Landscape of HER2-low metastatic breast cancer (MBC): results from the Austrian AGMT_MBC-Registry

Background About 50% of all primary breast cancers show a low-level expression of HER2 (HER2-low), defined as immunohistochemically 1+ or 2+ and lack of HER2 gene amplification measured by in situ hybridization. This low HER2 expression is a promising new target for antibody–drug conjugates (ADCs) currently under investigation. Until now, little is known about the frequency and the prognostic value of low HER2-expression in metastatic breast cancer (MBC). Patients and methods The MBC-Registry of the Austrian Study Group of Medical Tumor Therapy (AGMT) is a multicenter nationwide ongoing registry for MBC patients in Austria. Unadjusted, univariate survival probabilities of progression-free survival (PFS) and overall survival (OS) were calculated by the Kaplan–Meier method and compared by the log-rank test. Multivariable adjusted hazard ratios were estimated by Cox regression models. In this analysis, only patients with known HER2 status and available survival data were included. Results As of 11/15/2020, 1,973 patients were included in the AGMT-MBC-Registry. Out of 1,729 evaluable patients, 351 (20.3%) were HER2-positive, 608 (35.2%) were HER2-low and 770 (44.5%) were completely HER2-negative (HER2-0). Low HER2-expression was markedly more frequent in the hormone-receptor(HR)+ subgroup compared to the triple-negative subgroup (40% vs. 23%). In multivariable analysis, low HER2 expression did not significantly influence OS neither in the HR+ (HR 0.89; 95% CI 0.74–1.05; P = 0.171) nor in the triple-negative subgroup (HR 0.92; 95% CI 0.68–1.25; P = 0.585), when compared to completely HER2-negative disease. Similar results were observed when HER2 IHC 2+ patients were compared to IHC 1+ or 0 patients. Conclusion Low-HER2 expression did not have any impact on prognosis of metastatic breast cancer in this real-world population. Supplementary Information The online version contains supplementary material available at 10.1186/s13058-021-01492-x.


Introduction
Amplification of human epidermal growth factor receptor 2 (HER2) is a well-established negative prognostic factor both in early and metastatic breast cancer (MBC). HER2-directed therapies, however, have changed the natural course of this disease. Nowadays, adequately treated HER2+ /hormone-receptor(HR)+ breast cancer belongs to the subtypes with the most favorable prognosis both in the early and the advanced stage [1,2].
In contrast to HER2 positivity, defined as immunohistochemically (IHC) 3+ or IHC 2+ and HER2 gene amplification measured by in situ hybridization (ISH) [3], the significance of a low-level expression of HER2 (HER2low) is less clear. HER2-low is defined as IHC 1+ or IHC 2+ without HER2 gene amplification and compromises about 50 to 55% of all primary breast cancers [3,4]. In general, these tumors do not respond to trastuzumab [5] or T-DM1 [6], even if there seems to be a subgroup of patients-selected by a novel poly-ligand profiling technique-who might benefit from trastuzumab [7]. HER2low, however, is a potential target of new antibody-drug conjugates (ADCs). In contrast to T-DM1, these new ADCs show a higher bystander killer effect, by using cleavable linkers and a higher drug-to-antibody ratio [8,9] and are therefore not only active in HER2-overexpressing tumors [10] but also in tumors with low HER2 expression. Two of these ADCs have already shown promising activity in phase I trials including HER2-low MBC [11,12]: trastuzumab deruxtecan and trastuzumab duocarmazine. The former ADC is already approved by the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) for the treatment of HER2+ MBC pretreated with two or more anti-HER2based regimens. For the treatment of HER2-low MBC, however, none of the ADCs is approved today. Currently, two phase III trials are investigating trastuzumab deruxtecan in patients with HER2-low MBC (DESTINY-Breast04; ClinicalTrials.gov identifier: NCT03734029 and DB-06; NCT04494425). Furthermore, another novel HER2-targeting ADC (RC-48) is tested in HER2-low MBC in a phase I/II trial (NCT03052634). Besides several ADCs, bispecific antibodies as well as HER2 vaccines are under investigation in HER2-low breast cancer as reviewed by Tarantino et al. [13].
Several retrospective studies investigated the prognostic value of low HER2 expression in early breast cancer (EBC) [14][15][16][17][18][19][20]. Most of these studies showed a negative impact of a IHC 2+ HER2-expression on the risk of recurrence or survival but no or little influence on outcome by HER2 IHC 1+. In contrast to EBC, little is known about the real frequency and the prognostic significance of this new breast cancer subtypes in MBC. Only a few retrospective studies have currently been published, which investigated the prognostic value of low HER2 expression in MBC [20][21][22]. One trial showed a negative prognostic value of HER2 IHC 2+ only in patients older than 55 years, while there was no OS difference between IHC 2+ and 0 or 1+ tumors in the overall population [20]. The other two publications did not find any difference in OS between patients with HER2low tumors compared to patients with completely HER2negative tumors irrespective of the hormone receptor status [21,22], however the patient numbers in these studies were rather low.
Here, we present data from a large nation-wide registry for MBC in Austria. Both the incidence and the prognostic value of low HER2 expression were investigated in dependence of the HR status.

