Clinical evaluation of molecular surrogate subtypes in patients with ipsilateral multifocal primary breast cancer

In this retrospective study, women diagnosed and treated for primary multifocal breast cancer at Sahlgrenska University Hospital (Gothenburg, Sweden) between 2012 and 2017 were screened for inclusion. Inclusion criteria were specimens with ≥ 2 invasive foci and ≥ 2 tumors with information for the ER, PR, HER2 and Ki67 breast cancer biomarkers assessed with IHC. Exclusion criteria were prior breast cancer-related surgery in the same breast, patients who received neoadjuvant treatment, and distant metastases found at the time of diagnosis. Patient data were retrieved from the Swedish National Breast Cancer Register and local hospital records (i.e., Melior for patient medical records and Sympathy for local pathology reports). Ipsilateral multifocal primary breast cancer was defined as more than one invasive tumor reported in the pathology report regardless of the distance between the foci or multiple invasive tumors found within the same quadrant. The tissue between two invasive foci could consist of normal breast tissue or ductal carcinoma in situ (DCIS). The study was approved by the Regional Ethical Review Board in Linköping, Sweden (registration number 2016/387-31) and the Regional Ethical Review Board in Gothenburg, Sweden (registration number 479--18).

At the time of clinical evaluation, IHC was performed using 4-µm formalin-fixed, paraffin-embedded sections with the following antibodies: rabbit anti-ERα (Agilent Dako IR084, clone EP1), mouse anti-PR (Agilent Dako IR068, clone 636), mouse anti-KI-67 (Agilent Dako IR626, clone MIB-1), and rabbit anti-human HercepTest (Agilent Dako SK001). In Sweden, (2012–2017), ER and PR were considered positive with ≥ 10% immunostaining in neoplastic cells. Ki67 index was considered to be high with ≥ 14% immunostaining in neoplastic cells between 2012–2013, ≥ 30% immunostaining in neoplastic cells between 2013 and 2016, and ≥ 20% immunostaining in neoplastic cells in 2017. Samples with HercepTest scores of 2 + and 3 + were confirmed for amplification using silver in situ hybridization.

The receptor-based molecular surrogate subtypes (luminal A, luminal B HER2-negative, luminal B HER2-positive, non-luminal HER2-positive, triple-negative breast cancer [TNBC]) were determined for each focus using the applicable Swedish National guidelines for the biomarkers as mentioned above, i.e., luminal A: ER + , PR ± , HER2-, Ki67 low; luminal B HER2-negative: ER + , PR ± , HER2-, Ki67 high; luminal B HER2-positive: ER + , PR ± , HER2 + , any Ki67; non-luminal HER2-positive: ER-, PR-, HER2 + ; TNBC: ER–, PR–, HER2– [26]. Data on co-morbidities and postoperative adjuvant treatment were collected from patient medical records and applicable national treatment guidelines were considered when an additional multidisciplinary team meeting was held to reevaluate the treatment received by patients with discordant subtypes (IHC assessed samples) between the largest primary tumor (PT1) and the second tumor (PT2).

Statistical comparisons were performed only using two foci per specimen. Descriptive statistics as frequencies and percentages were presented for categorical variables and median with quartiles for continuous variables. Concordance between PT1 and PT2 with respect to categorical characteristics was explored by two-way crosstabs. Comparisons between PT1 and PT2 with respect to numerical variables were tested by nonparametric test for related measurements (Wilcoxon’s test), while comparisons between PT1 and PT2 with respect to dichotomous variable were tested by McNemar’s test. Independent groups were compared by Chi-square test or Mann–Whitney test. IBM SPSS v 28 was used for descriptive and analytic statistics.2. Kaplan–Meier plots were constructed for overall survival (OS) and disease-free survival (DFS) using the ggsurvfit package (v0.1.0) in R/Bioconductor (v4.1.1) [27]. OS was defined as the time from primary surgery to death of any cause, and DFS was defined as the time from primary surgery to local recurrence or distant metastasis or death. End of follow-up was 5th of October 2022. The tableone R script (version 0.13.2) was used to identify clinicopathologic features between different groups [28].

Between 2012 and 2017, 3362 primary invasive breast cancer specimens obtained by surgical resection were registered at Sahlgrenska University Hospital (Gothenburg, Sweden). Of the 3362 specimens, 347 (10.3%) contained ≥ 2 tumor foci. As only one focus was assessed with IHC, 125 of the 347 breast specimens (34.6%) were excluded. In total, 183 breast specimens from 180 patients (three patients had surgery done for bilateral IMBC) had ≥ 2 foci assessed with IHC and were therefore included in the study (Fig. 1).

Patient characteristics for the 180 patients and 183 specimens are shown in Table 1. Median age was 60 years (IQR 52–71). The most common surgery in the breast was mastectomy (71.6%), and in the axilla sentinel lymph node biopsy was the most common surgical procedure (70.0%). The majority of the patients received adjuvant endocrine therapy (93.3%) and 64 (35.5%) received chemotherapy. Most of the tumors were bifocal (77.0%).

The characteristics of PT1 and PT2 are shown in Table 2. Median size was 22.0 mm (IQR 14–30) for PT1 and 10.0 mm (IQR 7–15) for PT2. PT2 was the second largest focus in all specimens where two foci were assessed with IHC. For specimens where three or four foci were assessed with IHC, PT2 was the second largest tumor in all but two cases. In these two cases, PT2 was the focus with molecular surrogate subtype discordancy. The size of these two PT2s was only one and two mm smaller than the second largest. Both PT1 and PT2 were predominantly ductal (125 [68.3%] vs. 127 [69.4%]), Nottingham histologic grade 2 (111 [60.7%] vs. 106 [57.9%]), and luminal A (117 [63.9%] vs. 129 [70.5%]) for PT1 and PT2, respectively.

