cad expression, and two were E-cad negative. E-cad-
A.J.C.P.- •il 1996
SIITONEN ET AL.
397
E-Cadherin Expression in Breast Cancer
<
>
' . # *
•
'
•
&
FIG. 1. Immunohistochemical staining of breast carcinomas with a monoclonal antibody (HECD-1) to the E-cadherin. A, Invasive ductal carcinoma
with positive E-cadherin expression. Arrowhead indicates normal, positively stained mammary epithelium. B, Invasive ductal carcinoma with Ecadherin positive intraductal component and reduced E-cadherin expression in invasive component. C, Invasive ductal carcinoma with reduced Ecadherin expression due to heterogenous staining pattern. D, Invasive lobular carcinoma with E-cadherin negative carcinoma cells and positively
stained normal ductal epithelium. (Harris's hematoxylin counterstain; magnification, X320).
positive tumors also showed strong mRNA expression
as detected by dense autoradiographic labeling (Fig. 2A).
The amount of E-cad mRNA was also clearly visible in
normal ducts of otherwise negative ILCs. In five cases
where E-cad expression was classified as reduced or negative on the basis of immunohistochemistry, the amount
of mRNA was also reduced (Fig. 2B).
Correlation of E-cadherin Expression With
Pathobiologic Features of the Tumors
Twenty percent (4 of 20) of the pure intraductal carcinomas had abnormal E-cad expression. The proportion
of tumors with reduced or negative E-cad expression was
52% (124 of 239) in IDCs and 64% (18 of 28) in recurrent
ductal carcinomas. The trend toward abnormal E-cad
expression is statistically significant (chi square test for
trend, P = .004; Table 1). Among these 239 IDCs, there
were 43 tumors with both intraductal and invasive components in the specimen. In 15 of them (35%), E-cad expression was normal or reduced in the intraductal component and reduced or negative in the invasive
component. Most of the variants of invasive ductal carcinomas (85%; 17 of 20) were uniformly E-cad positive
(Table 1). None of the invasive lobular carcinomas (n =
55) had retained normal E-cad expression (Fig. ID; Table 1). When weak E-cad immunostaining was detected
on ILC cells, it often showed an abnormal staining pattern with diffuse cytoplasmic activity.
The association of E-cad expression with various pathobiologic features of the tumors was tested in 259 primary invasive ductal carcinomas (variant forms of IDC
included). For subsequent analyses, the groups with reduced and negative E-cad expression were combined be-
Vol. 105.No. 4
398
ANATOMIC PATHOLOGY
Article
FIG. 2. Messenger RNA in situ hybridization with a 33P-labeled oligonucleotide probe complementary to human E-cadherin. The grains in
the photographic emulsion are seen white in polarized light epiluminescence microscopy. The tissue structures are visualized with normal
transmitted light using hematoxylin counterstain. A, shows a strong
mRNA expression in carcinoma cells in an immunohistochemically Ecadherin positive tumor, whereas B shows only a background signal
in an immunohistochemically completely E-cadherin-negative tumor
(magnification, X450).
cause the clinicopathologic correlations of these groups
did not, in general, differ statistically from each other and
because the "negative group" was small. Abnormal Ecad expression was significantly associated with a high
histologic grade of the tumor (chi square test for trend,
P <.001; Table 2) and negative estrogen receptor status
(Fisher's exact test, P = .042; Table 2). E-cad expression
was also marginally related to the patient's nodal status
(Fisher's exact test, P = .063; Table 2). E-cad expression
was not related to the age of the patient, primary tumor
size, PR status, flow cytometric DNA ploidy, or S-phase
fraction, or to c-erbB2 oncoprotein over-expression (Table 2).
E-cadherin Expression and Clinical Outcome
Disease-free survival (DFS) rate of patients with primary breast carcinoma of the invasive ductal type (n =
109) was significantly higher if the tumors were E-cad
positive than if they showed reduced or negative E-cad
staining (Mantel-Cox test, P = .027; Fig. 3). Five-year
DFS of patients with E-cad positive tumors was 70% and
the rate of other patients was 38%. Invasive lobular carcinomas (ILCs) were excluded from the survival analyses
because they always have reduced E-cad expression, regardless of their other clinicopathologic and biologic tumor features, and variation in clinical outcome. Among
these 109 IDC cases, there were no tumors with any lobular features. The follow-up period was too short for reliable evaluation of distant disease-free or overall survival rates.