Methods
The MBC-registry of the Austrian Study Group of Medical Tumor Therapy (AGMT) is a multicenter nationwide ongoing retrospective and prospective registry for MBC patients in Austria.
HER2+ was defined as IHC 3+ or IHC 2+ and ISH+ according to the American Society of Clinical Oncology/College of American Pathologists Clinical Practice (ASCO/CAP) guidelines [3]. HER2-low was defined as IHC 1+ or IHC 2+ and ISH-. Completely, HER2-negative (HER2-0) was defined as IHC 0 and ISH-(if available). The classification was based on local pathology reports. No central pathology review was performed.
In this analysis, only patients with known HER2 status and available survival data were included. For progression-free survival (PFS) analyses, only patients with at least one line of therapy for metastatic disease and sufficiently documented medical records allowing calculation of PFS were included.
In patients with more than one available tumor sample, the following hierarchy was applied: if a tumor sample from a metastatic site was accessible, which was taken within 3 months after diagnosis of metastatic disease and included at least ER-and HER2-status, the receptor status (as well as grade, Ki-67 and histologic subtype) of this biopsy was used. Otherwise, the receptor status of the latest primary tumor (or local recurrence) diagnosed before (or within 3 months of ) the diagnosis of metastatic disease was used.
The primary goal of this analysis was to determine the frequency of HER2+, HER2-low and HER2-0 in this MBC population in dependency of the HR status. Furthermore, the impact of low HER2-expression on overall survival (OS) in the HR+ and triple negative population was investigated, respectively. Primarily, HER2-low was compared with HER2-0 regarding OS and first-line PFS, both in univariate and in multivariable analysis in the HR+ and triple-negative population, respectively. Secondarily, IHC 2+ was compared to IHC 1+ or IHC 0 in the two mentioned populations.
Overall survival was calculated from diagnosis of metastatic disease until death from any cause. First-line PFS was defined as time from start of first-line therapy until progression or death from any cause. In order to prevent falsely long PFS times in this retrospective analysis, patients who died more than two month after the end of therapy were censored with date of last dose.
Unadjusted and univariate PFS and OS probabilities between subgroups and were compared by the log-rank test. Multivariable adjusted hazard ratios (HR) were estimated by Cox proportional hazards models. Multivariable analysis was performed separately for HR+ and HR-subgroups with HER2-status (low vs. 0), diseasefree survival (DFS; ≥ 24 months or de novo metastatic vs. < 24 months) and visceral disease (yes vs. no) as minimum model. For inclusion of age (continuous), menopausal status (premenopausal vs. postmenopausal) and number of metastatic sites (2-3 vs. 1 and ≥ 4 vs. 1) stepwise backward selection according to Akaike's information criterion (AIC) were performed. As a result of the algorithm (Additional file 1: Table S5-S8), all variables were included in the final models. Subsequently model stability investigations were performed according to Heinze G et al. [14]. Due to nonlinear influence of age on survival, age was finally included according to menopausal status (interaction).
All tests were carried out at the 5% significance level, no p-value correction was applied. All statistical analyses were performed using R (version 4.0.2). Important packages: survminer (for survival analysis), bootStepAIC (for variable selection).

Frequency of low HER2 expression
As of 11/15/2020, 1,973 patients were included in the AGMT-MBC-Registry (Fig. 1). Out of 1,729 evaluable patients, who were diagnosed with MBC between November 2000 and August 2020, 351 (20.3%) were HER2-positive, 608 (35.2%) were HER2-low and 770 (44.5%) were completely HER2-negative (HER2-0) (Fig. 2  frequencies of all three HER2 subgroups in dependency of the HR status are shown in Fig. 2. When HER2patients were excluded, 44% of all HER2-negative patients, 48% of patients with HR+ /HER2-tumors and 33% of patients with triple-negative tumors showed a low-level expression of HER2, respectively. Compared to HER2-0 patients, patients with HER2low tumors were significantly older, were significantly more frequent de novo metastatic, HR+ and of no special type (NST) histology, respectively. Detailed patient characteristics for the overall population as well as for the three different expression levels of HER2 are provided in Table 1.  Fig. S1).

