Clinicopathologic implications of E-cadherin reactivity in patients with lobular carcinoma in situ of the breast




The current study addressed two questions pertaining to lobular carcinoma in situ (LCIS) of the breast. First, does the risk of a subsequent carcinoma decrease over time after an LCIS biopsy and second, what is the clinical significance of E-cadherin-reactive LCIS?


Eighty-two consecutive patients with a biopsy containing LCIS only, no prior history of breast carcinoma, and follow-up information available for the period 1955–1976 were reviewed. No patients underwent a mastectomy for LCIS. Four hundred eighty-six sections were stained with E-cadherin. E-cadherin reactivity was correlated with clinicopathologic features of the LCIS and subsequent tumors. The mean number of blocks stained per case was 5.9. The mean follow-up period was 21.6 years.


Sixteen patients (19.5%) developed 21 subsequent invasive carcinomas (9 ipsilateral, 2 contralateral, and 5 bilateral carcinomas). The 10-year and 20-year actuarial rates of developing subsequent carcinoma were 7.8% and 15.4%, respectively. Six of the 21 carcinomas (29%) developed after 20 years. Nine LCIS cases (10.9%) had focal E-cadherin reactivity. When compared with patients with nonreactive LCIS, patients with E-cadherin-reactive LCIS more frequently developed a subsequent ipsilateral carcinoma that had a ductal component (55.5% vs. 12.3%; P < 0.01). The subsequent carcinomas also developed after significantly shorter time periods (mean of 7.6 years vs. 19.6 years; P < 0.01).


LCIS appears to confer a persistent, increased risk of subsequent breast carcinoma that does not appear to decrease over time. E-cadherin reactivity appears to identify a subset of LCIS patients with risk factors for subsequent carcinoma similar to those of patients with low-grade intraductal carcinoma. Cancer 2001;92:738–47. © 2001 American Cancer Society.

Lobular carcinoma in situ (LCIS) of the breast is a unique process. Although termed “in situ carcinoma,” the premalignant nature of LCIS has been challenged since its inception as a diagnostic entity in 1942.1 Unlike ductal carcinoma in situ (DCIS), LCIS is not a direct precursor lesion of subsequent ipsilateral invasive carcinoma in the immediate vicinity of the LCIS biopsy.2, 3 Instead, LCIS consistently has been shown to be a marker of increased relative risk of bilateral subsequent carcinoma, which can be of ductal, lobular, or mixed types.4–19 To our knowledge, with the exception of a small group of studies that found an increased risk with increasing amounts of LCIS, no association between the features of LCIS and the risk of developing a subsequent carcinoma or its laterality has been identified to date.4–19 Currently, the usual management of LCIS is observation and the administration of tamoxifen.

To our knowledge the period of time after the diagnosis of LCIS during which patients have an increased risk of developing subsequent breast carcinoma is unclear.20, 21 One opinion is the greatest risk occurs within the first 5 years after the LCIS diagnosis, whereas another is that the risk remains constant over the patient's lifetime.12, 13, 18, 22–24 A contributing factor to the persistence of this controversy is the small number of LCIS patients who have retained both breasts and have been followed for extended periods.7, 9, 12–15, 18, 21, 25

E-cadherin is a transmembrane glycoprotein that is involved intimately in intercellular adhesion of similar-type cells.26–34 Two groups of authors, Moll et al. and Gamallo et al., reported in 1993 that invasive lobular carcinoma had decreased membrane E-cadherin immunohistochemical reactivity compared with invasive ductal carcinomas.35, 36 Both groups of authors briefly commented that LCIS also had no, or markedly decreased, E-cadherin reactivity compared with DCIS and normal breast epithelium. Other authors have reproduced and expanded on these observations.37–44 Absent or decreased E-cadherin membrane reactivity in LCIS denotes decreased intercellular adhesion, which most likely is an essential mechanistic process that contributes to or underlies its noncohesive morphology. In contrast, low-grade DCIS usually is reactive, which most likely facilitates intercellular adhesion and the formation of characteristic architectural patterns.35, 38 These mechanistic observations have extended onto the molecular level. LCIS usually has genetic mutations involving a region of the 16q chromosome that codes for the E-cadherin molecule, thus explaining the absence of immunohistochemically detectable E-cadherin.45–47 To our knowledge, genetic abnormalities in the 16q region have not been detected in ductal carcinomas.40, 46