In addition to E-cad expression (Mantel-Cox test, P =
.027; Fig. 3), nodal status (Mantel-Cox test, P = .068),
histologic grade (Mantel-Cox test, P = .013), ER status
(Mantel-Cox test, P = .043), and PR status (Mantel-Cox
test, P = .021) were associated with DFS, whereas primary tumor size (Mantel-Cox test, P = A 28), flow cytometric DNA ploidy (Mantel-Cox test, P = .339), S-phase
fraction (Mantel-Cox test, P = .282), and adjuvant therapy (Mantel-Cox test, P = .540) were not significantly
associated with clinical outcome in this set of patients.
When E-cad expression, nodal status, histologic grade,
ER, and PR were subjected to Cox multivariate regression analysis, PR (P = .018) and E-cad expression (P =
.072) were selected as independent predictors of DFS.
The relative risk of disease recurrence (RR) was 1.9 times
higher in the group with tumors showing reduced or negative E-cad staining (95% confidence interval, 0.9 to 3.9)
than in those tumors that were E-cad positive (Table 3).
DISCUSSION
The determination of E-cad expression by immunohistochemistry requires a carefully optimized and validated immunohistochemical method. We found it
difficult to validate the staining method for E-cad expression because the aberration to be detected is loss of staining rather than staining, which is usually the case with
proliferation antigens or activated oncogenes. We used
frozen tissue sections to overcome the problems related
to loss of immunoreactivity during formalin fixation and
paraffin embedding. Our attempts with paraffin-embedded tissue sections and commercially available antibodies failed to produce uniform and constantly reproducible stainings even though several antigen retrieval
methods were tested. When frozen sections were used,
the staining results were reproducibly scorable and could
be confirmed by mRNA in situ hybridization. Two previous studies have used paraffin-embedded tissue sections and validated the immunohistochemical methods
employed by comparing the results with frozen section
A.J.C.P.- •il 1996
399
SIITONEN ET AL.
E-Cadherin Expression in Breast Cancer
TABLE 1. E-CADHERIN EXPRESSION IN DIFFERENT HISTOLOGIC TYPES OF BREAST CARCINOMA
E-Cadherin Expression* [No. (%)]
Histologic Type
Pure intraductal carcinoma
Invasive ductal carcinoma
Recurrent ductal carcinomas
Invasive ductal variantsj
Invasive lobular carcinoma
n
Negative
Reduced
Positive
20
239
28
20
55
0(0)
22(9)
4(14)
0(0)
35 (64)
4(20)
102(43)
14(50)
3(15)
20 (36)
16(80)
115(48)
10(36)
17(85)
0(0)
P Value
.004f
<.0001§
* Classification of E-cadherin expression: positive = staining in tumor cells was equal to that of normal epithelium; reduced = staining of
cancer cells was clearly recognizable but weaker than that of normal epithelium; negative = expression was completely lost.
t Difference of the E-cadherin expression (negative + reduced vs. positive) between the groups of pure intraductal, invasive ductal, and
recurrent carcinomas as tested by chi-square test for trend. P value is two-tailed.
$ This group includes invasive medullary, tubular, mucinous, apocrine, epidermoid, and cribriform breast carcinomas.
§ Difference of E-cadherin expression (negative +reducedvs. positive) between invasive ductal and invasive lobular carcinomas was tested
by Fisher's exact test. P value is two-tailed.
