Purpose:

The prognostic utility of Breast Cancer Index (BCI) for risk assessment of overall (0–10 years), early (0–5 years), and late (5–10 years) distant recurrence (DR) in hormone receptor–positive (HR+) invasive lobular carcinoma (ILC) was evaluated.

Experimental Design:

BCI gene expression analysis was performed blinded to clinical outcome utilizing tumor specimens from patients with HR+ ILC from a multi-institutional cohort. The primary endpoint was time to DR. Kaplan–Meier analyses of overall, early, and late DR risk were performed, and statistical significance was evaluated by log-rank test and Cox proportional hazards regression. The prognostic contribution of BCI in addition to clinicopathologic factors was evaluated by likelihood ratio analysis.

Results:

Analysis of 307 patients (99% ER+, 53% T1, 42% N+, 70% grade II) showed significant differences in DR over 10 years based on BCI risk categories. BCI low- and intermediate-risk patients demonstrated similar DR rates of 7.6% and 8.0%, respectively, compared with 27.0% for BCI high-risk patients. BCI was a significant independent prognostic factor for overall 10-year DR [HR = 4.09; 95% confidence interval (CI), 2.00–8.34; P = 0.0001] as well as for both early (HR = 8.19; 95% CI, 1.85–36.30; P = 0.0042) and late (HR = 3.04; 95% CI, 1.32–7.00; P = 0.0224) DR. In multivariate analysis, BCI remained the only statistically significant prognostic factor for DR (HR = 3.49; 95% CI, 1.28–9.54; P = 0.0150).

Conclusions:

BCI is an independent prognostic factor for ILC and significantly stratified patients for cumulative risk of 10-year, early, and late DR. BCI added prognostic value beyond clinicopathologic characteristics in this distinct subtype of breast cancer.

Translational Relevance

Invasive lobular breast cancer (ILC) is a heterogenous disease with diverse clinical outcomes and considerable risk of late metastasis. Enhanced molecular approaches that provide information on the tumor biology of this distinct subtype of invasive breast cancer are needed to inform prognosis and individualized treatment. In this study, Breast Cancer Index (BCI) significantly stratified patients with ILC into risk groups based on risk of overall 10-year, early, and late distant recurrence. BCI provided distinct and additive prognostic information beyond clinicopathologic factors and reclassified a meaningful number of clinically low-risk tumors as high genomic risk and clinically high-risk tumors as low genomic risk. These findings demonstrate that BCI is an independent prognostic factor for ILC and suggest its potential role to enhance individualization of ILC treatment.

Invasive lobular carcinoma (ILC) is the second most common histologic subtype of invasive breast cancer and accounts for approximately 10% to 15% of all breast cancers (1). ILC displays distinct pathologic, molecular, and clinical characteristics compared with those of the more commonly diagnosed invasive ductal carcinoma (IDC; refs. 1–5). Loss of E-cadherin expression is a defining characteristic of ILC and results in reduced cell–cell adhesion and tumor morphology in which cells invade tissues in a chain-like single-file manner (1). ILC tumors are predominantly estrogen receptor positive (ER+), HER2 negative (HER2−), and of low grade and low proliferative index (1–5). Although these tumor characteristics are generally associated with favorable prognoses, ILC tumors have an increased risk of late distant recurrence (DR) and can display aggressive metastatic behavior associated with poorer long-term outcomes when compared with stage-matched IDC (2, 5–9). Despite the unique clinical challenges of ILC versus IDC, current clinical practice guidelines recommend similar treatment paradigms for both histologic subtypes (5, 10). Thus, there is an unmet medical need for enhanced approaches that interrogate underlying ILC tumor biology to better individualize treatment and long-term disease management (2, 11).

The Breast Cancer Index (BCI) is a gene expression–based signature that incorporates two functional biomarker panels: (i) the 2-gene ratio, HOXB13/IL17BR (H/I), and (ii) the 5-gene Molecular Grade Index (MGI). The BCI test is indicated for patients with early-stage, hormone receptor–positive (HR+) breast cancer and reports both a predictive and a prognostic result. The predictive component, BCI (H/I), reports a categoric prediction of high versus low likelihood of benefit from extended endocrine therapy (12–14), whereas the prognostic component, the BCI score, is based on the algorithmic combination of H/I and MGI and stratifies risk for overall (0–10 years) and late (5–10 years post-diagnosis) DR (12, 13, 15).

BCI prognostic models have been developed for both node-negative (N0) and node-positive (N+) disease (13, 16). The N0 prognostic model is based on gene expression alone and categorizes patients into low-, intermediate-, and high-risk groups (13), whereas the N+ prognostic model incorporates tumor size and grade with gene expression and dichotomizes patients with N+ tumors into low- and high-risk categories (16). BCI prognostic ability has been validated in multiple studies of breast cancer patients, which also included approximately 12% patients with ILC (13, 15–17). This study examines prognostic risk stratification of BCI specifically in ILC in a blinded multi-institutional analysis.

Study design and patient samples

In this retrospective study, formalin-fixed paraffin-embedded (FFPE) tumor specimens from 376 patients diagnosed with ILC between 1992 and 2011 were collected from The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University (JHU, N = 111), Dana-Farber Cancer Institute (DFCI, N = 95), Massachusetts General Hospital (MGH, N = 76), and the Pitt Biospecimen Core (PBC) at the University of Pittsburgh and University of Pittsburgh Medical Center (UPMC) Magee-Womens Hospital (UPMC, N = 94). Inclusion criteria included HR+ (≥1% positive stained cells for ER or progesterone receptor (PR) based on ASCO/CAP guidelines; ref. 18), stage I to III based on AJCC 7th Ed. (19), any nodal status, and pure lobular or mixed lobular/ductal histology. HER2 status was determined according to standard procedures at the time of patient diagnosis by IHC and FISH analysis. Patients were excluded if treated with neoadjuvant therapy, missing clinical information (i.e., tumor size or nodal status), or inadequate survival follow-up.