E-cadherin immunoreactivity in LCIS is not zero. Approximately 5% of LCIS cases will have E-cadherin membrane expression, usually in a patchy distribution and of lesser intensity than is noted in DCIS cases.36, 39, 41 If a lack of immunohistochemically detectable E-cadherin is a reflection of genetic abnormalities characteristic of LCIS, it appears possible that E-cadherin immunoreactivity by LCIS cells may be evidence of ductal differentiation by LCIS cells.35

The current study addresses the question of whether subsequent carcinoma risk decreases or persists over time after a diagnosis of LCIS by contributing additional LCIS patients with long-term follow-up to previous studies. It also examines whether E-cadherin reactivity by LCIS cells is associated with the time interval between diagnosis and development, degree of risk, location, and type of subsequent breast carcinoma.


The anatomic pathology database of the William Beaumont Hospital was searched for cases of breast biopsies with “lobular,” “atypical hyperplasia,” or “small cell proliferation” diagnostic codes accessioned between July 1, 1955 and July 30, 1976. Slides and pathology reports of the index breast biopsies were reviewed. Patients were excluded if they had concurrent or prior invasive breast carcinoma, DCIS, atypical ductal hyperplasia (ipsilateral or contralateral), coexistent DCIS in the index biopsy, or contralateral LCIS. One hundred eighteen patients had LCIS with or without additional benign lesions in a breast biopsy. All the patients underwent open biopsy for a palpable lesion that most often was due to fibrocystic changes and less often due to fibroadenoma or intraductal papilloma. The William Beaumont Hospital surgical pathology department evaluated 6482 benign breast specimens during the same period. The 118 LCIS biopsies comprised 1.8% of these breast specimens, which is similar to the prevalence rates of LCIS of 1.3–2.5% reported by other authors.10, 12, 18

Eighty-two of these 118 patients did not undergo mastectomy for the LCIS, had sufficient follow-up information, and had index-breast tissue blocks that were of adequate physical condition for immunohistochemical staining. Sixty-nine patients (84%) were followed until their deaths. These patients comprised the study population.

Clinical Features

The following features were recorded for each patient: 1) birthdate; 2) date and laterality of LCIS biopsy; 3) laterality of the subsequent carcinoma(s) (subsequent carcinoma was defined as DCIS only or invasive carcinoma [any type]. Pure LCIS in a subsequent ipsilateral or contralateral breast biopsy was not considered to be a subsequent carcinoma); 4) last contact date for patients who did not develop a subsequent carcinoma (the follow-up interval was the difference between the date of LCIS biopsy and the date of last contact); and 5) the date of subsequent carcinoma surgery. The time interval until the development of a subsequent carcinoma was the difference between the date of LCIS biopsy and the date the subsequent carcinoma was detected.

Pathologic Features of LCIS Biopsy

The following pathologic features of the LCIS biopsies were evaluated:

  • 1Confirmation of LCIS in the index biopsy. The histologic criteria of Rodgers for LCIS and its distinction from atypical lobular hyperplasia (ALH) and DCIS were applied rigidly.48 LCIS cells were small and monomorphic. Their nuclei predominantly were round or slightly oval and normochromatic, and their cytoplasm was inconspicuous or clear.48 The histologic diagnosis of LCIS required distended, distorted, and terminal duct lobular units (TDLUs) filled with noncohesive, lobular-type, uniform cells that did not form tubules or have orientation. The cells had to be monomorphic; fill the TDLUs in a regular, homogeneous manner; and comprise the entire population of cells in the involved acinus.48 A LCIS diagnosis and inclusion in the current study required all the acini in at least one TDLU were filled, distorted, and expanded by the uniform cells. The minimal number of LCIS acini in the current study was greater than the minimal requisite number used by Rodgers48 to ensure that all the patients in the current study had unequivocal LCIS. Distinction between LCIS and lobular involvement by low-grade, solid-pattern DCIS relied on the written criteria, comments, and photographs by Anderson et al., Rogers et al., Fechner et al., Rosen et al., and Haagensen et al.19, 48–52
  • 2The number of TDLUs with LCIS or ALH, defined as sufficient characteristic lobular-type cells within the expanded TDLU sufficient for an LCIS or ALH diagnosis if that TDLU was examined in isolation.48, 53 TDLUs containing a lobular proliferation below the diagnostic threshold for ALH were excluded.53
  • 3LCIS cell type, classified as Type A, B, or mixed, using the criteria of Haagensen et al.54
  • 4Number of blocks stained with E-cadherin.
  • 5The number of TDLUs containing E-cadherin-reactive LCIS cells.

Subsequent Carcinoma Pathologic Features

The following pathologic features of subsequent carcinomas were evaluated: 1) the type of carcinoma (classified as pure DCIS [nuclear grade also recorded], invasive ductal, invasive lobular, mixed ductal and lobular, tubulolobular, or mucinous;55 2) the presence of LCIS in association with invasive carcinoma; and 3) the presence and nuclear grade of DCIS in association with invasive carcinoma.

E-Cadherin Immunohistochemical Staining

Three 3-μm thick consecutive sections were cut from each block and each section was placed on a charged slide. One section was stained with hematoxylin and eosin, one was used for E-cadherin immunohistochemistry, and the third section was used as the negative immunohistochemistry control. Deparaffinization used sequential slide immersions into two xylene baths, three alcohol baths of decreasing concentration, and two water baths. The slides were washed in water for 1 minute after the deparaffinization, immediately immersed in ethylenediamine tetraacetic acid (EDTA) buffer (pH 7.0), and put into a commercial vegetable steamer at 95 °C for 30 minutes. The slides were allowed to cool on the counter, remaining immersed in the heated EDTA-filled containers for 5 minutes, followed by a 2-minute rinse with water also while the slides remained in the containers. The slides then were transferred into TRIS-filled containers (pH = 7.0) and allowed to undergo an additional 10 minutes of cooling on the countertop. The slides were transferred to a commercial immunohistochemistry autostainer (Dako Co., Carpinteria, CA). While on the autostainer, the slides first were washed with buffer followed by hydrogen peroxide incubation. The latter was rinsed off and the mouse, antihuman, E-cadherin antibody (Zymed Laboratories, San Francisco, CA), clone 4A2 C7 (dilution 1:50), was incubated over the sections for 40 minutes at room temperature. After the primary antibody was washed off, the components of the Envision-plus™ (Dako Co.) detection system were applied, including an antimouse polymer, two distilled-water washes, and a final diamino-benzene incubation for 5 minutes. Sections were counterstained lightly with hematoxylin and coverslipped. A positive control slide containing E-cadherin-reactive, normal breast lobules was included with each batch of study slides. A negative control slide for each study slide was incubated adjacent to the study slide on the autostainer. Staining procedures for the negative control slide were identical to those of the study slides with the exception of the E-cadherin antibody application.

Sections from 10 normal breast tissue blocks that were accessioned during the identical time period as the study cases were stained with E-cadherin antibody solutions of increasing dilutions prior to the start of the study. The least-concentrated E-cadherin antibody solution at which normal breast epithelium retained strong membrane reactivity was used in the study slides. All the slides demonstrated E-cadherin membrane reactivity in normal breast epithelium.