Discordance in ER, PR, HER2 and Ki67 was found in 2.7% (n = 5), 19.1% (n = 35), 7.7% (n = 14) and 16.9% (n = 31) of the 183 invasive foci, respectively. Ki67 was the only biomarker that showed significant changes in expression (p = 0.03) where 44.8% (24 of 54) changed from high expression in PT1 to low expression in PT2 (Table 3). The biomarker status was then used to determine the molecular surrogate subtypes, which were found to be concordant for 135/183 (73.8%) specimens and discordant for 48/183 (26.2%) specimens. The most common concordant molecular surrogate subtype between PT1 and PT2 was PT1 luminal A/ PT2 luminal A (102 of 135 [75.6.8%]), while the most common discordant molecular surrogate subtype was in the PT1 luminal B HER2-/PT2 luminal A group (21 of 48 [43.8.%]); Table 4).

The 48 discordant PT1 and PT2 specimens (48 patients) based on molecular surrogate subtyping were reevaluated for the previously recommended therapeutic options using a new multidisciplinary team board consisting of a medical oncologist (P.K), surgical oncologist (S.J) and a board-certified breast pathologist (A.K). Applicable guidelines at the time of surgery were considered. Eighteen patients (37.5%) had a more aggressive subtype in PT2 than PT1 of which PT1 luminal A/PT2 luminal B HER2- was the most common discordance (9 of 48 [18.6%]), and 30 (62.5%) had a less aggressive subtype in PT2 compared to PT1 (Table 5). Twenty patients (41.7% of the discordant 48 patients and 11.1% of the study cohort) had subtype changes in PT2, which potentially could have led to different therapy recommendations. Of the 20 patients with potential therapy changes, 11 (22.9% of the 48 discordant patients and 6.1% of the study cohort) were recommended therapy changes due to the subtype of PT2, i.e., two patients received additional endocrine therapy due to changes in ER status, three received additional chemotherapy due to changes in Ki67 status or tumor grade, and six received additional combined chemotherapy and trastuzumab due to HER2 positivity (Fig. 2). In addition, one patient would have received endocrine therapy, but due to the small tumor size of PT2 (5 mm) no treatment was recommended according to the Swedish national guidelines applicable for the current year. Another patient had a luminal B HER2 + PT2 but was not recommended trastuzumab due to the small tumor size (3 mm). In total, 37 patients (77.1% of the discordant patients and 20.1% of the study cohort) received no additional treatment due to either age, comorbidities, patient choice, or the patient received all possible treatment due to the status of PT1. Of the 11 patients where recommended therapy changed due to PT2, four patients had concordant histology and grade within the foci, and seven were discordant in histology and grade within the foci (Fig. 3).

Survival analysis showed that the median follow-up time for the study cohort was 79.7 months (IQR 1.7–128.4). In addition, univariate analysis showed that the molecular surrogate subtype-concordant group had significantly worse DFS than the molecular surrogate subtype-discordant group (p = 0.032), but no significant difference in OS was found between these groups (p = 0.18) (Fig. 4A, B). Recurrence was observed in 19 patients, with 18 in the molecular surrogate subtype-concordant group (13.6%) and 1 (2.1%) in the molecular surrogate–discordant group (p = 0.027). Deaths were observed in 28 patients, of which 24 (18.2%) were in the molecular surrogate subtype-concordant group and 4 (8.5%) in the molecular surrogate discordant–group (p = 0.147).

In the study cohort, 103 specimens (103 patients) had the same histologic type and grade in the different foci. Yet, 15 (14.6%) of these patients had discordant molecular surrogate subtyping for PT1 and PT2. Notably, 4 of these 15 patients were recommended therapeutic changes due to PT2 (Fig. 3). We also conducted a comparison between the 103 patients with concordant morphology and grade, where ≥ 2 foci were assessed with IHC, to 102 excluded patients with similarly concordant features, but where only one focus was assessed with IHC (Additional file 1: Table 1). Compared to the group of 103 patients, the 102 excluded patients more frequently underwent axillary lymph node clearances (47.1% compared to 27.2%) and had a larger lymph node burden: 49.0% were pN0 compared to 55.3%. In addition, the 102 excluded patients more frequently had Grade 3 tumors (30.4% vs. 17.5%, p = 0.031), higher Ki67 levels (44.1% vs. 25.2%, p = 0.004) and received more chemotherapy (51.0% vs. 33.0, p = 0.009) than the group of 103 patients. Recurrence was observed in 10 (9.8%) patients in the excluded group of 102 patients and in 13 (12.6%) patients in the 103 patient group (p = 0.526). Deaths were recorded in 23 (22.5%) patients in the excluded group of 102 patients and in 21 (20.4%) patients in the 103 patient group (p = 0.715). Although the excluded group of 102 patients had more aggressive tumors, higher nodal burden and received more chemotherapy, univariate survival analysis showed no statistically significant difference in neither DFS (p = 0.56) nor OS (p = 0.62) when comparing these two groups with shared features considering histology and grade between the different foci (Additional File 2: Fig. 1A, B).