Breast cancer index assay

BCI gene expression analysis of FFPE primary tumor specimens was performed blinded to clinical outcome as described previously (13). Briefly, macro-dissection was performed on FFPE sections to enrich tumor content followed by RNA extraction. Patients were excluded if their specimen had less than 40% tumor content according to assay requirements. Total RNA was reverse transcribed, and the resulting cDNA was pre-amplified by PCR using the PreAmp Master Mix Kit (Thermo Fisher Scientific) prior to TaqMan RT-PCR analysis. Calculation of risk scores and categorical risk stratification were performed using prespecified cut points for the N0 and N+ prognostic models (12, 13). The BCI prognostic model for N0 has three reporting categories (low-, intermediate-, and high-risk) whereas the N+ model (BCIN+) has two (low- and high-risk). To examine risk stratification in the overall cohort, patients were divided into three risk groups using both BCI models. The low-risk group for the overall cohort combined the low-risk patients from the N0 (using BCI gene expression alone) and N+ subsets (using BCIN+). The intermediate-risk group consisted of intermediate-risk patients from the N0 subset. The high-risk group combined the high-risk patients from the N0 and N+ subsets. For simplification, “BCI” throughout the manuscript refers to the combined risk groups in the overall cohort, which integrates prognostic risk categories from both the N0 (BCI gene expression alone) and N+ (BCIN+, gene expression with tumor size and grade) models. Calculation of a BCI predictive score [BCI (H/I)] for response to extended endocrine therapy was conducted using a pre-specified and validated assay cut-point to categorize patients as either BCI (H/I)-high (likely to respond) or BCI (H/I)-low (unlikely to respond; ref. 20).

Pilot study comparing laser capture microdissection with standard macro-dissection

To determine the analytical impact of tissue stroma and further define sample requirements based on cellularity, 23 tumor samples were analyzed in a pilot study comparing tissue processing by either the standard manual macro-dissection or laser capture micro-dissection (LCM). It was hypothesized that LCM would enrich for invasive tumors cells, thus impacting BCI results. Pre-specified criteria to determine equivalence between dissection methods were based on an empirically defined ≥80% concordance using the BCI assay variability threshold of two SDs of BCI score difference (1.2 BCI units).

Study endpoints

The primary endpoint was time to DR, defined as the time from diagnosis to first DR. Contralateral disease, local and regional recurrence, and other second primary cancers were not considered as events nor censored. The survival analysis was censored at 10 years. The primary objective was to evaluate the prognostic performance of BCI for risk of overall 10-year, early (0–5 years), and late (5–10 years) DR. The secondary objective was to evaluate the prognostic performance of BCI in clinically relevant subsets.

Statistical considerations and analyses

Kaplan–Meier analysis was used to estimate the overall 10-year, early (0–5 years), and late (5–10 years) DR risk for BCI risk groups. The log-rank test was used to test the equality of survival curves and a univariate Cox proportional hazards regression model was used to estimate HRs and the associated 95% confidence intervals (CI). A multivariate Cox proportional hazards regression model was used to evaluate whether BCI provided independent prognostic information in addition to standard clinicopathologic factors that were significant in the univariate analysis (age, adjuvant endocrine therapy, adjuvant chemotherapy, tumor size, tumor grade, nodal status) using Wald tests. Likelihood ratio statistics (ΔLR-χ2) were calculated on the basis of Cox proportional hazards regression models to measure the relative contributions of BCI gene expression alone or together with tumor size and grade. A two-sided P value of less than 0.05 was considered statistically significant. All analyses were performed using R statistical package (version 3.5.2, http://www.r-project.org).

BCI testing and clinicopathologic characteristics

Prior to standard BCI testing, a pilot study was conducted to evaluate the impact of tumor cellularity on BCI results comparing LCM to standard manual macro-dissection in 23 ILC tumor specimens. Tumor content in selected areas ranged between 15% to 85% (IQR: 30%) with two cases being excluded due to insufficient tumor quantity and three cases due to insufficient RNA yield in the LCM sample (Supplementary Fig. S1). In the remaining 18 paired samples, a high concordance (86%) in BCI scores was observed across both tissue dissection methods, with three samples being above the predetermined concordance threshold of >1.2 BCI units' difference (Supplementary Fig. S2). The two methods of LCM and manual micro-dissection were considered equivalent based on the prespecified concordance threshold. On the basis of this data, subsequent analysis for the study was performed following the standard manual macro-dissection. BCI results were generated in 307 patients (JHU, N = 80; DFCI, N = 76; MGH, N = 76; UPMC, N = 75) of 376 HR+ evaluated cases. Samples were excluded (N = 69) due to previous neoadjuvant chemotherapy, a history of remote breast cancer, insufficient tumor content, or missing clinical information for an attrition rate of 18% (Supplementary Fig. S3). BCI assay failure rate was 0%.

Patient and tumor characteristics are summarized in Table 1. Of the 307 patients, 71% were ≥50 years old and 42% had N+ tumors, with 95% of tumors exhibiting pure ILC histology and 5% with mixed ILC and IDC histology. Among patients with available tumor information, 100% were ER+ (306/306), 94% were HER2− (224/239), 47% were classified as T2 or higher (143/306), 70% had moderate-grade tumors (202/290), 55% were stage II (167/306), 50% were treated with adjuvant chemotherapy (140/282), and 93% received adjuvant endocrine therapy (263/284). HER2 status was not available for 68 patients either because the biomarker was not routinely assessed in older cases, accounting for the majority with unknown HER2 status, or that the data were lost when clinical information was migrated from medical record databases. Among the 233 patients with specific information on the type of endocrine therapy, 35% received tamoxifen monotherapy, 35% received aromatase inhibitor (AI) monotherapy, and 30% received tamoxifen-AI sequential treatment.

Table 1.

Clinicopathologic characteristics of ILC cohort (N = 307).