TABLE 2. ASSOCIATION OF E-CADHERIN EXPRESSION WITH VARIOUS PATHOBIOLOGIC FACTORS IN 259 INVASIVE
DUCTAL CARCINOMAS
E-Cadherin Expression* \no. (%)]
Pathobiologic Factor
Patient age
s50 years
>50 years
Tumor size
^2cm
>2cm
Nodal status
Axillary node negative
Axillary node positive
Histologic grade
Low
Intermediate
High
Estrogen receptor status
Negative
Positive
Progesterone receptor status
Negative
Positive
Ploidy
Diploid
Aneuploid
Flow cytometric S-phase
Below median (8.5%)
Above median (8.5%)
c-erbB-2 overexpression
Negative
Positive
Negative or Reduced
Positive
P Value
61
181
35(57)
83 (46)
26 (43)
98(54)
.139t
83
114
38 (45)
59(52)
45(55)
55(48)
.387f
122
106
51(42)
58(55)
71(58)
48 (45)
,063f
72
110
51
24(33)
56(51)
33 (65)
48(67)
54 (49)
18(35)
84
156
50(59)
70 (45)
34(41)
86(55)
.042f
145
96
78 (54)
42(43)
67 (46)
54(57)
.148f
97
139
46(47)
71(51)
51(53)
68 (49)
,599f
95
108
42 (45)
57(53)
53(55)
51(47)
.26 If
82
20
50(61)
10(50)
32(39)
10(50)
.450f
* Classification of E-cadherin expression: positive •= staining in tumor cells was equal to that of normal epithelium;staining
reducedof»
cancer cells was clearly recognizable but weaker than that of normal epithelium; negative = expression was completely lost,
t Two-tailed P value of Fisher's exact test.
% P value of chi-square test for trend.
Vol. 105-No. 4
.0005
400
ANATOMIC PATHOLOGY
Original Article
100
positive E-cad
expression (n=66)
90
x
80 -
«3
>
70
60 H
0)
50
0.027
40 H
30
to
0)
negative or reduced
E-cad expression |n=43)
20 H
10
~i—
12
24
—i—
— i —
36
48
60
72
Follow-up (months)
FIG. 3. Disease-free survival of 109 primary invasive ductal breast carcinoma patients according to E-cadherin expression. The difference between E-cad positive tumors (n = 66) and tumors with reduced and
negative E-cadherin expression (n = 43) is significant (P = .027).
immunostainings. ' A weaker staining with a real heterogeneity has been reported, 20 which is consistent with
our experience. Our argument is that the evaluation of
E-cad expression with the antibodies commercially
available can be done more reliably on frozen sections
than on paraffin-embedded sections.
In keeping with the results of previous studies, 1617 ' 20 ' 21
our analysis of a large patient material showed that reduced or negative E-cad expression is a uniform finding
in invasive lobular carcinomas (ILCs). In invasive ductal
carcinomas (IDCs), reduced E-cad expression correlated
with high histologic grade, negative estrogen receptor status, and marginally with positive axillary node status.
Our results on the correlation between reduced E-cad expression and tumor dedifferentiation are consistent with
those reported by Moll and colleagues,16 Gamallo and
associates,17 and Oka and coworkers18 all of whom used
frozen sections. Studies using paraffin-embedded tissue
sections have failed to find any correlation with histologic grade, and the correlation with ER was in the opposite direction to that observed in our study. 2021 These
differences probably reflect not only methodologic
differences, but also the small number of patients in
many studies. 81617 ' 21 Supported by the findings from
studies using frozen sections, our large material suggests
that molecular defects leading to impaired E-cad expression coincide with tumor dedifferentiation and the development of hormone insensitivity.
The findings of this study indicated that impaired Ecad expression may be directly associated with tumor
progression. First, we found a significant proportion of
tumors whose intraductal carcinoma component had re-
tained their E-cad expression, whereas it was reduced or
lost in the invasive carcinoma component (Fig. 1B). Furthermore, reduced E-cad expression was significantly
more frequent in invasive and especially in recurrent carcinomas than in noninvasive, pure intraductal carcinomas. These findings suggest that reduced E-cad expression is involved in the development of an invasive
carcinoma phenotype. Because most of the recurrent tumors were local (nonmetastatic), the question remains
open as to whether loss of normal E-cad expression facilitates the development of true metastases. We found
only a marginally significant association between reduced E-cad expression and axillary node status, which
lends support to the hypothesis that reduced E-cad expression plays a more central role in invasiveness than in
the formation of metastases.
