Pathologic findings from the National Surgical Adjuvant Breast and Bowel Project

Twelve-year observations concerning lobular carcinoma in situ

Authors

  • Edwin R. Fisher M.D.,

    Corresponding author
    1. Pathology Center, National Surgical Adjuvant Breast and Bowel Project Pathology Center, Pittsburgh, Pennsylvania
    2. Allegheny General Hospital, West Penn Allegheny Health System, Pittsburgh, Pennsylvania
    • Allegheny Cancer Center, 5th Floor, Allegheny General Hospital, 320 East North Avenue, Pittsburgh, PA 15212
    Search for more papers by this author
    • Fax: (412) 359-5100

  • Stephanie R. Land Ph.D.,

    1. Biostatistical Center, National Surgical Adjuvant Breast and Bowel Project Biostatistical Center, Pittsburgh, Pennsylvania
    2. Department of Biostatistics, University of Pittsburgh, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Bernard Fisher M.D.,

    1. National Surgical Adjuvant Breast and Bowel Project Operations Center, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Eleftherios Mamounas M.D.,

    1. Aultman Cancer Center, Canton, Ohio
    Search for more papers by this author
  • Linda Gilarski,

    1. Biostatistical Center, National Surgical Adjuvant Breast and Bowel Project Biostatistical Center, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Norman Wolmark M.D.

    1. National Surgical Adjuvant Breast and Bowel Project Operations Center, Pittsburgh, Pennsylvania
    Search for more papers by this author

Abstract

BACKGROUND

The current report represents a 12-year clinicopathologic update of an earlier 5-year analysis of 180 patients with lobular carcinoma in situ (LCIS) who were treated with local excision and subsequent surveillance only.

METHODS

Nineteen pathologic characteristics of LCIS were assessed as potential predictors of invasive and noninvasive ipsilateral breast tumor recurrence (IBTR) and contralateral breast tumor recurrence (CBTR) as well as mortality.

RESULTS

Overall, only 26 IBTRs (14.4%) and 14 CBTRs (7.8%) were observed. Nine IBTRs (5.0% of the total cohort) and 10 CBTRs (5.6% of the total cohort) were invasive carcinomas. Eight of 9 IBTRs (88.9%) and 6 of 8 invasive CBTRs (75%) that had histologic sections available for review were of the lobular invasive type. Ninety-six percent of all IBTRs and 100% of invasive IBTRs occurred within the same site as the index LCIS. The numbers of invasive IBTRs were comparable within and after 5 years (5 invasive IBTRs vs. 4 IBTRs). Recurrences of invasive CBTR occurred later than recurrences of invasive IBTR, with 70% of invasive CBTRs recognized after 5 years compared with 44% of invasive IBTRs. It was found that Grade 2–3 LCIS was significantly predictive for invasive IBTR when combined with the number of recurrences of ductal carcinoma in situ (DCIS) alone or with LCIS. Only 2 patients in the cohort (1.1%) succumbed to breast carcinoma; 1 patient had a prior invasive IBTR, and the other patient had an invasive CBTR. The reasons for the lower frequency of invasive recurrences and the higher proportions of the lobular invasive phenotype than noted by others are discussed along with the impact of the findings on the nomenclature, precursor nature, and treatment of LCIS.

CONCLUSIONS

LCIS is a more indolent form of in situ breast carcinoma than DCIS, with which it shares other features of its natural history, particularly very low mortality rates. There is no compelling reason to surgically treat LCIS other than conservatively. The values of other adjuvant modalities in the management of LCIS are discussed. The authors acknowledge that their findings are based on relatively few events and, even at 12 years, may be regarded as ‘preliminary’. Nonetheless, their findings may reflect the true biologic nature of LCIS. Cancer 2004;100:238–44. © 2003 American Cancer Society.

