Lobular neoplasia (LN) is a term used to encompass a spectrum of atypical proliferative lesions that includes atypical lobular hyperplasia (ALH) and lobular carcinoma-in-situ (LCIS). LCIS was initially considered to be a form of mammary carcinoma and as such was treated with mastectomy.1 However, long-term results of clinical follow-up studies suggested that LCIS and ALH represented markers of generalized increased breast cancer risk, not true precursor lesions, and surgery fell out of favor.2 – 4 Patients with ALH have a relative risk of breast cancer that is 4- to 5-fold higher than the general population, and patients with LCIS have a relative risk as high as 8- to 10-fold.

LN lacks a distinct radiographic correlate, and thus it is frequently diagnosed as an incidental finding in otherwise benign breast biopsy specimens, with a reported incidence ranging from 0.3 to 4 %.3,5 – 9 In the era of image-guided core biopsy (CB), a diagnosis of LN led to concern that the imaging abnormality in question was potentially undersampled, and numerous studies have evaluated the upgrade rate to ductal carcinoma-in-situ (DCIS) or invasive breast cancer after surgical excision of LN.9,10 These rates vary widely, with some studies reporting no upgrades and others reporting upgrades in up to 50 % of patients.11 – 13 Unfortunately, most early studies were retrospective and described small numbers of patients, not all of whom underwent excision.6,8,11 – 18 In some studies, LN in the CB was accompanied by another lesion, which by itself would warrant excision (such as atypical ductal hyperplasia), and in some reports, cases with pleomorphic LCIS or radiographic-pathologic discordance (e.g., ALH or LCIS on CB but a highly suspicious mass on imaging studies) were included.8,14,19 With the exception of a study by Irfan and Brem, many of the earlier studies also failed to report the Breast Imaging Reporting and Data System (BI-RADS) classification of the initial imaging lesion.6 – 8,11 – 21

Given the limitations of the available literature, current National Comprehensive Cancer Network guidelines remain conservative and recommend excision after a CB diagnosis of LCIS.22 However, among providers, this issue remains a matter of debate. The purpose of this study was to determine the upgrade rate to DCIS or invasive cancer after a CB diagnosis of pure LN in a single-arm multicenter prospective study.

Materials and Methods

The protocol received institutional review board approval at the Dana Farber/Harvard Cancer Center in November 2004 and was expanded in 2012 to included five additional sites in the Translational Breast Cancer Research Consortium (TBCRC) (Table 1). Eligibility criteria are listed in Fig. 1. In all cases, the breast imaging-detected lesion had a BI-RADS score of 4 or lower, and all pathologic diagnoses of LN were deemed concordant with imaging findings. Image-guided core biopsies were performed as per institutional standards; neither device type nor gauge were standardized for the study. Eligible patients signed informed consent, completed registration, and underwent surgical excision. Hematoxylin and eosin-stained sections (and in some cases immunostains for E-cadherin) were evaluated by the local pathologist to establish the diagnosis and subsequently underwent central pathology review by two breast pathologists (L.G., S.J.S.).

Table 1 Accrual by translational breast cancer research consortium site
Full size table
Fig. 1
figure 1

Eligibility criteria. ALH atypical lobular hyperplasia, LCIS lobular carcinoma-in-situ

Full size image

Statistical Considerations

The primary objective was to estimate the frequency with which LN found on core needle biopsy (by local pathology report) was upgraded to invasive breast cancer or DCIS on surgical excision. We aimed to accrue 78 patients based on the desire to use a decision rule with the power to differentiate between a true upgrade rate of 5 versus 15 %. The protocol-specified decision rule stated that if ≤7 of 78 patients were upgraded, we would conclude that the chance of upgrade on excision was too low to continue research on this subject (and that we would not recommend surgical excision for LN). Conversely if ≥8 patients were upgraded, we would continue to investigate which patients required surgical excision. With this design, if the true probability of upgrade was 5 %, there would be a 4 % chance that we would continue research on this issue, and if the true probability of an upgrade was 15 %, there would be a 91 % chance that we would continue to do research on this issue. This decision rule and the calculation of 95 % confidence intervals were based on the exact binomial distribution.