Patients, n (%)
Age 
 <50 88 (29%) 
 ≥50 219 (71%) 
Primary surgery 
 Lumpectomy 130 (46%) 
 Mastectomy 148 (52%) 
 Other 7 (2%) 
 Unknown 22 
ER status 
 Positive 306 (100%) 
 Unknown 
PR status 
 Positive 273 (92%) 
 Negative 25 (8%) 
 Unknown 
HER2 status 
 Positive 15 (6%) 
 Negative 224 (94%) 
 Unknown 68 
Histological type 
 Lobular 292 (95%) 
 Mixed 15 (5%) 
T stage 
 T1 163 (53%) 
 T2 116 (38%) 
 T3 27 (9%) 
 Unknown 
Tumor grade 
 Well 59 (20%) 
 Moderate 202 (70%) 
 Poor 29 (10%) 
 Unknown 17 
Stage 
 Stage I 107 (35%) 
 Stage II 167 (55%) 
 Stage III 32 (10%) 
 Unknown 
Nodal status 
 N0 179 (58%) 
 N1 103 (34%) 
 N2 15 (5%) 
 N3 10 (3%) 
Adjuvant chemotherapy 
 No 142 (50%) 
 Yes 140 (50%) 
 Unknown 25 
Adjuvant endocrine therapy 
 No 21 (7%) 
 Yes 263 (93%) 
 Unknown 23 
Patients, n (%)
Age 
 <50 88 (29%) 
 ≥50 219 (71%) 
Primary surgery 
 Lumpectomy 130 (46%) 
 Mastectomy 148 (52%) 
 Other 7 (2%) 
 Unknown 22 
ER status 
 Positive 306 (100%) 
 Unknown 
PR status 
 Positive 273 (92%) 
 Negative 25 (8%) 
 Unknown 
HER2 status 
 Positive 15 (6%) 
 Negative 224 (94%) 
 Unknown 68 
Histological type 
 Lobular 292 (95%) 
 Mixed 15 (5%) 
T stage 
 T1 163 (53%) 
 T2 116 (38%) 
 T3 27 (9%) 
 Unknown 
Tumor grade 
 Well 59 (20%) 
 Moderate 202 (70%) 
 Poor 29 (10%) 
 Unknown 17 
Stage 
 Stage I 107 (35%) 
 Stage II 167 (55%) 
 Stage III 32 (10%) 
 Unknown 
Nodal status 
 N0 179 (58%) 
 N1 103 (34%) 
 N2 15 (5%) 
 N3 10 (3%) 
Adjuvant chemotherapy 
 No 142 (50%) 
 Yes 140 (50%) 
 Unknown 25 
Adjuvant endocrine therapy 
 No 21 (7%) 
 Yes 263 (93%) 
 Unknown 23 

The median follow-up for the overall population was 10 years (7.4, 14.4, 10.0, and 11.0 years for the JHU, DFCI, MGH, and UPMC cohorts, respectively). There were 41 DR events in the entire cohort (13% of patients), 63% of which occurred more than 5 years after diagnosis. For each BCI risk group (BCI low/intermediate or BCI high), differences in clinicopathologic variables were evaluated (Supplementary Table S1). As expected, the group of patients classified as BCI low/intermediate were more likely to be T stage T1 (60% vs. 46%; P = 0.0496), well-differentiated (37% vs. 1%; P < 0.0001), stage I (53% vs. 14%; P < 0.0001), and N0 (87% vs. 24%; P < 0.0001; Supplementary Table S1).

BCI is a significant prognostic factor in ILC

Significant differences in outcome based on BCI risk stratification were observed. In the overall cohort (N = 307), BCI classified 40% of patients (N = 123) as low-risk with a 10-year DR rate of 7.6% (95% CI, 2.0%–12.9%), 14% of patients (N = 44) as intermediate-risk with a 10-year DR rate of 8.0% (95% CI, 0.0%–16.4%), and 46% of patients (N = 140) as high-risk with a 10-year DR rate of 27.0% (95% CI, 18.3%–34.9%; Fig. 1A). The low- and intermediate-risk patients displayed similar rates of DR, and therefore were combined into a single low/intermediate-risk group with a 10-year DR rate of 7.8% (95% CI, 3.0%–12.4%; Fig. 1B). BCI significantly stratified patients with ILC into high- and low/intermediate-risk groups based on overall 10-year (HR = 4.09; 95% CI, 2.00–8.34; P = 0.0001), early (0–5 years) (HR = 8.19; 95% CI, 1.85–36.30; P = 0.0042), and late (5–10 years) DR (HR = 3.04; 95% CI, 1.32–7.00; P = 0.0224; Fig. 1B–D). Of the 248 patients that remained free of DR for at least 5 years, 57% were classified as low/intermediate-risk with a late DR rate of 6.5% (95% CI, 2.0%–10.9%) compared with a DR rate of 18.7% (95% CI, 10.4%–26.3%) in the high-risk group (Fig. 1D). Overall, the low/intermediate-risk group had a favorable prognosis with DRs predominantly occurring late (post-5 years; 6.5%; 95% CI, 1.3%–7.0%) rather than early (1.4%; 95% CI, 0.0%–3.2%). In contrast, the high-risk group demonstrated a persistent and increasing risk of DR over the entire 10-year period (Fig. 1B–D). Results in patients that received adjuvant endocrine therapy (N = 263) were similar to the overall cohort (Supplementary Fig. S4). The overall cohort was then stratified by BCI (H/I) to evaluate the likelihood of benefit from extended endocrine therapy. Of the BCI low/intermediate-risk patients, 34% were classified as BCI (H/I)-high and predicted to benefit from extended endocrine therapy (Supplementary Fig. S5). In addition, 49% of the BCI high-risk patients were classified as BCI (H/I)-low and not predicted to benefit from extended endocrine therapy (Supplementary Fig. S5).

Figure 1.

Prognostic performance of BCI for overall 10-year, early (0–5 years), and late (5–10 years) DR rate for all patients in the lobular cohort utilizing (A) BCI risk stratification by three prognostic risk groups (low, intermediate, high) and (B–D) BCI risk stratification by two prognostic groups (low/intermediate, high).

Figure 1.

Prognostic performance of BCI for overall 10-year, early (0–5 years), and late (5–10 years) DR rate for all patients in the lobular cohort utilizing (A) BCI risk stratification by three prognostic risk groups (low, intermediate, high) and (B–D) BCI risk stratification by two prognostic groups (low/intermediate, high).