Although lobular carcinoma in situ (LCIS) of the breast was described more than 6 decades ago, uncertainty remains concerning some of its pathologic features, natural history, and treatment. Our review of the literature of LCIS up to 1996 revealed only 14 studies that included a total of 703 patients who were treated only by excisional biopsy.1 The periods of follow-up were quite variable, and the sites and types of ipsilateral breast tumor recurrences (IBTRs) were noted only rarely. Most important, the pathologic criteria used for the diagnosis of LCIS and putative risk factors for IBTR also varied. Our own investigation of LCIS, published in 1996,1 involved 182 patients who were treated with excisional biopsy only and who were registered and followed by the National Surgical Adjuvant Breast Project (NSABP). Results after 5 years of observation revealed 13 IBTRs (7%). Four of those recurrences (2% of the total cohort) were invasive malignancies. Two invasive recurrences (1%) occurred in the contralateral breast (CBTR). All invasive IBTRs and 1 of 2 invasive CBTRs were of the lobular invasive histologic type. Nineteen pathologic characteristics of LCIS were assessed as predictors of recurrence. Although the sample sizes were small, it was found that only the Type 3 form and, to a lesser extent, the Type 2 form of the disease were significant predictors for these events. In addition, possible roles of margin status, administration of tamoxifen, or irradiation in the treatment of LCIS were suggested along with the precursor nature of LICS for invasive disease. However, the results and conclusions were considered preliminary, because the number of events was limited and the follow-up was relatively short. In fact, it has been suggested that approximately 50% of recurrences subsequent to excision of LCIS occur after 10 years.2

Examination of the literature after our 1996 publication has revealed 3 additional studies,3–5 encompassing 214 examples of LCIS that had been treated with local excision only. These reports indicated frequencies of recurrence of invasive carcinoma in approximately 10% and 20% of patients at 10 years and 20 years, respectively. The survival rate at 15 years, which was mentioned in only 1 of the studies,3 was 100% in a cohort of 32 patients.

Studies of LCIS reported after 1996 exhibit shortcomings similar to those noted prior to that time. These considerations prompted us to perform a 12-year update of our previous findings and to incorporate information derived from recent studies from others that may have an impact on the pathology, natural history, and treatment of LCIS.

MATERIALS AND METHODS

The materials and methods used in the current study are described in detail elsewhere.1 In brief, hospital pathologists who participated in the study submitted routinely stained sections and, occasionally, paraffin blocks from 218 women who were diagnosed with LCIS and who, after informed consent was obtained, were treated with local excision only. These sections were submitted to the NSABP pathology headquarters for review by one of the authors (E.R.F.). Thirty sections were excluded for a variety of reasons, including inadequate pathologic materials, performance of mastectomy, and the presence of LCIS as well as ductal carcinoma in situ (DCIS) independently in the same specimen. Six sections exhibited lesions that we designated ductolobular carcinoma in situ (DCLIS).1, 6 Those sections were characterized by the presence of classic DCIS in large ducts and LCIS in the distal portion of the same ductolobular unit. In other sections, cells characteristic of LCIS appeared in both the lobular component and the large ductal component of the same ductolobular unit. These latter sections often were difficult to classify as ductal or lobular in nature. Compounding this difficulty was the presence of comedo necrosis or microacini, features of DCIS, among the ‘LCIS-type’ cells. Some investigators7 have referred to these forms as indeterminate carcinoma in situ. Immunohistochemical staining studies with the antibody for E-cadherin recently have supported the dual nature of some by demonstrating negative and positive elements consistent with their lobular and ductal identification, respectively, although other samples appear to reflect only a lobular derivation.7–9 The treatment for patients with DLCIS has not been established. However, we have suggested that local breast irradiation is appropriate for patients with DLCIS, but not for patients with LCIS, because of the putative ductal component of the former.1, 6 This therapeutic consideration prompted their exclusion. Two patients who withdrew consent to be followed also were removed from the study.

After all exclusions, there were 180 patients remaining for analyses, 2 fewer than the 182 patients on whom the previous report was based, reflecting the absence of the 2 patients who withdrew from the study. The pathologic materials from all except two recurrences were available for pathologic review.