Results

In total, 79 patients were registered on the study. At the time the 78th patient was registered, it was known that one of the previously registered patients was unable to undergo surgical excision due to concomitant illness; the accrual of the 79th patient was thus permitted. An additional patient cancelled surgery and rescheduled to a date long after her CB, prompting her treating physician to remove her from the study. Thus, 77 of 79 patients who underwent surgical excision per protocol guidelines were included in this analysis. Patient characteristics of the 77 eligible participants are listed in Table 2. Median patient age was 51 years (range 27–82 years), and median time between CB and surgical excision was 2 months (range 0.5–15 months). Per local pathology interpretation, the CB diagnosis was ALH in 49 patients (64 %), LCIS in 17 (22 %), and both ALH and LCIS in 11 (14 %) (Table 3).

Table 2 Characteristics of study cohort (n = 77)
Full size table
Table 3 Summary of local pathology and central pathology interpretation of initial CB results
Full size table

Primary End Point

On the basis of local pathology review, the diagnosis of 2 of 77 patients was upgraded after surgical excision: one patient was found to have a grade I invasive tubular carcinoma (patient 6), and another patient was found to have intermediate nuclear grade DCIS (patient 59). Therefore, the upgrade rate based on the local pathology diagnosis was 2 (3 %) of 77 (95 % CI 0.3–9). The case upgraded to tubular carcinoma (patient 6) represented an magnetic resonance imaging–guided CB diagnosis of LCIS. The biopsy was performed for an area of abnormal enhancement in a 49-year-old patient undergoing magnetic resonance imaging screening because of a strong family history of breast cancer. In the case upgraded to DCIS, the original mammographic abnormality represented an area of calcifications spanning 8 mm in a 49-year-old woman with distant family history of postmenopausal breast cancer. The CB revealed LCIS, and the subsequent excision demonstrated solid, cribriform, and micropapillary DCIS of intermediate nuclear grade. Of the 18 excision slides, DCIS was present in 3 slides, one of which had CB site changes.

Central Pathology Review

CB slides were available for central pathology review in 76 of 77 cases, and LN was confirmed in 74 cases (96 %). In one case (patient 16), the local pathology report diagnosed LCIS in the CB; however, no LCIS (or ALH) was identified on central pathology review, and patient 6 (whose disease was upgraded to tubular carcinoma at excision per local pathology) was found to harbor a focus of tubular carcinoma in the CB specimen that was similar to that identified in the surgical excision. In a post hoc analysis, based on the 74 cases for which LN and absence of cancer in the CB could be confirmed by central pathology review, there was only 1 upgraded case (1 %, 95 % CI 0.01–7).

Discussion

Although many studies have looked at upgrade rates to carcinoma after excision of LN, various methodologic issues, including their retrospective design with inherent selection bias, failure to describe the BI-RADS classification and/or radiographic-pathologic concordance, and inclusion of pleomorphic LCIS, have made it difficult to interpret the findings. Table 4 summarizes the findings of the more contemporary, larger (comprising more than 20 cases) series on this subject and demonstrates that upgrade rates remain highly variable, ranging from 0 to 27 %.7 , 14,19 – 21,23 – 31 In our prospective study with strict eligibility criteria and central pathology review, we found an upgrade rate for concordant, BI-RADS 4 or lower, pure LN lesions to be 3 % based on the local pathology diagnosis and 1 % based on central pathology review. These findings are in agreement with more recent studies reporting the presence of concordance for the LN lesions in question, and they demonstrate that routine excision for a concordant CB diagnosis of pure LN is not warranted.