Impact of low HER2 expression on OS
In this analysis, we focused on the prognostic differences between the HER2-low and HER2-0 subgroup. The median follow-up in this population was 68.2 months (95% CI 61.6-72.3 months). In univariate analysis, HER2-low was significantly associated with a longer OS compared to completely HER2-negative disease (HR 0.84; 95% CI 0.73-0.95; P = 0.006; Additional file 1: Fig. S2). Given the unalterable influence of the HRstatus on prognosis and the uneven distribution of HRpositivity between the two HER2-subgroups, all further analyses were performed in HR+ and HR-patients separately. In the triple-negative subgroup, median OS was 16.6 months in HER2-low patients and 12.7 months in HER2-0 patients (HR 0.92; 95% CI 0.72-1.18; P = 0.535; Fig. 3A). In the HR+ subgroup, the median OS was 38.9 months both in the HER2-low and in the HER2-0 subgroup (HR 0.90; 95% CI 0.77-1.04, P = 0.160; Fig. 3B).
Similarly, in multivariable analysis including the known prognostic factors age (according to menopausal status), duration of disease-free survival and presence of visceral disease and metastatic sites at diagnosis of metastatic disease, we did not observe a statistically significant difference between patients with HER2-low and HER2-0 tumors both the HR+ (HR 0.89; 95% CI 0.74-1.05; P = 0.171; Table 2) and in the triple-negative subgroup (HR 0.92; 95% CI 0.68-1.25; P = 0.585; Table 3).

Discussion
Previous studies have shown a negative prognostic impact of HER2 IHC 2+ expression in EBC even in the absence of HER2 amplification [14][15][16][17][18][19][20]. Tumor biologic findings in early breast cancer cannot simply be transferred to the metastatic stage, since the genetic background is different between the early and the advanced disease [23] and prognostic factors can behave differently depending on the context. For example, androgen receptor (AR) expression predicted a better prognosis in HR+ EBC [24], while there was no influence on time-to-progression (TTP) on first-line endocrine therapy in MBC [25].
Here we provide, complementary evidence for the incidence of HER2-low in MBC an its influence on prognosis in the HR+ and triple-negative subgroup. Out of the whole MBC cohort (n = 1,729), 35% of patients had HER2-low tumors defined as HER2 IHC 1+ or IHC 2+ and ISH-. This corresponds to 44% of all HER2-negative patients, 48% of patients with HR+/HER2-tumors and 33% of patients with TNBC.
Patients with low HER2 expression did not have a significantly different OS compared to patients without any HER2 expression both in the triple-negative cohort (HR 0.92; 95% CI 0.72-1.18; P = 0.535; Fig. 3A) and in the HR+ cohort (HR 0.90; 95% CI 0.77-1.04, P = 0.160;  The major limitation of our analysis is that the receptor status was extracted from the pathology report and no central HER2 (and ER) testing was performed. The known inter-pathologist variability [26] could have potential impact on our results. Furthermore, the technique of staining and the details of interpretation have slightly changed over time [3]. Since the diagnosis of MBC ranged in a timeframe of 20 years, this is another potential confounder.
Currently, two phase III trials are randomizing between trastuzumab deruxtecan and investigator's choice chemotherapy in patients with HER2-low MBC. DESTINY-Breast04 (ClinicalTrials.gov identifier: NCT03734029) includes patients with HR+ and HR-disease pretreated with one or two lines of chemotherapy. DB-06 (NCT04494425), instead, recruits only patients with HR+/HER2-low MBC who have had disease progression on at least two previous lines of endocrine therapies but are naïve for chemotherapy.
Our data are well in line with those from other retrospective analyses, showing no difference in OS between patients with low HER2 expression compared to patients with HER2 0 (or 1+) [20][21][22]. In one analysis, patients older than 55 years had a statistically significant worse prognosis in case of 2+ HER2 expression compared to HER2 0/1+ patients (HR 1.45; 95% CI 1.01-2.07; P = 0.044) [20]. This observation was not confirmed in our study. In an exploratory multivariable analysis, we did not find a statistically significant influence of HER2low on OS neither in premenopausal (HR 1.10; 95% CI 0.67-1.82; P = 0.705; Additional file 1: Table S3) nor in postmenopausal women (HR 0.84; 95% CI 0.70-1.01; P = 0.069; Additional file 1: Table S4) with HR+ MBC.

Conclusion
In our analysis, about 44% of all patients with MBC, defined as HER2-negative according to the ASCO/CAP guidelines [3], showed a low expression of HER2. HER2low was more frequently in patients with HR+ tumors compared to patients with TNBC (48% vs. 33%). This potentially new target for anti-HER2 ADCs, however, did