Table 1 lists the 19 pathologic characteristics of LCIS that were assessed as potential predictors for overall IBTR and CBTR as well as those that were exclusively predictors of invasive carcinoma or noninvasive DCIS. LCIS was graded according to the degree of involvement of the lobules, as reported previously,1 with slight modifications according to the criteria proposed and illustrated by Tavassoli10 for lobular neoplasia (LN). LCIS 1 denotes the presence of the characteristic cytologic elements of LCIS that involve a portion or all of the lobular ductular lumens with slight or no distension. According to this definition, LCIS 1 includes some lesions designated as atypical lobular hyperplasia (ALH). LCIS 2 resembles LCIS 1 except that all ductules exhibit modest distension. Overt ductular and lobular distension with areas that exhibit little or no interductular stroma were designated LCIS 3. It should be emphasized that residual ductules may be encountered in all types. Tavassoli10 regards lesions in which the ductules are replaced completely by signet ring cells, regardless of the degree of distension, as a subtype of LCIS 3 (LCIS 3-2). We have not encountered such a lesion in this cohort, although the presence of lesser numbers of signet ring cells as well as so-called targetoid forms are not uncommon. It should be noted that the highest grade of LCIS was used for analyses when more than one was encountered in a particular sample. Margins were regarded as free when tumor was not transected, regardless of how ‘close’ to the line of excision they may have appeared.

Table 1. Pathologic Variables of Lobular Carcinoma In Situ Analyzed
  1. LCIS: lobular carcinoma in situ.

Variable
No. of involved lobules
Duct extension
LCIS grade
Intralesional calcification
Focality
Lobular hyperplasia
Proliferative fibrocystic disease
Nuclear grade
Cell size
Signet ring cells
Pagetoid cells
Histiocytoid cells
Loss cohesion
Mitoses
Mucin
Glycogen
Lymphocytic infiltration
Free margins
Persistence of ductules

The average annual rates of occurrence of invasive IBTR and CBTR were calculated by categories of the 19 pathologic characteristics examined. Those characteristics were compared using a two-sided log-rank test.11 The results were based on a mean time on study of 146 months. The mean follow-up was computed by the Kaplan–Meier method, using each patient's most recent follow-up as the event of interest and death as the censoring event.

RESULTS

Table 2 lists the pathologic type, site, and time of recurrences. At the time of the current report, there were 26 patients (14.4%) with IBTRs of all types, including 1 patient with 2 metachronous tumors in the same breast, 3 patients with metachronous IBTR and a CBTR, and 1 patient with a synchronous IBTR and CBTR Nearly 85% of the IBTRs were detected by mammography. Nine of 26 IBTRs (34.6%; 5.0% of the total cohort) were invasive malignancies, and 8 of 9 IBTRs (88.9%) were of the lobular invasive histologic type. The remaining 17 IBTRs included of DCIS alone in 6 patients (3.3% of the total cohort), LCIS only in 10 patients (5.6% of the total cohort), and both forms of carcinoma in situ in 1 patient (0.6% of the total cohort). All of the invasive IBTRs and those consisting of ‘pure’ DCIS and DCIS combined with LCIS were found in the same quadrant as the index LCIS. Only 1 of 10 IBTRs characterized as pure LCIS was found at a different site.

Table 2. Types, Times, and Locations of Breast Tumor Recurrences: Lobular Carcinoma In Situa
Study no.Pathologic typeTime to recurrence (yrs)Recurrence site-quadrant
  • LOB; lobular; INV: invasion; IBTR-S: ipsilateral breast tumor recurrence in the same quadrant; LCIS: lobular carcinoma in situ; DCIS: ductal carcinoma in situ; CBTR: contralateral breast tumor recurrence; NOS: not otherwise specified; IBTR-D: ipsilateral breast tumor recurrence in a different quadrant; MUC: mucinous.