Table 4 Summary of existing literature on upgrade rates found on excision of lobular neoplasia diagnosed on core biopsy
Full size table

Lewis et al. reviewed 201 cases of LN and reported an upgrade rate of 13 %; however, BI-RADS and/or pathologic-radiographic concordance information for these cases was not available.26 Niell et al. reported an upgrade rate of 11 % in 63 concordant LN lesions.28 Shah-Khan et al., however, reported an upgrade rate of 2 % among 101 LN lesions, 90 % of which were concordant with imaging findings.30 Rendi et al. reported three upgrades (4 %) among 68 BI-RADS 4 lesions, without specific concordance information provided.29 Forty of these patients were designated as high risk on the basis of their family history, and all three upgrades were in high-risk patients. Although pathologic-radiographic concordance is likely more important than pure BI-RADS classification, studies reporting both have consistently reported the lowest upgrade rates. Atkins et al. reported no upgrades among 38 concordant lesions.23 Chaudhary et al. reported an upgrade rate of 3 % among 87 concordant lesions.25 Finally, similar to our study, Murray et al. conducted a prospective study of 80 excisions after a CB diagnosis of pure LN and reported two upgrades (3 %) among 72 concordant cases.32 In contrast, in the discordant group, the upgrade rate was 3 (38 %) of 8. Among the three upgraded discordant cases, two were thought to have had insufficient sampling of the target lesion by CB, both of which were upgraded to DCIS, and one represented a spiculated mass, classified as BI-RADS 5, which was upgraded to invasive carcinoma.

As demonstrated here and in other studies of LN, there is a degree of interobserver variability for a diagnosis of ALH versus LCIS, and the distinction is to some degree subjective and dependent on the amount of submitted material. In a series of breast specimens sent for formal consultation at the Universidade Federal de Minas Gerais School of Medicine in Brazil, Gomes et al. reported agreement for a diagnosis of ALH in 8 (47 %) of 17 cases and an agreement for a diagnosis of LCIS in 31 (69 %) of 45 cases.33 In our series of CB specimens, among 49 cases diagnosed as ALH by the local pathologist, there was agreement by central pathology review in 46 cases (94 %), and among 28 cases diagnosed as LCIS, with or without ALH, there was agreement in 14 cases (50 %). Although clinical follow-up studies demonstrate a greater magnitude of subsequent breast cancer risk associated with LCIS than with ALH, clinical management algorithms remain the same; therefore, the clinical significance of the distinction between these two lesions on CB may be less relevant. Patients with ALH and LCIS should be counseled regarding their cumulative increased risk of subsequent breast cancer, which is conferred equally to both breasts and significantly reduced with the use of chemoprevention.34

Our study, with its unique prospective design, predetermined statistical end points, strict eligibility criteria, and central pathology review, represents the most comprehensive evaluation of this subject to date. Although the confidence intervals for the upgrade rates in our study range from 0 to 9 %, the predetermined statistical end points were clear, and our goal was to differentiate between a true upgrade rate of 5 versus 15 %. Considering that a BI-RADS 3 designation implies a less than 2 % risk of upgrade to a cancer diagnosis, our findings suggest that a concordant CB diagnosis of LN is amenable to follow-up imaging at 6 months rather than excision, as the upgrade rates are similar.35 Prospective data to document the yield of 6-month follow-up imaging in patients with concordant diagnosis of pure LN on CB are not available, yet it is our opinion that this practice should be considered in the management algorithms for patients with pure LN who do not undergo excision to allow multidisciplinary groups to collect this information and report their outcomes.

In conclusion, in this prospective multi-institutional study of patients with BI-RADS 4 or lower concordant lesions with LN on CB, we found a very low upgrade rate to invasive cancer or DCIS on excision. These results indicate that routine excision after a diagnosis of LN on CB is not indicated. Patients with a concordant CB diagnosis of pure LN should be counseled regarding their increased lifetime risk of breast cancer and offered participation in high-risk surveillance programs, including a full discussion of lifestyle interventions (diet, exercise) as well as the risks and benefits of chemoprevention.