In addition to the association detected with tumor progression, reduced E-cad expression was found to be related to shortened disease-free survival time in the sample of 109 primary IDCs. There is one previous report
on the prognostic impact of E-cad expression on overall
survival in breast cancer.20 Lipponen and colleagues20
studied 208 breast carcinomas and found a weak association between reduced E-cad expression and shortened
overall survival time. However, this association was lost
when the patients were adjusted for other known prognostic factors in multivariate analysis. In our series, Ecad expression retained its independent prognostic impact after adjusting for nodal status, histologic grade, and
hormone receptor status. Our findings warrant further
studies to examine the role of E-cad expression as a prognostic factor in larger patient materials with clinical follow-up data.
There are several possible molecular mechanisms that
lead to reduced E-cad expression. Frixen and colleagues6
were unable to show deletions or gross rearrangements
in the E-cad gene on nonexpressing cell lines. However,
loss of heterozygosity on the q arm of chromosome 16 is
one of the most frequent allelotype changes, not only in
TABLE 3. INDEPENDENT PREDICTORS OF DISEASE-FREE
SURVIVAL IN 109 PRIMARY INVASIVE DUCTAL BREAST
CARCINOMAS ACCORDING TO COX
REGRESSION ANALYSIS
Prognostic Parameter
PR status
Histoscore<;100/>100
E-Cadherin expression
Positive/reduced + negative
RR95%
Relative Confidence
Risk
Interval P Value
2.5
1.0-6.1
.018
1.9
0.9-3.9
.072
Variables tested in the Cox model were as follows: nodal status, histologic grade, ER and
PR status, and E-cadherin expression. ER = estrogen receptor; PR = progesterone receptor.
A.J.C.P.-April 1996
SIITONEN ET AL.
401_
E-Cadherin Expression in Breast Cancer
breast cancer, but also in prostate adenocarcinoma and
hepatocellular carcinoma.32"34 The loss of E-cad expression may also be due to transcriptional down-regulation
of the gene.6 Behrens and associates35 have described the
E-cad promoter that stimulated E-cad transcription in
differentiated breast carcinoma cells, but was less active
in their poorly differentiated counterparts. Recently, a
novel human E-cad gene transcription promoter sequence has been found by Bussenmaker and coworkers.36 The promoter is not active in the E-cad nonexpressing human prostate cancer cell lines. The authors
suggest that this may be due to the binding of a repressor
protein to the promoter. 36
There are also other adhesion processes related to the
normal function of E-cad that contribute to the development of invasive and metastatic phenotypes of carcinoma cells. One of these processes is the expression of
a-catenin, a cytoplasmic protein linking E-cad to actin
filaments and cytoskeleton of the cell.4 Recently, Takayama and colleagues37 reported a significant correlation
between a-catenin and E-cad expression, and showed
that a-catenin expression is decreased in breast carcinoma even more often than E-cad expression. Both Takayama and colleagues37 and Shiozaki and associates38
have suggested that impaired E-cad function may be due
to the down-regulation of a-catenin expression in human carcinomas. These suggestions are supported by
studies with cell lines: mutational inactivation of the acatenin gene has been shown to be one of the mechanisms responsible for the loss of adhesive function in carcinoma cells with normal E-cad expression.39,40 Although loss of E-cad expression, inactivation, or
abnormal E-cad function facilitates cell detachment,
other adhesion molecules mediating cell to matrix interactions play important roles in the metastatic process.
The question of how these processes operate as a multicomponent system and are altered to affect metastases,
remains to be answered.
REFERENCES
1. Takeichi M. Cadherin cell adhesion receptors as a morphogenetic
regulator. Science 1991; 251:1451 -1455.
2. Eidelman S, Damsky CH, Wheelock MJ, Damjanov I. Expression
of the cell-cell adhesion glycoprotein Cell-CAM 120/80 in normal human tissues and tumors. Am J Pathol 1989; 135:101110.
3. Shimoyama Y, Hirohashi S, Hirano S, et al. Cadherin cell-adhesion molecules in human epithelial tissues and carcinomas.