  • a

    In relation to the index lesion.

  • b

    Ipsilateral breast tumor recurrence.

  • c

    Contralateral.

  • d

    First.

  • e

    Second.

  • f

    No slides were available for review.

237LOB INV≤ 5IBTR-S
730LOB INV and LCIS≤ 5IBTR-S
834LOB INV and LCIS≤ 5IBTR-S
517DCIS≤ 5IBTR-S
518DCIS,b DCISc≤5,b ≤5cIBTR-S, CBTR
1008DCIS≤5–10IBTR-S
509DCIS and LCIS≤ 5IBTR-S
363LCIS≤ 5IBTR-S
380LCIS≤ 5IBTR-S
799LCIS≤ 5IBTR-S
929LCIS≤ 5IBTR-S
996LCIS≤ 5IBTR-S
462LCIS≤ 5IBTR-S
803LOB INV and LCIS5–10IBTR-S
902LCIS5–10IBTR-S
942LCIS5–10IBTR-S
614DCIS10+IBTR-S
843LOB INV and LCIS5–10IBTR-S
816LCIS≤ 5IBTR-S
828NOS INV and DCIS5–10IBTR-S
539LOB INV5–10IBTR-S
300LOB INVb and LCISc≤ 5,b ≤5cIBTR-S,b CBTRc
955DCIS,d DCISe5–10,c 5–10eIBTR-S × 2
312LCIS10+IBTR-D
602DCIS,b and INVcf5–10,b 5–10cIBTR-S,b CBTRc
788LOB INV≤ 5CBTR
1005MUC INV≤ 5CBTR
829DCIS and LCIS≤ 5CBTR
126DCIS,d DCISe10+,d 10+eCBTR × 2
127NOS INV and DCIS10+CBTR
984LOB INV5–10CBTR
1014LOB INV (tubular)5–10CBTR
404INVf10+CBTR
506LOB INV5–10CBTR
864LOV INV and DCIS10+CBTR
860LOB INV, LOB INV≤ 5,c ≤ 5bCBTR, IBTR-S

For all types of IBTR, including the invasive form, recurrence was only slightly more common within the first 5 years of observation than after that interval (15 IBTRs vs. 11 IBTRs, respectively). Of the 9 IBTRs that were invasive, almost equal numbers occurred within 5 years and after 5 years (5 IBTRs vs. 4 IBTRs, respectively). The number of noninvasive IBTRs during the first 5 years and after the first 5 years also was almost equal, with 10 IBTRs within 5 years and 7 IBTRs after that period.

All CBTRs were less frequent than IBTRs, occurring in only 14 patients (7.8%) in the entire cohort, including the 4 patients discussed above with metachronous and synchronous IBTRs. However, the frequency of patients in the total cohort who had invasive carcinomas that were CBTRs (10 patients; 5.6%) was similar to the frequency noted for IBTRs (9 patients; 5.0%). The time to recurrence for invasive carcinomas that were CBTRs differed from the time noted for IBTRs, occurring after 5 years in 7 patients (70.0%) who had CBTRs compared with 4 patients (44.4%) who had IBTRs. Four of the former recurrences occurred after 10 years. The lobular histologic type of invasive CBTR was evident in 6 of 8 patients (75.0%) for whom pathologic material was available for review. The remaining four CBTRs included of DCIS only in two patients, a combination of DCIS and LCIS in one patient, and LCIS only in one patient. Three of the 4 CBTRs of noninvasive carcinoma occurred within 5 years