Close modal

In the N0 subset (N = 179), BCI gene expression classified 81% of patients as low/intermediate-risk and 19% as high-risk, and the 10-year DR rates were 8.2% (95% CI, 2.8%–13.3%) and 24.7% (95% CI, 6.5%–39.5%), respectively (HR = 3.85; 95% CI, 1.43–10.35; P = 0.0158; Supplementary Fig. S6A). In contrast, BCIN+ classified 17% of N+ patients (N = 128) into the low-risk group compared with 83% in the high-risk group (Supplementary Table S2), and the 10-year DR rates were 5.6% (95% CI, 0.0%–15.6%) and 27.9% (95% CI, 17.6%–37.0%), respectively (HR = 5.90; 95% CI, 0.80–43.64; P = 0.1419; Supplementary Fig. S6B).

Patients with well- and moderately-differentiated tumors accounted for 90% of the overall cohort, and within this group 53% were N0 and 47% were N+. Among these patients with well- and moderately-differentiated tumors, the 10-year rate of DR was 16.3% (95% CI, 11.0%–21.2%), and BCI stratified 45% of patients into a high-risk group with a 10-year rate of DR of 26.0% (95% CI, 16.6%–34.4%), and stratified 55% of patients into a low/intermediate-risk group with a 10-year rate of DR of 8.2% (95% CI, 2.8%–13.4%; HR = 3.78; 95% CI, 1.76–8.09; P = 0.0012). In addition, BCI significantly stratified patients for both early (HR = 11.64; 95% CI, 1.47–91.90; P = 0.0126) and late (HR = 2.78; 95% CI, 1.19–6.49; P = 0.0484) DR (Fig. 2A).

Figure 2.

Prognostic performance of BCI for overall 10-year, early (0–5 years), and late (5–10 years) DR rate for patients with (A) well- and moderately-differentiated tumors, (B) stage II and III tumors, and (C) chemotherapy treatment.

Figure 2.

Prognostic performance of BCI for overall 10-year, early (0–5 years), and late (5–10 years) DR rate for patients with (A) well- and moderately-differentiated tumors, (B) stage II and III tumors, and (C) chemotherapy treatment.

Close modal

In the 65% of patients with stage II and III breast cancer (36% N0, 64% N+), the overall 10-year DR rate was 29.5% (95% CI, 19.7%–38.1%) for the high-risk group and 10.5% (95% CI, 1.9%–18.3%) for the low/intermediate-risk group (HR = 3.37; 95% CI, 1.40–8.12; P = 0.0158). These patients with stage II and III disease could also be further stratified for both early (HR = 4.09; 95% CI, 0.91–18.27; P = 0.1352) and late (HR = 3.01; 95% CI, 1.01–8.95; P = 0.1139) DR (Fig. 2B) with 39% classified as low/intermediate-risk. Patients treated with chemotherapy were similarly stratified for overall 10-year (HR = 4.62; 95% CI, 1.60–13.36; P = 0.0081), early (P = 0.0246), and late (HR = 2.73; 95% CI, 0.89–8.38; P = 0.1876) DR risk (Fig. 2C).

BCI was prognostic in patients ≥50 years of age for overall (HR = 4.16; 95% CI, 1.88–9.23; P = 0.0007), early (HR = 4.97; 95% CI, 1.06–23.41; P = 0.0789), and late (HR = 3.89; 95% CI, 1.53–9.86; P = 0.0087) DR. Since patients under 50 years of age primarily had early DRs, BCI was significantly prognostic for early DR (P = 0.0414) but not late DR (P = 0.9235) within this age group (Fig. 3).

Figure 3.

Prognostic performance of BCI for overall 10-year, early (0–5 years), and late (5–10 years) DR rate for patients (A) <50 years old and (B) ≥50 years old, respectively.

Figure 3.

Prognostic performance of BCI for overall 10-year, early (0–5 years), and late (5–10 years) DR rate for patients (A) <50 years old and (B) ≥50 years old, respectively.

Close modal

BCI is an independent prognostic factor beyond clinicopathologic parameters

Univariate analysis in the overall cohort showed that adjuvant chemotherapy, tumor size, tumor grade, nodal status, and BCI each provided significant prognostic information for overall 10-year DR rate, whereas age and adjuvant endocrine therapy did not. However, in the multivariate analysis, after adjusting for other significant prognostic factors identified in the univariate analysis, BCI remained the only significant and independent prognostic factor for overall 10-year DR rate (HR = 3.49; 95% CI, 1.28–9.54; P = 0.015; Table 2).

Table 2.

Univariate and multivariate Cox regression analysis of prognostic performance of BCI for overall 10-year DR rate in the ILC cohort.

Univariate analysisMultivariate analysisa
VariableHRP valueHRP value
Age (years) 
 ≥50 vs. <50 1.82 (0.84–3.95) 0.129 — — 
Adjuvant endocrine therapy 
 Yes vs. no 0.53 (0.19–1.49) 0.229 — — 
Adjuvant chemotherapy 
 Chemo vs. no chemo 1.96 (1.03–3.74) 0.041 1.21 (0.58–2.51) 0.611 
Tumor size (cm) 
 T2–3 vs. T1 2.22 (1.17–4.22) 0.015 1.65 (0.84–3.24) 0.149 
Tumor grade (differentiation) 
 Moderate vs. well 5.06 (1.21–21.13) 0.026 2.09 (0.44–9.93) 0.352 
 Poor vs. well 9.33 (1.94–44.93) 0.005 2.76 (0.48–15.77) 0.254 
Nodal status 
 N+ vs. N0 2.24 (1.19–4.19) 0.012 0.77 (0.33–1.81) 0.545 
BCI 
 High vs. low/Intermediate 4.09 (2.00–8.34) 0.0001 3.49 (1.28–9.54) 0.015 
Univariate analysisMultivariate analysisa
VariableHRP valueHRP value
Age (years) 
 ≥50 vs. <50 1.82 (0.84–3.95) 0.129 — — 
Adjuvant endocrine therapy 
 Yes vs. no 0.53 (0.19–1.49) 0.229 — — 
Adjuvant chemotherapy 
 Chemo vs. no chemo 1.96 (1.03–3.74) 0.041 1.21 (0.58–2.51) 0.611 
Tumor size (cm) 
 T2–3 vs. T1 2.22 (1.17–4.22) 0.015 1.65 (0.84–3.24) 0.149 
Tumor grade (differentiation) 
 Moderate vs. well 5.06 (1.21–21.13) 0.026 2.09 (0.44–9.93) 0.352 
 Poor vs. well 9.33 (1.94–44.93) 0.005 2.76 (0.48–15.77) 0.254 
Nodal status 
 N+ vs. N0 2.24 (1.19–4.19) 0.012 0.77 (0.33–1.81) 0.545 
BCI 
 High vs. low/Intermediate 4.09 (2.00–8.34) 0.0001 3.49 (1.28–9.54) 0.015 

aOnly significant prognostic factors in the univariate analysis were included in the multivariate model.