CancerRes 1989;49:2128-2133.
4. Birchmeier W, Behrens J. Cadherin expression in carcinomas:
Role in the formation of cell junctions and prevention of invasiveness. Biochim BiophysActa 1994; 1198:11-26.
5. Behrens J, Mareel MM, Van Roy FM, Birchmeier W. Dissecting
tumor cell invasion: Epithelial cells acquire invasive properties
after the loss of uvomorulin-mediated cell-cell adhesion. J Cell
Biol 1989; 108:2435-2447.
6. Frixen UH, Behrens J, Sachs M, et al. E-cadherin-mediated cellcell adhesion prevents invasiveness of human carcinoma cell. J
CellBiol 1991;113:173-185.
7. Vleminckx K. Vakaet L, Mareel M, Fiers W, Van Roy F. Genetic
manipulation of E-cadherin expression by epithelial tumor cells
reveals an invasion suppressor role. J Cell Biol 1991;66:107119.
8. Shiozaki H, Tahara H, Oka H, et al. Expression of immunoreactive
E-cadherin adhesion molecules in human cancers. Am J Pathol
1991;139:17-23.
9. Schipper JH, Frixen UH, Behrens J, et al. E-cadherin expression in
squamous cell carcinomas of head and neck: Inverse correlation
with tumor dedifferentiation and lymph node metastasis. Cancer Res 1991;51:6328-6337.
10. Dorudi S, Sheffield JP, Poulsom R, Northover JMA, Hart IR. Ecadherin expression in colorectal cancer: An immunohistochemical and in situ hybridization study. Am J Pathol
1993;142:981-986.
11. Umbas R, Isaacs WB, Bringuier PP, et al. Decreased E-cadherin
expression is associated with poor prognosis in patients with
prostate cancer. CancerRes 1994;54:3929-3933.
12. Mayer B, Johnson JP, Leitl F, et al. E-cadherin expression in primary and metastatic gastric cancer: Down-regulation correlates
with cellular dedifferentiation and gladular disintegration. Cancer Res 1993;53:1690-1695.
13. Sakuragi N, Nishiya M, Ikeda K, et al. E-cadherin expression in
endometrial carcinoma is associated with tumor dedifferentiation and deep myometrial invasion. Gynecol Oncol 1994;53:
183-189.
14. Kadowaki T, Shiozaki H, Inoue M, et al. E-cadherin and a-catenin
expression in human esophageal cancer. Cancer Res 1994; 54:
291-296.
15. Mattijssen V, Peters HM, Schalkwijk L, et al. E-cadherin expression in head and neck squamous-cell carcinoma is associated
with clinical outcome. Int J Cancer 1993;55:580-585.
16. Moll R, Mitze M, Frixen UH, Birchmeier W. Differential loss of
E-cadherin expression in infiltrating ductal and lobular breast
carcinomas. Am J Pathol 1993; 143:1 -12.
17. Gamallo C, Palacios H, Suarez A, et al. Correlation of E-cadherin
expression with differentiation grade and histological type in
breast carcinoma. Am J Pathol 1993; 142:987-993.
18. Oka H, Shiozaki H, Kobayashi K, et al. Expression of E-cadherin
cell adhesion molecules in human breast cancer tissue and its
relationship to metastasis. Cancer Res 1993; 53:1696-1701.
19. Bringuier PP, Umbas R, Schaafsma HE, et al. Decreased E-cadherin immunoreactivity correlates with poor survival in patients
with bladder tumors. CancerRes 1993;53:3241-3245.
20. Lipponen P, Saarelainen E, Ji H, Aaltomaa S, Syrianen K. Expression of E-cadherin (E-CD) as related to other prognostic factors
and survival in breast cancer. J Pathol 1994; 174:101 -109.
21. Rasbridge SA, Gillet CE, Sampson SA, Walsh FS, Millis RR. Epithelial (E-) and placental (P-) cadherin cell adhesion molecule
expression in breast carcinoma. J Pathol 1993; 169:245-250.