Statistical Analysis

All LCIS biopsy features and E-cadherin reactivity were evaluated without knowledge of the patient's outcome. Statistical analyses were performed with SAS Version 8.0 statistical software (SAS Institute Inc., Cary, SC). The Fisher exact test (two-tailed), linear regression, and logistic regression tests were used for the direct comparison of variables. The Kaplan–Meier method was used for actuarial survival analyses and Cox regression analysis was used for comparing E-cadherin reactivity and risk of failure over time.



The overall mean and median follow-up periods were 21.6 years and 22.4 years, respectively (range, 2.2–40.4 years; standard deviation, 9.6 years). The mean and median follow-up periods of the patients who did not develop a subsequent carcinoma were 23.5 years and 22.5 years, respectively (range, 2.1–40.4 years; standard deviation, 9.6 years). The mean and median ages of the patients at the time of the LCIS diagnosis were 44.2 years and 44.9 years, respectively (range, 33.6–59.9 years; standard deviation, 4.8 years).

Sixteen patients (19.5%) developed a subsequent carcinoma. Nine patients developed ipsilateral-only carcinoma, five patients developed bilateral carcinoma, and two patients developed contralateral-only carcinoma. Overall, 14 cases (67%) were ipsilateral and 7 cases (33%) were contralateral. The mean and median time periods between the diagnosis of LCIS and the development of all subsequent carcinomas were 15.0 years and 14.2 years, respectively (range, 4.7–24.5 years; standard deviation, 6.7 years). The mean and median time periods between the diagnosis of LCIS and the development of subsequent ipsilateral carcinomas were 15.3 years and 14.2 years, respectively (range, 4.7–24.5 years; standard deviation, 7.0 years). The mean and median time periods between the diagnosis of LCIS and the development of subsequent contralateral carcinomas were 14.6 years and 13.8 years, respectively (range, 9.0–19.0 years; standard deviation, 6.4 years). The mean age at the time of diagnosis of LCIS of those patients who developed a subsequent carcinoma was 42 years, compared with 45 years in patients who did not develop a subsequent carcinoma (P = 0.08). When the patient age at the time of LCIS diagnosis was analyzed as a continuous variable using Cox regression analysis, there was a nonsignificant trend toward younger patient age and an increased risk of subsequent carcinoma (P = 0.10).

The overall, ipsilateral, and contralateral 20-year actuarial rates of subsequent carcinoma were 15.4%, 14.5%, and 8.8%, respectively (Fig. 1) (Table 1). The interval time period between the diagnosis of LCIS and the development of a subsequent carcinoma was distributed evenly over the follow-up period (Table 2). Only 2 subsequent carcinomas (10%) developed within the first 5 years after the LCIS biopsy, but 6 subsequent carcinomas (29%) developed ≥ 20 years after the LCIS biopsy. A taper point, or specific point at which the rate of subsequent carcinomas declined, was not identified.

Figure 1.

Kaplan–Meier curves demonstrating freedom from any subsequent carcinoma for all patients, and freedom from subsequent ipsilateral carcinoma in E-cadherin-reactive versus nonreactive cases.

Table 1. Actuarial Rates of Subsequent Carcinoma
Time periodOverallIpsilateralContralateral
Table 2. Time Interval until Subsequent Carcinoma
Time after LCIS biopsy (yrs)No. of carcinomasPercent of subsequent carcinomas
  1. LCIS: lobular carcinoma in situ.

< 5.029%
≥ 20.0629%


LCIS biopsies

The mean number of blocks examined per case for all patients was 5.9 (range, 2–11; standard deviation, 2.0). Forty-six patients (56.1%) had pure type-A cell LCIS, 21 patients (25.6%) had a mixture of type-A and type-B cell LCIS, and 15 patients (18.3%) had pure type-B cell LCIS. The majority of involved TDLUs were filled by pure type-A cell LCIS in those cases with mixed cell LCIS. The mean number of TDLUs with ALH or LCIS was 16.8 (range, 3–40; standard deviation, 9.4). The mean number of TDLUs with ALH or LCIS was nearly identical among those patients who developed a subsequent carcinoma (mean, 16.3) and those patients who remained free of disease (mean, 17.0).