To further examine the additive prognostic value of BCI gene expression versus tumor size and grade, likelihood ratio statistics (ΔLR-χ2) were calculated for the N0 and N+ subsets. As shown in Fig. 4, BCI gene expression was highly prognostic in both N0 (ΔLR-χ2 = 6.28) and N+ (ΔLR-χ2 = 6.82) patients. In the N0 subset, BCI was more prognostic than tumor size (ΔLR-χ2 = 2.52), but not more than tumor grade (ΔLR-χ2 = 7.74) or tumor size plus grade (ΔLR-χ2 = 10.05). BCI added independent prognostic information to tumor grade, tumor size, and tumor size plus grade, and conversely, tumor grade and tumor size plus grade added independent prognostic information to BCI gene expression. In the N+ subset, BCI provided greater prognostic information versus tumor size (ΔLR-χ2 = 0.02), grade (ΔLR-χ2 = 2.10), or tumor size plus grade (ΔLR-χ2 = 2.14), and added more prognostic information to tumor size and grade (ΔLR-χ2 = 5.35) than these variables added to BCI (ΔLR-χ2 = 0.67).

Figure 4.

Prognostic value of BCI (gene expression only) and improvement by incorporation of tumor size and grade as measured by the change in likelihood ratio statistic, χ2 (ΔLR-χ2) for patients with N0 (A) and N+ (B) tumors in the lobular cohort. Bars represent relative contributions of BCI (black) and clinicopathologic factors, including tumor size and/or grade (white).

Figure 4.

Prognostic value of BCI (gene expression only) and improvement by incorporation of tumor size and grade as measured by the change in likelihood ratio statistic, χ2 (ΔLR-χ2) for patients with N0 (A) and N+ (B) tumors in the lobular cohort. Bars represent relative contributions of BCI (black) and clinicopathologic factors, including tumor size and/or grade (white).

Close modal

Findings from this study, which investigated the prognostic ability of BCI in a multi-institutional cohort of patients with ILC treated with endocrine or chemo-endocrine therapy, demonstrate that BCI is a significant and independent prognostic factor in ILC and provides risk stratification for cumulative 10-year DR, as well as both early (0–5 years) and late (5–10 years) DR. Patients with ILC were stratified into a low/intermediate-risk group comprising 54% of the patients, with a DR rate of approximately 7.8%, and a high-risk group including 46% of patients with a DR rate of approximately 27.0%. Importantly, BCI also significantly stratified patients with ILC for late DR, with prognostic performance consistent across all clinical subsets examined. It has been previously reported that lymph node status, tumor size, age, S-phase, PR status, and ER status are significant clinical factors independently associated with recurrence and survival (21). In addition, increased tumor size and nodal status have been shown to be associated with risk of late DR (22). However, results from the current study underscore the utility of genomic risk assessment in ILC, as BCI classified 19% of patients with N0 tumors and 45% of patients with well/moderately-differentiated tumors as high-risk. Conversely, 17% of patients with N+ tumors and 39% of patients with stage II/III tumors were classified as low/intermediate-risk by BCI, indicating these patients have a favorable long-term prognosis despite their high-risk clinicopathologic features.

As ILC has a propensity for late DR, an apparent feature of BCI prognostic stratification is that DRs in low/intermediate-risk patients occurred almost entirely in the late follow-up period (post-5 years from diagnosis), whereas recurrences in high-risk patients showed a steady increase in cumulative risk over 10 years (Fig. 1). Similar recurrence patterns to those of the overall cohort were also observed in high- and low/intermediate-risk patients with well- and moderately-differentiated tumors (Fig. 2A), stage II/III tumors (Fig. 2B), and N+ tumors (Supplementary Fig. S6B). In addition, patients with ILC did not demonstrate any reduction in risk or shift in prognostic profile with chemotherapy treatment, although this should be interpreted with caution given the retrospective nature of this study (Fig. 2C).

Consistent with the distinct biological characteristics of ILC, there are notable differences between BCI risk stratification of patients with ILC versus IDC. Previous studies have shown that BCI stratified N0 IDC tumors into low-, intermediate-, and high-risk groups comprising approximately 55–60%, 25–30%, and 15–20% of patients, respectively (13, 15). For patients with N+ IDC, BCI stratified ∼20% as low-risk and ∼80% as high-risk (16). In contrast, only two risk groups were classified by BCI in ILC in patients with both N0 and N+ tumors. Thus, BCI identifies two clinical entities in ILC: a low/intermediate-risk category consisting of 54% of patients and a high-risk category comprising 46% of patients. Importantly, findings from this study indicate that tumor biology and genomic classification provide increased resolution to prognostication of ILC, which is highly heterogenous and therefore likely requires individualized assessment of gene expression to accurately assess DR risk. Although nodal status showed correlation with BCI risk groups, in multivariate analysis, BCI remained the only significant independent prognostic factor for risk of DR.