22. Hartmann WH, Ozzello L, Sobin LH, et al, eds. Histological typing
of breast tumors, ed 2. Geneva, Switzerland: World Health Organization, 1981.
23. Bloom HJG, Richardson WW. Histological grading and prognosis
in breast cancer. Br J Cancer 1957; 11:359-377.
24. Peacock CS, Thompson IW, van Norden S. Silver enhancement of
polymerized diaminobenzidine: Increased sensitivity for immunoperoxidase staining. J Clin Pathol 1991 ;44:756-758.
25. Sommers CL, Thompson EW, Torri JA, et al. Cell adhesion molecule uvomorulin expression in human breast cancer cell lines:
Relationship to morphology and invasive capacities. Cell
Growth Differ 1991;2:365-372.
26. Dagerlind A, Friberg K, Bean AJ, Hokfelt T. Sensitive messenger
RNA detection using unfixed tissue: Combined radioactive and
non-radioactive in situ hybridization histochemistry. Histochemistry 1992;98:39-49.
Vol. 105-No. 4
402
ANATOMIC PATHOLOGY
Original Article
27. Rimm DL, Morrow JS. Molecular cloning of human E-cadherin
suggests a novel subdivision of the cadherin superfamily. Biochem Biophys Res Commun 1994; 200:1754-1761.
28. Kallioniemi O-P. Comparison of fresh and paraffin-embedded tissue as starting material for DNAflowcytometry. Cytometry
1988;9:164-169.
29. Helin HJ, Helle MJ, Helin ML, Isola JJ. Immunocytochemical
detection of estrogen and progesterone receptors in 124 human
breast cancers. Am J Clin Pathol 1988; 89:137-142.
30. Kallioniemi O-P, Holli K, Visakorpi T, et al. Association of c-erbB2 protein over-expression with high rate of cell proliferation, increased risk of visceral metastasis and poor long-term survival
in breast cancer, lnt J Cancer 1991 ;49:650-655.
31. Dixon WJ. BMDP Statistical Software. Los Angeles: University of
California Press, 1981.
32. Sato T, Tanigami A, Yamakawa K, et al. Allelotype of breast cancer: Cumulative allele losses promote tumor progression in primary breast cancer. Cancer Res 1990; 50:7184-7189.
33. Carter BS, Ewing CM, Ward WS, et al. Allelic loss of chromosome
16q and lOq in human prostate cancer. Proc Natl Acad Sci
1990;87:8751-8755.
34. Tsuda H, Zhang W, Shimosato Y, et al. Allele loss on chromosome
16 associated with progression of human hepatocellular carcinoma. Proc Natl Acad Sci 1990;87:6791-6794.
35. Behrens J, Lowrick O, Klein-Hitpass L, Birchmeier W. The Ecadherin promoter: Functional analysis of a G-C-rich region
and an epithelial cell-specific palindromic regulatory element.
Proc Natl Acad Sci 1991;88:11495-11499.
36. Bussenmakers MJG, Giroldi LA, van Bokhoven A, Schalken JA.
Transcriptional regulation of the human E-cadherin gene in human prostate cancer cell lines: Characterization of the human
E-cadherin gene promoter. Biochem Biophys Res Commun
1994;203:1284-1290.
37. Takayama T, Shiozaki H, Inoue M, et al. Expression of E-cadherin
and a-catenin molecules in human breast cancer tissues and associations with clinicopathological features. Int J Oncol 1994; 5:
775-780.
38. Shiozaki H, Iihara K, Oka H, et al. Immunohistochemical detection of a-catenin expression in human cancers. Am J Pathol
1994;144:667-674.
39. Morton RA, Ewing CM, Nagafuchi A, Tsukita S, Isaacs WB. Reduction of E-cadherin levels and deletion of the a-catenin gene
in human prostate cancer cells. Cancer Res 1993;53:3585—
3590.
40. Shimoyama Y, Nagafuchi A, Fujita S, et al. Cadherin dysfunction
in a human cancer cell line: Possible involvement of loss of acatenin expression in reduced cell-cell adhesiveness. Cancer Res
1992;52:5770-5774.
A.J.C.P.. April 1996