Subsequent Carcinomas


Of the 14 subsequent ipsilateral carcinomas, 8 (57%) had an invasive ductal carcinoma component (Table 3). Six were pure invasive ductal carcinomas. Four of these pure invasive ductal carcinomas (three of which were well differentiated and one of which was moderately differentiated) were usual-type invasive ductal carcinomas, and two were invasive tubular carcinomas. Nuclear Grade 1 cribriform DCIS and LCIS was admixed in four of these six cases, coexistent pure LCIS was present in one case, and one case had no in situ component. Two of the eight invasive carcinomas with a ductal component were mixed lobular and well differentiated usual-type invasive ductal carcinoma. One was invasive tubulolobular and the other was well differentiated usual-type ductal and lobular carcinoma. One of two mixed invasive ductal and lobular carcinomas had coexistent LCIS only and the other had LCIS and Grade 1 DCIS. Five of the 14 subsequent ipsilateral carcinomas (36%) were pure lobular-type. All had coexistent in situ carcinoma, three had pure LCIS, and two had LCIS and nuclear Grade 1 DCIS. One of the 14 subsequent ipsilateral carcinomas (7%) was a nuclear Grade 1 cribriform and solid DCIS with abundant LCIS.

Table 3. Characteristics of Subsequent Carcinomas
Invasive carcinomasNo.Associated carcinoma in situ
DCIS onlyDCIS and LCISLCIS onlyNone
  • DCIS: ductal carcinoma in situ; LCIS: lobular carcinoma in situ.

  • a

    Includes two invasive tubular carcinomas.

  • b

    Includes one mixed invasive tubular and lobular carcinoma (tubulolobular carcinoma).

All ipsilateral140941
 Pure ductal6a0411
 Mixed ductal and lobular2b0110
 Pure lobular50230
 DCIS only10100
All contralateral70340
 Pure ductal30210
 Mixed ductal and lobular10010
 Pure lobular30120


Four of the seven subsequent contralateral carcinomas (57%) had an invasive ductal component (Table 3). Three were pure, usual-type invasive ductal carcinomas (two were well differentiated and one was poorly differentiated). Two of these three cases had coexistent LCIS and DCIS, and the third case had LCIS only. One of the seven subsequent contralateral invasive carcinomas (14%) was a mixed, well differentiated usual-type ductal and lobular carcinoma with LCIS but no DCIS. Three of the seven subsequent contralateral carcinomas (29%) were pure invasive lobular carcinomas. Two of these had pure LCIS and one had both LCIS and DCIS.

E-Cadherin Reactivity

Four hundred eighty-six sections were stained with E-cadherin. Nine patients (10.9%) had E-cadherin membrane reactivity in the LCIS cells. The E-cadherin reactivity was focal and was confined to the LCIS in a single region of one slide. No cases demonstrated LCIS E-cadherin reactivity on more than one slide. The mean number of TDLUs with E-cadherin-reactive LCIS cells per case was 2.3 (range, 1–4 TDLUs). The intensity was weak to moderate in all cases and was less intense than the reactivity of the adjacent normal breast epithelial cells (Figs. 1–6). Focal E-cadherin reactivity was a membranous brown linear reaction product that usually only involved a portion of the cell membrane. None of the E-cadherin reactive LCIS cases had the large, irregular, globular reactivity characteristic of myoepithelial cell processes.

Figure 2.

Lobular carcinoma in situ involving several, closely situated terminal duct lobular units (H & E, ×24).

Figure 3.

Lobular carcinoma in situ, higher magnification of Figure 3 (H & E, ×460).

Figure 4.

E-cadherin stained section of FIGURE 2.

There was moderate E-cadherin reactivity in the lobular carcinoma in situ, producing a brown hue (H & E, ×24).

Figure 5.

High magnification of the photomicrography in FIGURE 4. The majority of the lobular carcinoma in situ cells had faint, partial membrane reactivity, producing a thin line of brown staining (H & E, ×512).