In addition to BCI, several other prognostic classifiers have been investigated in patients with ILC (22–28). Similar prognostic classifications to those reported here have been reported for the 70-gene MammaPrint assay (24) and EndoPredict (27), each of which classified patients with ILC into a high-risk and a low-risk category. The 97-gene Genomic Grade Index (GGI) outperformed histologic grade in patients with ILC patients, classifying 64% of tumors as GG low (GG1), 17% as GG high (GG3), and 19% as equivocal (not classified as GG1 or GG3; ref. 23). Studies evaluating the 21-gene OncotypeDX score in ILC tumors have also been reported wherein a majority of tumors (71%) were classified as intermediate-risk with very limited differences observed in breast cancer–specific survival between low/intermediate-risk (99%) and high-risk (96%) groups (25). In the prospective PlanB study, the prevalence of high recurrence score (RS) was 3-fold lower in patients who had lobular breast cancer compared with those who had nonlobular breast cancer, but 5-year disease-free survival estimates for lobular and nonlobular breast cancer were similar, suggesting that RS alone may not add the same prognostic information in ILC (29). The prognostic performance of Prosigna was compared in patients with ILC and patients with IDC, in a cohort of postmenopausal women receiving 5 years of endocrine therapy, with significant prognostic value for Prosigna demonstrated in N0 and N+ subsets; however, 28% of ILC tumors classified as intermediate-risk (26). Finally, Conforti and colleagues reported the prognostic performance of Clinical Treatment Score post-5 years (CTS5), which incorporates age, tumor size, nodal status, and tumor grade (22). In this study, 95% of patients were classified as low-risk, with only 3 out of 1,361 patients (0.2%) classified as high-risk; therefore the clinical utility of CTS5 in patients with ILC remains to be clearly established (22). Overall, it is notable that the classifiers described above, including BCI, were not developed specifically for prognostication of lobular cancer, which may provide a basis for the variability in performance.

Knowledge of both prognostic risk of recurrence as well as predicted response to extended endocrine therapy may be useful for treatment decisions. Aside from BCI prognostic results, the BCI assay also provides a predictive result that reports whether the patient has a high or low likelihood to benefit from extended endocrine therapy. It is notable that 34% of the BCI low/intermediate-risk patients were predicted to benefit from extended endocrine therapy and, conversely, that 49% of the BCI high-risk patients were not predicted to benefit from extended endocrine therapy (Supplementary Fig. S5). For patients with a high risk of recurrence but low likelihood of benefit from extended endocrine therapy, alternative therapies may warrant consideration. For example, the monarchE study demonstrated that addition of abemaciclib, a cyclin-dependent kinase 4/6 inhibitor, to adjuvant endocrine therapy in patients with HR+, HER2−, high-risk, early-stage breast cancer resulted in improved invasive disease-free survival (30). This study enrolled patients with poor prognosis, including those with one to three positive nodes and grade 3 tumors or a Ki-67 ≥20%. Although outcomes for patients with ILC were not disclosed in the monarchE trial, there is a growing importance for biologic markers to select patients who remain at high risk for recurrence despite optimal standard therapy and for whom addition of novel agents may be particularly beneficial. This study highlights the value of BCI in selecting this higher risk population among patients with ILC, a subtype of breast cancer whose biology remains understudied.

This study has several key strengths and limitations. This was a multi-institutional study, conducted in a well-annotated cohort in which 95% of patients had pure lobular histology. The study was prospectively defined and BCI testing was performed blinded to clinical outcome. Results from the comparative tissue dissection study showed a high concordance of BCI-risk group classification between methods, adding meaningful confidence in the study findings. Nevertheless, due to the unique growth pattern of lobular breast cancers, approximately 10% of samples were excluded since they had a tumor cellularity below 40%. Although useful information was gained from subset analyses overall, further examination of BCI prognostic activity in patients with N0 ILC is warranted given the low event rate in this study. Approximately 22% of patients had unknown HER2 status, although patients with HER2+ tumors represented ∼5% of the overall cohort. Finally, details regarding chemotherapy and endocrine therapy were unavailable for 11% and 7% of patients, respectively, and specific information on duration of endocrine therapy was not available for a majority (70%) of patients in the study.

In summary, the findings presented here indicate that BCI is a significant and independent prognostic factor for risk stratification in ILC. Findings from this study may help individualize prognosis and inform treatment approaches. Furthermore, although the addition of novel agents may result in improved outcome, it can also result in increased side effects and adverse events. BCI may be a useful tool in selecting a population of patients with ILC and a high risk of recurrence for whom escalation of therapy may be particularly beneficial; conversely, BCI may be useful in identifying patients with a low risk of recurrence who may be appropriate for de-escalation of treatment. Overall, these results support a role for BCI to define prognostic risk groups based on individual tumor biology to better inform treatment strategies for patients with ILC.

R. Nunes reports institutional grants from Biotheranostics, Inc. during the conduct of the study. T. Sella reports grants from American Physicians Fellowship for Medicine in Israel, and Pinchas Borenstein Talpiot Medical Leadership Program, Chaim Sheba Medical Center, Tel-Hashomer, Israel, during the conduct of the study; T. Sella also reports personal fees from Roche outside the submitted work. K. Treuner reports personal fees from Biotheranostics, Inc. outside the submitted work, as well as employment and stock ownership with Biotheranostics, Inc. J. Wong reports personal fees from Biotheranostics, Inc. outside the submitted work, as well as employment and stock ownership with Biotheranostics, Inc. Y. Zhang reports personal fees from Biotheranostics, Inc. outside the submitted work, as well as employment and stock ownership with Biotheranostics, Inc. Y. Zhang also has a patent for Predicting Likelihood of Response to Combination Therapy: 14/724,732 issued, a patent for Integration of Tumor Characteristics with Breast Cancer Index: 15/349,915 pending, a patent for Predicting Breast Cancer Recurrence: 14/483,108 pending, and a patent for Post-Treatment Breast Cancer Prognosis: 15/298,128 pending. A.L. Richardson reports a patent for Predictor of Therapy Response licensed and with royalties paid from Myriad Genetics. C.A. Schnabel reports personal fees from Biotheranostics, Inc. outside the submitted work, as well as employment and stock ownership with Biotheranostics, Inc. C.A. Schnabel also has a patent for Predicting Likelihood of Response to Combination Therapy: 14/724,732 issued, a patent for Integration of Tumor Characteristics with Breast Cancer Index: 15/349,915 pending, a patent for Predicting Breast Cancer Recurrence: 14/483,108 pending, and a patent for Post-Treatment Breast Cancer Prognosis: 15/298,128 pending. D.C. Sgroi is the coinventor of an issued patent for the use of BCI as a biomarker in breast cancer licensed to BioTheranostics, Inc. S. Oesterreich reports grants from BCRF during the conduct of the study. A. Cimino-Mathews reports grants and personal fees from Bristol-Myers Squibb outside the submitted work. O. Metzger reports grants from Pfizer, Roche, and Genentech, as well as personal fees from Grupo Oncoclinicas outside the submitted work. No disclosures were reported by the other authors.

