HER2-low status and response to neoadjuvant chemotherapy in HER2 negative early breast cancer

Knowledge on whether low expressions of HER2 have prognostic impact in early-stage breast cancer (BC) and on its response to current chemotherapy protocols can contribute to medical practice and development of new drugs for this subset of patients, changing treatment paradigms. This study aims to evaluate the impact of HER2-low status on response to neoadjuvant chemotherapy (NACT) and survival outcomes in early-stage HER2-negative BC. Records from all BC patients treated with NACT from January 2007 to December 2018 in a single cancer center were retrospectively reviewed. HER2-negative (immunohistochemistry [IHC] 0, + 1, or + 2 non-amplified by in situ hybridization [ISH]) patients were included. HER2-low was defined by IHC + 1 or + 2 ISH non-amplified and HER2-0 by IHC 0. The coprimary objectives were to compare pathological complete response (pCR) and relapse-free survival (RFS) between luminal/HER2-low versus luminal/HER2-0 populations and between triple negative (TNBC)/HER2-low versus TNBC/HER2-0. In total, 855 HER2-negative patients were identified. The median follow-up was 59 months. 542 patients had luminal subtype (63.4%) and 313 had TNBC (36.6%). 285 (33.3%) were HER2-low. Among luminal patients, 145 had HER2 IHC + 1 (26.8%) and 91 had IHC + 2/ISH non-amplified (16.8%). In TNBC, 36 had HER2 IHC + 1 (11.5%) and 13 had IHC + 2/ISH non-amplified (4.2%). Most patients had locally advanced tumors, regardless of subtype or HER2-low status. For luminal disease, pCR was achieved in 13% of HER2-low tumors versus 9.5% of HER2-0 (p = 0.27). Similarly, there was no difference in pCR rates among TNBC: 51% versus 47% in HER2-low versus HER2-0, respectively (p = 0.64). HER2-low was also not prognostic for RFS, with 5-year RFS rates of 72.1% versus 71.7% (p = 0.47) for luminal HER2-low/HER2-0, respectively, and 75.6% versus 70.8% (p = 0.23) for TNBC HER2-low/HER2-0. Our data does not support HER2-low as a biologically distinct BC subtype, with no prognostic value on survival outcomes and no predictive effect for pCR after conventional NACT.


Introduction
HER2-positive status has for more than 20 years defined a subgroup of breast cancer (BC) patients who benefit from anti-HER2 therapy [1,2]. Beyond that, HER2 status defines a distinct BC subtype with aggressive biological behavior and historically worse prognosis, a reality that was changed after the incorporation of HER2 therapy [3][4][5]. Patients with HER2-low status-defined as immunohistochemistry 1 3 (IHC) + 1 or + 2/ISH non-amplified, have not, however, benefitted from the use of either trastuzumab or pertuzumab [6][7][8].
Recently, the results of non-randomized trials with novel antibody-drug conjugates (ADC) targeting HER2 (trastuzumab-deruxtecan and trastuzumab-duocarmazine) have suggested a level of efficacy in HER2-low patients with advanced breast cancer, with objective response rates ranging between 32 and 37% in a heavily pretreated population [9,10]. This led to the hypothesis that HER2-low tumors might represent a separate disease subset, distinct from other luminal and triple-negative breast cancers (TNBC), with distinct clinical properties and thus the potential for new targeted therapies. Indeed, several trials are currently exploring the potential of anti-HER2 agents in HER2-low patients.
Little is known about HER2-low early BC and data on its prognosis are scarce [11]. Besides, its response to standard chemotherapy has not been previously reported. More data on this patient population are urgently needed in order to guide the integration of these new drugs into treatment guidelines.
The majority of large phase 3 trials conducted in the past involving luminal or triple-negative BC did not collect detailed HER2 status information, making retrospective cohorts particularly valuable in this scenario. Therefore, the purpose of this study was to evaluate the impact of HER2-low status on response to neoadjuvant chemotherapy (NACT) and survival outcomes in early-stage HER2-negative BC.