Figure 6.

E-cadherin reactivity of lobular carcinoma in situ. There was moderate, complete membrane reactivity around a minority of cells (H & E, ×624).

E-cadherin statistical analyses

E-cadherin staining was found to be associated strongly with the development of a subsequent ipsilateral carcinoma (P < 0.01) (Table 4). Five of the 9 patients with E-cadherin-reactive LCIS cells (55.5%) developed a subsequent ipsilateral invasive carcinoma compared with 9 of the 73 patients with no E-cadherin-reactive LCIS cells (12.3%). The 10-year actuarial rates for the development of ipsilateral subsequent carcinomas among patients with E-cadherin-reactive and nonreactive LCIS were 61% and 0%, respectively (Fig. 1) (P < 0.01). The mean time intervals between the diagnosis of LCIS and the development of subsequent ipsilateral carcinomas in patients with E-cadherin-reactive LCIS was 7.6 years (range, 4.7–9.0 years; standard deviation, 1.65 years). This interval was significantly shorter than the mean of 19.6 years (range, 13.8–24.5 years; standard deviation, 4.5 years) noted in patients with E-cadherin nonreactive LCIS (P < 0.01). The 20-year actuarial rate of developing a subsequent ipsilateral carcinoma was 7% for E-cadherin nonreactive LCIS (Fig. 1).

Table 4. E-Cadherin Reactivity and Subsequent Carcinoma
Patient groupNo. of patientsLCIS E-cadherin reactivityP value
  1. LCIS: lobular carcinoma in situ.

Ipsilateral carcinoma14 (17%)9 (12%)5 (56%)0.01
Contralateral carcinoma7 (9%)5 (9%)2 (22%)0.17
All patients82 (100%)73 (100%)9 (100%)

E-cadherin-reactive LCIS was not found to be associated with a subsequent contralateral carcinoma using the Fisher exact test (P = 0.17), but there did appear to be a significant relation when the Cox regression analysis was used (P = 0.02). Of the nine E-cadherin-reactive patients, two developed contralateral failures.

The subsequent invasive carcinoma that developed subsequent to E-cadherin-reactive LCIS was found to have a ductal component in all 7 patients (100%). Of the five ipsilateral subsequent carcinomas, three were pure, usual-type well differentiated ductal carcinomas; one was a tubular carcinoma; and one was a mixed ductal and lobular carcinoma. One of the two subsequent contralateral carcinomas was a pure, well differentiated, usual-type invasive ductal carcinoma and the other was a mixed, well differentiated, usual-type ductal and lobular carcinoma. In contrast, only 38% of the 14 patients with E-cadherin nonreactive LCIS had a ductal component detected in their subsequent carcinomas (P < 0.01, Fisher exact test).

Patients with E-cadherin-reactive LCIS had a younger mean age at the time of LCIS diagnosis than those with nonreactive LCIS (37.6 years [range, 33.6–49.1 years] vs. 45.0 years [range, 35.6–58.4 years]; P < 0.01). However, there were no statistically significant differences with regard to the mean age between patients who did and those who did not develop a subsequent carcinoma within the E-cadherin-reactive LCIS (P = 0.36) and nonreactive LCIS patient groups (P = 0.20). Similar results were obtained when mean patient ages were compared against subsequent ipsilateral carcinomas. There also was no independent association found between mean patient age (P = 0.71) and the development of subsequent ipsilateral carcinoma when E-cadherin status was included on the multivariate Cox regression analysis. Only E-cadherin-reactive LCIS was found to be significantly related to patient age (P < 0.01).

E-cadherin reactivity in LCIS also was found to be associated with the number of blocks available for staining (P < 0.01). The mean number of stained blocks per case in E-cadherin-reactive LCIS was 8.2 (range, 7–10 blocks; standard deviation, 1.20) compared with 5.6 blocks per case in E-cadherin nonreactive LCIS (range, 2–11 blocks; standard deviation, 1.92). All the E-cadherin-reactive LCIS cases had more than seven blocks available for staining.