R. Nunes: Conceptualization, validation, investigation, visualization, writing–original draft, project administration, writing–review and editing. T. Sella: Validation, investigation, visualization, writing–original draft, project administration, writing–review and editing. K. Treuner: Conceptualization, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. J.M. Atkinson: Conceptualization, resources, supervision, funding acquisition, investigation, visualization, methodology, project administration, writing–review and editing. J. Wong: Resources, data curation, software, formal analysis, validation, writing–original draft, writing–review and editing. Y. Zhang: Conceptualization, data curation, software, formal analysis, validation, visualization, methodology, writing–original draft, writing–review and editing. P. Exman: Conceptualization, resources, data curation, software, formal analysis, validation, visualization, methodology, writing–review and editing. D. Dabbs: Resources, writing–review and editing. A.L. Richardson: Resources, writing–review and editing. C.A. Schnabel: Conceptualization, resources, supervision, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. D.C. Sgroi: Conceptualization, supervision, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. S. Oesterreich: Visualization, writing–review and editing. A. Cimino-Mathews: Resources, visualization, writing–review and editing. O. Metzger: Conceptualization, resources, supervision, validation, investigation, visualization, methodology, project administration, writing–review and editing.

Funding was provided by Biotheranostics, Inc., and in part by the Breast Cancer Research Foundation (to A.L. Richardson, D.C. Sgroi, and S. Oesterreich). T. Sella is supported by The American Physicians Fellowship for Medicine in Israel and the Pinchas Borenstein Talpiot Medical Leadership Program, Chaim Sheba Medical Center, Tel-Hashomer, Israel. A.L. Richardson receives support from the Judy and Peter Blum Kovler Foundation. Special thanks to Ranelle Salunga, Tristan Harris, Jose Ramirez, Yen Tran, Veena Singh, MD, and Peter Gray, PhD, for their support of the trial and excellent technical expertise. This project used the UPMC Hillman Cancer Center and Tissue and Research Pathology/Pitt Biospecimen Core shared resource which is supported in part by award P30CA047904, as well as resources from the DF/HCC Breast SPORE: Specialized Program of Research Excellence (SPORE), an NCI-funded program, Grant 1P50CA168504. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health/NCI.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Sledge
GW
,
Chagpar
A
,
Perou
C
. 
Collective wisdom: lobular carcinoma of the breast
.
Am Soc Clin Oncol Educ Book
2016
;
35
;
18
21
.
2.
Barroso-Sousa
R
,
Metzger-Filho
O
. 
Differences between invasive lobular and invasive ductal carcinoma of the breast: results and therapeutic implications
.
Ther Adv Med Oncol
2016
;
8
:
261
6
.
3.
McCart Reed
AE
,
Kutasovic
JR
,
Lakhani
SR
,
Simpson
PT
,
McCart Reed
AE
,
Kutasovic
JR
, et al
Invasive lobular carcinoma of the breast: morphology, biomarkers and 'omics
.
Breast Cancer Res
2015
;
17
:
1
11
.
4.
Ciriello
G
,
Gatza
ML
,
Beck
AH
,
Wilkerson
MD
,
Rhie
SK
,
Pastore
A
, et al
Comprehensive molecular portraits of invasive lobular breast cancer
.
Cell
2015
;
163
:
506
19
.
5.
Rakha
EA
,
Ellis
IO
. 
Lobular breast carcinoma and its variants
.
Semin Diagn Pathol
2010
;
27
:
49
61
.
6.
Pestalozzi
BC
,
Zahrieh
D
,
Mallon
E
,
Gusterson
BA
,
Price
KN
,
Gelber
RD
, et al
Distinct clinical and prognostic features of infiltrating lobular carcinoma of the breast: combined results of 15 International Breast Cancer Study Group clinical trials
.
J Clin Oncol
2008
;
26
:
3006
14
.
7.
Engstrøm
MJ
,
Opdahl
S
,
Vatten
LJ
,
Haugen
OA
,
Bofin
AM
. 
Invasive lobular breast cancer: the prognostic impact of histopathological grade, E-cadherin and molecular subtypes
.
Histopathology
2015
;
66
:
409
19
.
8.
Adachi
Y
,
Ishiguro
J
,
Kotani
H
,
Hisada
T
,
Ichikawa
M
,
Gondo
N
, et al
Comparison of clinical outcomes between luminal invasive ductal carcinoma and luminal invasive lobular carcinoma
.
BMC Cancer
2016
;
16
:
248
.
9.
Iorfida
M
,
Maiorano
E
,
Orvieto
E
,
Maisonneuve
P
,
Bottiglieri
L
,
Rotmensz
N
, et al
Invasive lobular breast cancer: subtypes and outcome
.
Breast Cancer Res Treat
2012
;
133
:
713
23
.
10.
Gradishar
WJ
,
Anderson
BO
,
Abraham
J
,
Aft
R
. 
Breast Cancer, Version 3.2020, NCCN Clinical Practice Guidelines in Oncology
.
J Natl Compr Canc Netw
2020
;
18
:
452
78
.
11.
Mamtani
A
,
King
TA
. 
Lobular breast cancer: different disease, different algorithms?