Study design and participants
This is a retrospective cohort which enrolled all early breast cancer patients treated with neoadjuvant chemotherapy from January 2007 to December 2018 in a single cancer center in South America (A. C. Camargo Cancer Center, São Paulo, Brazil). Patients had to meet the following inclusion criteria: age ≥ 18 years old, histologically proven invasive carcinoma of the breast, clinical stage I-III, and treatment with neoadjuvant chemotherapy followed by surgery with curative intent. Patients with HER2-positive BC-IHC + 3 or IHC + 2 FISH/ ISH amplified or patients with IHC + 2 FISH/ISH equivocal, were excluded. Other exclusion criteria included insufficient data about tumor pathological characteristics or treatment received, history of synchronous or metachronous breast cancer, and diagnosis of other synchronous invasive cancer.
Subtype was defined by IHC for hormonal receptors (HR) and HER2 and/or FISH when HER2 staining was + 2, on pre-treatment biopsy specimens. Tumors were defined as luminal if estrogen (ER) and/or progesterone (PgR) receptors were ≥ 1% or TNBC, if ER < 1% and PgR < 1%. HER2-low status was defined as HER2 IHC + 1 or + 2/ISH non-amplified. Other patients were classified as HER2-0. HER2 was accessed using standard antibodies and ISH techniques and classified according to the ASCO/CAP guidelines available during the different time periods [12][13][14]. Pathological review of specimens was not performed.
The patient flow diagram for the study is shown on Supplementary figure S1. A total of 1,495 consecutive BC patients were screened, and 303 HER2-positive and 2 HER2 + 2/ISH equivocal were excluded. Other 337 patients were excluded for other reasons. The following baseline clinicopathological characteristics were collected from medical records: age at diagnosis, gender, menopausal status, histologic subtype and grade, Ki67, TNM staging, and anatomic stage groupings. We also collected data regarding treatments offered (chemotherapy, surgery, and radiotherapy) and sites of first disease recurrence. The choice of chemotherapy agents used was at the discretion of the attending physician and usually followed national and international guidelines for BC treatment and the best evidence from clinical trials. This included the use of anthracyclines and/or taxanes for most patients, carboplatin preferentially for TNBC [15], and dose-dense schemas (corresponding to biweekly doxorubicin and cyclophosphamide administration) initially to higher-risk tumors [16], but after 2018 to most patients following data on overall survival gain from this practice [17]. This study was approved by the institutional Internal Ethics Review Board.

Statistical analysis
The primary objective of this study was to investigate whether HER2-low status impacts efficacy outcomes, including pathological complete response (pCR) and relapse-free survival (RFS) for patients with luminal/TNBC undergoing neoadjuvant chemotherapy. pCR was defined as no invasive carcinoma in the breast and in axillary lymph nodes at the time of surgery (ypT0/is ypN0). RFS was defined as the time from breast cancer diagnosis to locoregional or contralateral relapse, distant metastasis, or death (from any cause), whichever occurred first.
Secondary objectives were included to describe the demographics and clinical-pathological characteristics of this population and to assess the impact of low HER2 expression on overall survival (OS) and patterns of recurrence.
The clinical-pathological and demographical variables were represented as relative and absolute frequencies. Fisher's exact test and Mann-Whitney's test were used to explore differences in categorical and continuous variables between luminal HER2-0/HER2-low and TNBC HER2-0/HER2-low patients.
Survival curves were generated using the Kaplan-Meier method and were compared using the log-rank test. Tests were considered statistically significant when p-values (two-tailed) < 0.05. Baseline clinicopathological and demographical variables associated with OS or RFS with Wald's p-value ≤ 0.2 in the univariate analysis were selected for multivariate analysis. All statistical analyses were performed with software SPSS 23.0 (SPSS, Chicago, IL).
The baseline characteristics of this population and the patterns of treatment offered, stratified by subtype and the presence or absence of HER2-low status, are described in Table 1. Most patients were premenopausal and had invasive ductal carcinomas and grade II/III tumors, irrespective of subtype or HER2-low status. Among the luminal subtype tumors, there was a slightly high proportion of clinical T3/4, node positive, and stage III tumors, with no statistical difference between the luminal/HER2-low and the luminal/ HER2-0 population. Among TNBC tumors, the same was true despite a nonstatistically significant larger number of more advanced tumors among TNBC/HER-low as compared to TNBC/HER2-0. Median Ki67 was higher for TNBC as compared to luminal BC, irrespective of HER2 status.
Anthracyclines plus taxanes were the preferred chemotherapy regimen, with a higher proportion of dose-dense regimens and neoadjuvant carboplatin being used in TNBC/ HER2-low and TNBC/HER2-0 (Table 1). Mastectomy plus axillary dissection was performed in more than 70% of luminal/HER2-low and luminal/HER2-0 patients. TNBC/ HER2-0 patients had slightly less mastectomies when compared to the rest of the population, probably reflecting the higher proportion of clinical T2 tumors and clinically negative axillae. Most patients in this cohort received adjuvant radiotherapy.
HER2-low status was not associated with RFS both in univariate and multivariate analysis for the entire cohort (HR 0.83, 95%CI 0.6-1.11, p = 0.21), as shown in Table 2. In univariate analysis for OS, HER2-low status was associated with improved OS, but this association was lost after adjusting for other factors in the multivariate model for all patients (Table 3). Survival curves considering the whole population stratified by HER2-0, HER2 + 1, and HER2 + 2 or HER2-0/ HER2-low follow the same behavior and are shown in supplementary figure S2. An exploratory analysis considering the luminal and TNBC population separately was conducted and also did not review the HER2-low status as a prognostic factor (Supplementary Tables S2, 3, 4, and 5).
Sites of recurrence-bone, visceral, and central nervous system (CNS)-were not significantly different among HER2-low patients, although CNS and visceral recurrences occurred in a greater proportion among the HER2-0 population (Supplementary figure S3).