Table 3 reveals the frequencies and average annual rates per 100 patient-years according to the grades of LCIS for all IBTRs and the combined total of IBTRs that consisted of invasive carcinoma and DCIS with or without a component of LCIS. The average annual rate for patients with Grade 3 LCIS was high (1.85) and was similar to the average annual rate for 15 patients who had Grade 2 LCIS (1.83). The rate for 4 patients with Grade 1 lesions was comparatively low (0.55). These differences did not reach statistical significance (P = 0.075). However, the rates for all IBTRs were statistically significant when patients who had Grade 2 and 3 LCIS together were compared with patients who had Grade 1 LCIS (P = 0.023) (Table 3). The grade of LCIS was not significantly related to invasive IBTR, per se; however, as is also shown in Table 4, all grades of LCIS (P = 0.037) and combined Grade 2–3 LCIS versus Grade 1 LCIS (P = 0.010) were predictive for the endpoint of invasive IBTR and for IBTR consisting of DCIS with or without a component of LCIS. None of the other pathologic variables were significantly related to IBTR; and none of those variables, including grades of LCIS, were related to all CBTRs or to CBTRs consisting of invasive carcinoma or its total with IBTR consisting of DCIS.

Table 3. Relation of Lobular Carcinoma In Situ to Ipsilateral Breast Tumor Recurrence
GradeNo. of patients (%)aNo. of IBTRsbAverage annual rate per 100 patient-yrsP value
  • IBTR: ipsilateral breast tumor recurrence; DCIS: ductal carcinoma in situ.

  • a

    Frequency of grades/total cohort (n = 180)

  • b

    Frequency of IBTR/grade.

  • c

    Log-rank test comparing all grades of lobular carcinoma in situ.

  • d

    Log-rank test comparing Grade 2 with Grades 2 and 3 combined.

All IBTRs
 163 (34.6) 4 (6.3)0.550.075,c 0.023d
 280 (44.4)15 (18.7)1.83
 337 (20.6) 7 (18.9)1.85
IBTR: Total of invasive IBTRs and DCIS alone
 1 1 (1.5)0.140.037,c 0.010d
 210 (12.5)1.22
 3 5 (13.5)1.32
Table 4. Lobular Carcinoma In Situ: Mortality
DeathsNo. of patients (%)
  • SLE: systemic lupus erythematosus; ALS: amyotrophic lateral sclerosis.

  • a

    Total cohort (n = 180).

  • b

    Includes one patient with invasive ipsilateral breast tumor recurrence and one patient with invasive contralateral breast tumor recurrence.

  • c

    Includes one patient with contralateral breast tumor recurrence.

  • d

    No evidence of breast cancer.

Total16 (9)a
Breast carcinoma 2 (1)b
Heart disease3c
Second primary3d
Pulmonary disease2d
SLE1d
ALS1d
Unknown4d

Table 4 summarizes the 16 deaths observed in the cohort of 180 patients with LCIS. There were only 2 patients (1.1%) in the entire cohort whose deaths were considered to be related to breast carcinoma. One of those patients had an invasive IBTR, and the other patient had an invasive CBTR. The remaining 14 deaths were attributed to heart disease (n = 3), a second primary malignancy (n = 3), pulmonary disease (n = 2), systemic lupus erythematosus (SLE) (n = 1), and amyotrophic lateral sclerosis (ALS; n = 1). For 4 patients, the cause of death was unknown, but the records revealed that these patients, as well as 9 of 10 patients who died of causes other than breast carcinoma, failed to reveal any evidence of breast carcinoma. The remaining patient who died of ischemic heart disease had a previous CBTR. The infrequency of deaths related to LCIS precluded analysis of their correlation with the pathologic features studied.