Surg Oncol Clin N Am
2018
;
27
:
81
94
.
12.
Sgroi
DC
,
Carney
E
,
Zarrella
E
,
Steffel
L
,
Binns
SN
,
Finkelstein
DM
, et al
Prediction of late disease recurrence and extended adjuvant letrozole benefit by the HOXB13/IL17BR biomarker
.
J Natl Cancer Inst
2013
;
105
:
1036
42
.
13.
Zhang
Y
,
Schnabel
CA
,
Schroeder
BE
,
Jerevall
P-LL
,
Jankowitz
RC
,
Fornander
T
, et al
Breast cancer index identifies early-stage estrogen receptor-positive breast cancer patients at risk for early- and late-distant recurrence
.
Clin Cancer Res
2013
;
19
:
4196
205
.
14.
Bartlett
JMS
,
Sgroi
DC
,
Treuner
K
,
Zhang
Y
,
Ahmed
I
,
Piper
T
, et al
Breast Cancer Index and prediction of benefit from extended endocrine therapy in breast cancer patients treated in the Adjuvant Tamoxifen—To Offer More? (aTTom) trial
.
Ann Oncol
2019
;
30
:
1776
83
.
15.
Sgroi
DC
,
Sestak
I
,
Cuzick
J
,
Zhang
Y
,
Schnabel
CA
,
Schroeder
B
, et al
Prediction of late distant recurrence in patients with oestrogen-receptor-positive breast cancer: a prospective comparison of the breast-cancer index (BCI) assay, 21-gene recurrence score, and IHC4 in the TransATAC study population
.
Lancet Oncol
2013
;
14
:
1067
76
.
16.
Zhang
Y
,
Schroeder
BE
,
Jerevall
PL
,
Ly
A
,
Nolan
H
,
Schnabel
CA
, et al
A novel breast cancer index for prediction of distant recurrence in HR+ early-stage breast cancer with one to three positive nodes
.
Clin Cancer Res
2017
;
23
:
7217
24
.
17.
Sestak
I
,
Buus
R
,
Cuzick
J
,
Dubsky
P
,
Kronenwett
R
,
Denkert
C
, et al
Comparison of the performance of 6 prognostic signatures for estrogen receptor–positive breast cancer a secondary analysis of a randomized clinical trial
.
JAMA Oncol
2018
;
4
:
545
53
.
18.
Allison
KH
,
Hammond
MEH
,
Dowsett
M
,
McKernin
SE
,
Carey
LA
,
Fitzgibbons
PL
, et al
Estrogen and progesterone receptor testing in breast cancer: American Society of Clinical Oncology/College of American Pathologists guideline update
.
Arch Pathol Lab Med
2020
;
144
:
545
63
.
19.
Edge
SB
,
Byrd
DR
,
Compton
CC
,
Fritz
AG
,
Greene
FL
,
Trotti
A
,
editors
.
AJCC Cancer Staging Manual
. 7th ed.; 
2010
.
20.
Ma
X-J
,
Hilsenbeck
SG
,
Wang
W
,
Ding
L
,
Sgroi
DC
,
Bender
RA
, et al
The HOXB13/IL17BR expression index is a prognostic factor in early-stage breast cancer
.
J Clin Oncol
2006
;
24
:
4611
9
.
21.
Arpino
G
,
Bardou
VJ
,
Clark
GM
,
Elledge
RM
. 
Infiltrating lobular carcinoma of the breast: tumor characteristics and clinical outcome
.
Breast Cancer Res
2004
;
6
:
7
11
.
22.
Conforti
F
,
Pala
L
,
Pagan
E
,
Viale
G
,
Bagnardi
V
,
Peruzzotti
G
, et al
Endocrine-responsive lobular carcinoma of the breast: features associated with risk of late distant recurrence
.
Breast Cancer Res
2019
;
21
:
1
11
.
23.
Metzger-Filho
O
,
Michiels
S
,
Bertucci
F
,
Catteau
A
,
Salgado
R
,
Galant
C
, et al
Genomic grade adds prognostic value in invasive lobular carcinoma
.
Ann Oncol
2013
;
24
:
377
84
.
24.
Beumer
IJ
,
Persoon
M
,
Witteveen
A
,
Dreezen
C
,
Chin
S-F
,
Sammut
S-J
, et al
Prognostic value of MammaPrint® in invasive lobular breast cancer
.
Biomark Insights
2016
;
11
:
139
46
.
25.
Kizy
S
,
Huang
JL
,
Marmor
S
,
Tuttle
TM
,
Hui
JYC
. 
Impact of the 21-gene recurrence score on outcome in patients with invasive lobular carcinoma of the breast
.
Breast Cancer Res Treat
2017
;
165
:
757
63
.
26.
Lænkholm
AV
,
Jensen
MB
,
Eriksen
JO
,
Rasmussen
BB
,
Knoop
AS
,
Buckingham
W
, et al
PAM50 risk of recurrence score predicts 10-year distant recurrence in a comprehensive danish cohort of postmenopausal women allocated to 5 years of endocrine therapy for hormone receptor-positive early breast cancer
.
J Clin Oncol
2018
;
36
:
735
40
.
27.
Sestak
I
,
Filipits
M
,
Buus
R
,
Rudas
M
,
Balic
M
,
Knauer
M
, et al
Prognostic value of EndoPredict in women with hormone receptor positive, HER2-negative invasive lobular breast cancer
.
Clin Cancer Res
2020
;
26
:
4682
7
.
28.
McCart Reed
AE
,
Lal
S
,
Kutasovic
JR
,
Wockner
L
,
Robertson
A
,
de Luca
XM
, et al
LobSig is a multigene predictor of outcome in invasive lobular carcinoma
.
NPJ Breast Cancer
2019
;
5
:
1
11
.
29.
Christgen
M
,
Gluz
O
,
Harbeck
N
,
Kates
RE
,
Raap
M
,
Christgen
H
, et al
Differential impact of prognostic parameters in hormone receptor–positive lobular breast cancer
.
Cancer
2020
;
126
:
4847
58
.
30.
Johnston
SRD
,
Harbeck
N
,
Hegg
R
,
Toi
M
,
Martin
M
,
Shao
ZM
, et al
Abemaciclib combined with endocrine therapy for the adjuvant treatment of HR+, HER2-, node-positive, high-risk, early breast cancer (monarchE)
.
J Clin Oncol
2020
;
38
:
3987
98
.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs International 4.0 License.