Discussion
In this cohort, 33.3% of patients were classified as HER2low, a somewhat lower frequency than observed in previous reports where the prevalence of this emerging subtype ranged between 40 and 50% [18,19]. This discrepancy seems to be largely driven by the lower proportion of HER2 + 1 in this cohort (21.2%), when compared with other  [20]. One possible explanation would be that the distinction between HER2-0 and HER2 + 1 until recently had no clinical implications, which may impact the precision with which pathologists evaluate low scores. Indeed, previously published data showed that for cases considered as HER2-0 by local pathologists, there was only 15% concordance after central review, with as many as 76% being reclassified to HER2 + 1 [21]. Concordant with more recent data, the vast majority of HER2-low patients (82.8%) had hormone receptor (HR)positive breast cancer [20]. However, differently from other publications [20,22], HER2-low status did not imply more advanced tumors among luminal BC, with only slightly more T4 tumors among TNBC. When looking at the entire cohort there were more node-positive cases among HER2low, but this was possibly biased given that the majority of this population comprised luminal patients for which neoadjuvant chemotherapy is commonly indicated in locally advance disease.
Regarding the HER2-low status impact on survival outcomes, it initially looked like HER2-low which led to better overall survival, but this did not hold true when adjusting for disease subtype and other prognostic factors on multivariate analysis. The prognostic role of HER2-low status remains controversial, with some retrospective studies suggesting worse prognosis for HER2-low in localized node positive and luminal early BC, although these relatively older works made comparisons only between HER2-0/HER2 + 1 versus HER2 + 2/ISH negative [19,22].
A recent paper, however, that used the current HER2-low definition, did not find a prognostic impact for HER2-low status on overall survival in a large cohort of 3,689 patients with predominantly advanced BC [20]. Interestingly, in this paper, comprehensive molecular analysis was performed using PAM50 and individual gene expression data, and the authors were able to differentiate HR-positive/HER2-low BC as a more distinct biologic entity pertaining to Luminal A intrinsic subtype with a higher ERBB2 expression level versus HR-positive/HER2-0 BC, a group that encompassed a larger proportion of Luminal B and Basal-like subtypes with lower ERBB2 expression. This was not observed for TNBC that was composed predominantly of Basal-like subtype and had low levels of ERBB2 expression regardless of being HER2-low or not. The underlying biological intrinsic subtypes might explain why in our cohort, although not significant, more CNS and visceral recurrences were seen in the HER2-0 population, which was possibly enriched for the Basal-like subtype.
HER2-low status did not predict for response to standard chemotherapy in this cohort. A similar finding was preliminarily presented at San Antonio 2020, in a retrospective series of 331 patients, with no difference in the pCR rates between HR-positive HER2-0 and HER2-low subgroups (8% vs 13%, p = 0.35), but a higher pCR rate, although non-statistically significant, among TNBC HER2-0 versus HER2-low tumors (56% vs. 39%, p = 0.09) [23]. Diverging from this data, results from a pooled analysis of 2310 patients from four prospective neoadjuvant trials showed that HR-positive HER2-low tumors had lower pCR rates when comparing to HER2-0 (17.5% versus 23.6%, respectively, p = 0.024), but no difference was seen in HR-negative HER2-low versus HER2-0 patients [24]. However, this predictive power of HER2-low status on pCR in the HR-positive subgroup was not sustained after performing a multivariable logistic regression model, corroborating our findings.
Although the activity of new ADCs is being extensively studied in advanced HER2-low BC, clinical trials testing  theses agents in the early-stage setting are lacking [25]. In this scenario, the only data presented were the adjuvant trial of the HER2-targeting vaccine, Nelipepimut-S, associated with Trastuzumab with negative results [26,27]. Our study has several limitations, including relatively small sample size, its retrospective nature, absence of central pathological review, and patient inclusion across a large time period, during which different guidelines for HER2 testing and interpretation were in use. Still it sheds some light on the clinical and biological behavior of this large BC population which usually accounts for young individuals with locally advanced HR positive, who have very poor pCR rates following NACT and therefore could largely benefit from the incorporation of new treatment approaches.
For the abovementioned subset new classes of drugs, like CDK4/6 inhibitors, have not improved outcome [28] and more importantly than for the TNBC subgroup, for which the field is rapidly moving toward neoadjuvant chemoimmunotherapy with outstanding pCR rates [29], it represents an area of unmet clinical need. Addressing this issue In conclusion, although our data does not support HER2low as a biologically distinct BC subtype, it may reflect a specially interesting population in which to test new HER2targeting ADC. Therefore, efforts should be made to better identify those patients in real-life settings and to better determine how to incorporate these drugs into (neo)adjuvant treatment plan.