DISCUSSION

The overall incidence rates of IBTR and of invasive carcinoma in the ipsilateral breast at 12 years after excision of LCIS remains low: the latter incidence is approximately 50% of that observed in the updated total of 17 studies encompassing a total of 917 patients with LCIS who were treated with local excision only and were followed for approximately 12 years. Whether this dichotomy will continue after subsequent periods of observation remains to be demonstrated. It is noteworthy that the incidence of IBTR consisting of invasive carcinoma was almost identical for the first 5-year period and for the subsequent 7-year period of evaluation in the current cohort. The reasons for the differences in frequency of subsequent IBTRs consisting of invasive carcinoma in our experience and that of others are uncertain. Page et al.12 have suggested that low estimates of this event may be due to the inclusion of examples of ALH in such studies, with their purportedly lower frequency compared with LCIS for the subsequent development of ipsilateral invasive carcinoma. However, Wheeler et al.,13 using a definition of ALH comparable to the one used by Page et al.,12 found a higher frequency. Surprisingly, Page et al.,14 in a more recent update of their previous data that included additional samples of ALH, found that IBTRs of invasive carcinoma at 10 years were approximately 3 times more common compared with our findings. It should be noted, however, that at least 10% of their database contained examples of ALH with ‘well-developed’ forms of LCIS. Page et al. used the designations LCIS, LN, ALH, and another somewhat vague term, atypical lobular carcinoma, synonymously. Although our results suggest that Grade 1 LCIS may be regarded as nosologic verification of ALH as an entity, we continue to consider the distinction of ALH from LCIS highly subjective or, perhaps, expressed more easily in words than in actual practice. Because of this subjectivity, we have chosen not to discriminate between the two entities. It is our practice to indicate that Grade 1 LCIS may be regarded by some as ALH; however, in our experience, it has not been deemed predictive of an increase in the incidence of subsequent recurrence of invasive carcinoma.

The low frequency of invasive IBTRs in our cohort may be related to the ‘free’ excision margins universally observed in our specimens of LCIS, as we suggested previously.1 This feature has received little attention in studies relating to the outcome of patients after local excision of LCIS. Schnitt and Morrow15 considered this suggestion to be ‘counterintuitive’, because LCIS is multicentric and bilateral, and these are well-accepted features of LCIS. However, those authors failed to recognize that invasive carcinomas16 and DCIS17 of the breast also may be multicentric and bilateral, yet clear margins are considered mandatory in their surgical treatment. Furthermore, both of the tumor types described above, as well as LCIS, as demonstrated in this study, most frequently appear in the same quadrant as the index lesion. Indeed, recognition of this important biologic feature minimizes the clinical significance of the multicentricity of LCIS, as it does with DCIS or invasive carcinoma. Finally, evidence that LCIS is a precursor (vide infra) supports the view that margin status may play a role in the treatment of patients with LCIS. The similarity in frequency of invasive carcinoma in IBTRs and CBTRs and the later time of occurrence of the latter, of course, do not relate to the status of the margins of the index lesion. However, it should be noted that the need for free margins that we intimated previously and above was based on inference rather than firm data. At this time, we do not consider it warranted to advocate free margins as a routine consideration in the surgical treatment of patients with LCIS, particularly if it would interfere with cosmesis.

Although the designation LN18 has merit, it should be recognized that the term was advocated as a surrogate for the designation LCIS, because it was hoped that elimination of the word carcinoma might dissuade the use of mastectomy for its surgical treatment. The decision also was prompted by the authors' observations of an unusually low frequency of recurrences and mortality after removal of the index lesion. Whether the margins were free is uncertain. This view prompted many to regard the designation LN as a risk factor rather than a precursor. In addition, in support of a risk factor was the relatively high frequency of invasive and noninvasive CBTR. However, this latter decision appears unwarranted, because such CBTRs may arise from the well recognized contralateral occurrence of LCIS in patients with ipsilateral disease. Conversely, as indicated above, recognition that the invasive IBTRs in the vast majority of instances occur at the same site as the index LCIS lesion and that the very high frequency of its lobular invasive histologic type are more in keeping with its nature as a precursor, a view supported by other studies,9, 16, 19 including recent ones that revealed a similar lack of reaction to E-cadherin in LCIS and lobular invasive carcinoma.9 The high frequency of IBTRs and CBTRs of the latter histologic phenotype is approximately two-to-three times more frequent than that cited in the literature. The reason for such a dichotomy is not certain. The diagnostic experience of one of the authors (E.R.F.) suggests that this may be related in large part to the infrequent practice of histologic subtyping of invasive breast carcinoma as well as the failure to recognize a tubulolobular form.20 In any event, the frequency of lobular invasive carcinoma noted in this report, and even that recorded by others, far exceeds the frequency of only 10% of lobular invasive carcinoma found in its pure or mixed forms in 1000 consecutive patients with invasive breast carcinoma unrelated to LCIS21; frequencies that obviously are more than coincidental. The above considerations have prompted us to retain the designation LCIS rather than LN. This view regarding the nomenclature for LCIS is supported by Lishman and Lakhani22 in their recent review of ALH and LCIS.