Additional Analyses

There were no significant differences (P > 0.10) with regard to the distribution of LCIS cytology, the number of TDLUs with LCIS, the number of stained sections, and the cytologic subtype of LCIS between patients who did and patients who did not develop a subsequent carcinoma. There also was no significant difference with regard to the mean time interval until the development of a subsequent carcinoma between patients who developed ipsilateral (15.3 years) and contralateral (14.7 years) subsequent carcinomas (P = 0.84, Student t test).


LCIS is associated with an increased risk of subsequent carcinoma in either breast.4–19 There is controversy among authors regarding the interval at which the risk of developing a subsequent carcinoma decreases.20, 21 One group of authors found that the greatest rate of development of subsequent carcinoma occurred within 1–5 years after the initial LCIS diagnosis, with the rate decreasing after longer follow-up periods.12, 13, 22, 23 Another group reported that the subsequent carcinoma risk persisted with the length of the follow-up interval.18, 24 One of the confounding elements contributing to this issue is the small number of patients with < 15 years of follow-up.7, 9, 12–15, 18, 21, 25 The current study of 82 patients, with a mean follow-up interval of 21.6 years, found that the incidence of subsequent carcinoma did not decrease with an increasing post-LCIS biopsy time interval. Approximately 29% of the subsequent carcinomas developed > 20 years after the LCIS biopsy. These results support the conclusions of Rosen et al. that there was a persistent, increased risk of developing subsequent bilateral breast carcinoma over the patient's lifetime after an LCIS biopsy.18, 24

E-cadherin-reactive LCIS purportedly is uncommon. Loss of E-cadherin membrane expression appears to be an early event in the majority of LCIS cases.35, 45–47, 56 This point is supported by our observation that E-cadherin is absent from breast biopsies containing only minimally proliferative lobular lesions (unpublished data). With increasing experience, E-cadherin appears to be a useful antibody for assisting in the classification of epithelial lesions of the breast as ductal-type or lobular-type. Nearly all ductal lesions are diffusely and strongly reactive whereas only 0–10% of LCIS cases are reactive. The majority of authors who have studied the utility of E-cadherin in this regard have found that < 5% of LCIS cases are E-cadherin-reactive, and only possess focal reactivity in positive cases.35–44 We also found that the E-cadherin reactivity of LCIS also was always focal, usually only in a single or few closely situated TDLUs. Unlike other studies, 10.9% of the 82 LCIS cases in the current study were E-cadherin-reactive, which we believe is a greater frequency than the collective experience of other authors reported to date. Several issues may shed light on the contributing factors that account for these differences. First, E-cadherin-reactive LCIS was found to be associated significantly with the number of blocks stained with E-cadherin antibody. At least 7 blocks were stained in all the E-cadherin-reactive LCIS cases compared with an overall mean of 5.9 blocks per case. These values are substantially greater than the one representative tissue block per case used in the majority of other studies. Together, these results suggest that the prevalence of E-cadherin-reactive LCIS may be due in part to the intensity of the search. Second, all the E-cadherin-stained tissue sections in the current study were 3-μm thick and the hematoxylin counterstain was kept at a light-blue hue. Based on our experience with other membrane-reactive antibodies, we have found that these two technical modifications greatly facilitate the identification of weak membrane-based immunoreactions.

E-cadherin appears to identify a unique subset of LCIS patients. Patients with E-cadherin-reactive LCIS developed a subsequent ipsilateral carcinoma significantly more often than patients with nonreactive LCIS. These subsequent ipsilateral carcinomas always had a ductal component and appeared to develop after a significantly shorter time interval in E-cadherin-reactive LCIS patients than in nonreactive patients. These findings run contrary to the recognized associations between LCIS and an increased risk of subsequent carcinoma in either breast, some cases of which develop after a long time interval after the LCIS biopsy.