There are many similarities concerning the biologic nature of LCIS and DCIS, as noted above. However, recurrences are less common and appear later in patients who have LCIS compared with DCIS, indicating that the natural history of the former is more indolent. Both present an apparent conundrum in that, despite recurrences of invasive carcinomas, mortality rates are very low (1.6% at 8 years for DCIS17; 1.1% at 12 years for LCIS). However, it should be recognized that follow-up for many of the invasive IBTRs and CBTRs is relatively short after their occurrence.

The lack of a prospective randomized trial related to therapy for patients with LCIS prevents us from drawing more absolute conclusions concerning such therapy. Nevertheless, our data have provided some useful information in this regard. The infrequency of the development of invasive carcinoma after local excision of LCIS, almost exclusively located in the ipsilateral breast at the site of excision of the index lesion, as noted above, and its very low mortality rate are supportive of views that were suggested several decades ago by two of the authors (E.R.F. and B.F.)23 advocating the use of breast conservation and observation rather than mastectomy for the surgical management of patients with LCIS. There now is a general consensus among surgical oncologists to treat LCIS with local excision and subsequent surveillance only,24 a surgical view we support, as indicated above. It generally is agreed that,25–27 with rare exception,28 the presence of LCIS accompanying invasive carcinoma has no effect on local breast recurrence or survival rates after patients undergo surgical excision and local breast irradiation, which also militates against the need for mastectomy for its treatment. There is no unanimity concerning the treatment of a patient with a core biopsy that contains only LCIS, which is a rare event. Some investigators29, 30 advocate subsequent excision, whereas others31–33 believe that it is unnecessary. One group of investigators34 considers it necessary only when there also is independent DCIS or features of an ‘overlap’ of both LCIS and DCIS, a lesion we designate as DLCIS and for which we have been advising treatment with excision and local breast irradiation. We personally have not encountered any recurrences after this treatment for DLCIS, but a more formal study of its behavior is necessary, along with further pathologic examination of this lesion with E-cadherin staining, which also may represent a significant guide to treatment because of its apparently clear delineation of ductal and lobular lesions. At least currently, unlike DCIS, patients with LCIS are not treated with irradiation after undergoing local excision. The results of at least 1 comprehensive study of E-cadherin9 in this regard revealed a very high degree of sensitivity for a positive reaction in patients with the ductal histologic phenotype (97.7%) and for LCIS with its lack of staining (94.0%). This high level of concordance between immunohistochemical staining with E-cadherin and the histologic phenotypes of LCIS and DCIS suggests that the former may be most appropriate only for those instances in which there is uncertainty concerning their histopathologic distinction.

We are unaware of any scrutable information concerning the effects of irradiation on LCIS per se. Our pathologic study35 of the effects of preoperative chemotherapy revealed that LCIS is morphologically unaltered by this treatment method, suggesting that radiation may be ineffective for the treatment of LCIS. In contrast, the universal presence of estrogen and progesterone receptors in cells of LCIS1 indicates the propriety of estrogen antagonists for its treatment according to a preventative regimen that has demonstrated at least a twofold reduction in the development of invasive breast carcinoma for patients with LCIS.36 Therefore, it appears to be worth considering the administration of estrogen antagonists to patients with Grade 2 and 3 LCIS, but surveillance is recommended only for patients with Grade 1 disease (ALH).

Ancillary