To our knowledge the pathogenetic mechanisms underlying these clinical findings are unknown. We were struck by the associations between E-cadherin-reactive LCIS and the increased development of subsequent ipsilateral carcinomas with a ductal component after relatively short time periods, which we believe are akin to the features of residual low-grade intraductal carcinoma arising after incomplete eradication.57–62 These results can be viewed as being supportive of the theory that the lesion that appears to be E-cadherin-reactive LCIS is a subtype of DCIS. LCIS and well differentiated DCIS have a high degree of genetic homology.63, 64 This has been interpreted by some clinicians as support for the hypotheses that LCIS and well differentiated DCIS are closely related neoplastic lesions that evolve from one neoplastic cell. Loss of E-cadherin expression in LCIS may represent a molecular switch to the loosely cohesive growth pattern.63, 64 The most frequent (but not universal) group of genetic abnormalities in lobular carcinomas are somatic frame mutations in the E-cadherin gene locus on chromosome 16q.45–47 These mutations are associated with loss of heterozygosity of the wild-type allele, abnormal truncated E-cadherin molecules, and lack of membrane E-cadherin immunoreactivity.45–47, 56 Truncated E-cadherin molecules are believed to interfere with the normal cell-cell function, resulting in the characteristic dyshesive cell pattern of lobular carcinoma.35, 46 Mutations in the 16q, E-cadherin gene locus have not been identified in ductal carcinomas.40, 46 It is possible that E-cadherin reactivity in LCIS may imply the presence of an intact 16q E-cadherin gene and a DCIS genotype, with the accompanying risk factors of DCIS. In this scenario, the characteristic cellular dyshesion of LCIS may be the result of low-grade DCIS that has sustained alterations in one of the molecules related to the E-cadherin binding system, such as the α-, β-, or γ- catenin, but the E-cadherin molecule is left intact.27, 30, 39, 65–67 Defects in the α- or β-E-catenin or their genes have resulted in loss of E-cadherin-mediated intercellular adhesion.66, 67 These issues require intensive investigation. Advanced molecular genetic techniques may shed considerable light on these questions.

Finally, we wish to issue several cautionary notes pertaining to the application of these results in routine surgical pathology. First, the results of the current study need to be confirmed by other authors. Second, blanket staining of all LCIS-containing tissue blocks with E-cadherin most likely is not realistic or practical. Third, we do not advocate the classification of in situ carcinoma as lobular-type or ductal-type based solely on the presence or absence of E-cadherin membrane reactivity. Although a lack of intact E-cadherin may underlie the cellular dyshesion of LCIS, we believe that standard morphologic criteria should be applied rigorously until additional studies provide more insight. Last, we believe that these results should not be misconstrued to suggest that E-cadherin-reactive LCIS “predicts” a definite subsequent carcinoma. In the current study, a subsequent carcinoma did not develop in nearly 50% of patients (45%) with E-cadherin-reactive LCIS.

One major limitation of the current study was the lack of analysis and statistical control of other known risk factors for breast carcinoma such as family history. The distant histories of these patients precluded acquisition of this information.

The risk of developing a subsequent carcinoma remains constant over the lifetimes of patients with LCIS, with 29% of the subsequent carcinomas occurring > 20 years after the LCIS biopsy in the current study. Focal (usually weak) E-cadherin reactivity is present in approximately 10% of LCIS cases. Differences in the reported incidence of E-cadherin-reactive LCIS may be a reflection of the number of sections stained per case and technical factors. Patients with E-cadherin-reactive LCIS were significantly younger and more frequently developed subsequent ipsilateral carcinomas that occurred after a significantly shorter time period compared with patients with nonreactive LCIS. The pathogenic mechanisms underlying these findings are uncertain and will require additional investigation.


The authors thank Mamtha Balasubramaniam, M.S., for her assistance with statistical analysis and Paul Rigo of the Beaumont Tumor Registry for his assistance with follow-up data.