HER2 Nuclear Translocation and Decreased T-DM1 Ecacy in HER2 Positive Advanced Breast Cancer Treated with Dual HER2 Blockade. The SePHER Study.

Background: ErbB2-targeting agents have dramatically changed the therapeutic landscape of ErbB2+ advanced breast cancer (ABC). However, their optimal sequence of administration deserves further investigation. Methods: The biology of ErbB2 was investigated through sequential treatments in vitro, in ErbB2+ breast cancer cell lines resistant to trastuzumab, pertuzumab, and their combination. We analyzed data from 555 ErbB2+ ABC patients treated with trastuzumab emtansine (T-DM1) and explored the ecacy of T-DM1 in the 371 patients who received it in second-line. Survival estimates were graphically displayed in Kaplan Meier curves, compared by log rank test and, when possibile, conrmed in multivariate models. Results: We show here lower activity of T-DM1 in two HER2+ breast cancer cell lines resistant to trastuzumab+pertuzumab, as compared to trastuzumab-resistant cells. Reduced T-DM1 ecacy is associated with a marked reduction of HER2 expression on the cell membrane and its nuclear translocation. Membrane-HER2 downregulation was conrmed in biopsies of four trastuzumab/pertuzumab-pretreated patients. Among 371 patients treated with second-line T-DM1, median overall survival (mOS) from diagnosis and median progression-free survival to second-line treatment (mPFS2) were 52 and 6 months in 177 patients who received trastuzumab/pertuzumab in rst-line, and 74 and 10 months in 194 pertuzumab-naïve patients (p=0.0006 and 0.03 for OS and PFS2, respectively). Conclusions: Our data support the hypothesis that the addition of pertuzumab to trastuzumab reduces the amount of available plasma membrane HER2 receptor, limiting the binding of T-DM1 to cancer cells. This may help interpret the less favorable outcomes of second-line T-DM1 in trastuzumab/pertuzumab pretreated patients compared to their pertuzumab-naïve counterpart.


Background
Human epidermal growth factor receptor 2 (HER2) is a member of the HER family of receptor tyrosine kinases, also including the epidermal growth factor receptor (EGFR), HER3, and HER4. Ligands binding to EGFR, HER3 and HER4 induce homo-and heterodimerization among the family members. Despite the lack of speci c ligands, activated HER2 homodimerizes in HER2 positive (HER2+) breast cancer (BC) cells and is then recruited as a preferred partner in heterodimers, resulting in the activation of cancer-driving pathways (1). HER2 overexpression and/or gene ampli cation occurs in approximately 15-20% of BC, and is associated with a more aggressive behavior, with high rates of cell proliferation and metastasis, and poor patient outcomes (2). HER2 + advanced breast cancer (ABC) has signi cantly bene ted from the approval of several HER2-targeting agents in the last decades. Trastuzumab, a monoclonal antibody targeting HER2, has revolutionized the therapeutic landscape of HER2 + ABC (3). Despite this, up to 40% of ABC patients show innate trastuzumab-resistance, and most patients develop acquired resistance whithin the rst year of trastuzumab treatment (4,5). The approval of three additional anti-HER2 agents, i.e., lapatinib, pertuzumab, and trastuzumab emtansine (T-DM1), has converted HER2 + ABC into a highly treatable disease, with more favorable outcomes (6)(7)(8)(9). Pertuzumab is a monoclonal antibody binding HER2 at a different site compared with trastuzumab. A more comprehensive signaling blockade underlies the noticebly enhanced antitumor activity of trastuzumab and pertuzumab combination treatment (10).
Results from the CLEOPATRA trial showed an unprecedented median overall survival (mOS) advantage of 15.7 months in the pertuzumab arm, and the double-block combination has therefore become the new standard rst-line treatment in HER2 + ABC (7). T-DM1 is an antibody-drug conjugate of trastuzumab with emtansine (DM1), an antimicrotubule maytansine derivative (11,12). The activity of T-DM1 depends on both trastuzumab antitumor effects and intracellular DM1. Following T-DM1 binding to membrane HER2 receptor, the HER2-T-DM1 complex enters into the cell via receptor-mediated endocytosis (13). Subsequently to the release from the lysosome, DM1-containing metabolites inhibit microtubule assembly, causing cell death (14).
In second-line treatment, after treatment with taxane and trastuzumab or as rst-line in patients with rapid progression after adjuvant trastuzumab (≤ 6 months), TDM-1 has shown greater e cacy than lapatinib and capecitabine in the phase III EMILIA trial (8). Consequently, TDM-1 has become the standard secondline treatment in HER2 + ABC patients.
To the aims of the present study, it is noteworthy that patients accrued in randomized trials of T-DM1 had not received prior pertuzumab. Therefore, we lack solid evidence on T-DM1 e cacy following trastuzumab/pertuzumab-containing regimens. Data from observational trials are limited. We have previously shown evidence of lower T-DM1 e cacy in trastuzumab/pertuzumab-pretreated patients providing data from a retrospective, multicentric study of 250 HER2 + ABC patients (15). In a further retrospective evaluation of T-DM1 activity as second-line or later treatment from Dzimitrowicz and colleagues (16), results in terms of tumor response rates and progression free survival (PFS) appeared less favorable than those reported in randomized trials of T-DM1.
These ndings suggest the need to investigate the biology of HER2 through sequential treatments in order to de ne the molecular basis for the appropriate therapeutic approach.
Based on the above reported evidence, we explored the effects of the exposure to trastuzumab and/or pertuzumab on HER2 receptor expression and cellular localization in HER2 + BC cell lines, and their effects on T-DM1 activity. The pre-clinical experiments were paralleled by the conduct of a large, multicentric, observational study, i.e., the SePHER study, Administration Sequence in Pertuzumab-pretreated HER2 + ABC patients, aimed to explore the e cacy of T-DM1 in light of prior trastuzumab/pertuzumab treatment in the real-world setting.

Study approval
The SePHER study is a multicenter, observational trial with retrospective design including HER2 + ABC Patients' selection Information on demographics, clinical, histopathological and immunohistochemical (IHC) features, antitumoral therapies and related outcomes were retrieved from patients' medical records by speci cally trained research assistants. All included patients were treated for metastatic disease. Each patient was evaluated during treatment according to the follow-up strategies of each center. Clinical response was evaluated by response evaluation criteria in solid tumours (RECIST) criteria, version 1.1. Anonymized data were entered into a dedicated database with a SPSS operating interface. Median follow up was calculated starting from diagnosis of metastatic disease to death or date at last follow up. Endpoints for e cacy outcome included progression free survival (PFS) and overall survival (OS). Progression free survival for any speci c line of treatment was calculated from the time of treatment start to the time of disease progression, interruption of treatment for toxicity, death or loss to follow-up. Overall survival was calculated starting from diagnosis of metastatic disease to death or last follow-up. Median PFS (mPFS) and OS (mOS) were calculated using the Kaplan-Meier product limit estimator method. Patients treated in rst-line with T-DM1 (N: 25). The groups de ned upon treatments' sequence, as previously speci ed, were selected by hypothesizing a possible effect of the administration sequence on the main clinical outcomes, i.e., OS and PFS to rst-line, second-line and third-line of treatment (PFS1, PFS2 and PFS3, respectively). Survival estimates were rst computed for the whole cohort and then by treatments' sequence. Subsequently, data on OS and PFS2 were also analyzed across strata de ned by IHC tumor features, namely, estrogen receptor (ER) and progesterone receptor (PgR) positive (triple positive, TP), ER or PgR positive, and (both) hormonal receptors (HRs) negative.
Pathology assessment was performed in surgical specimens of primary tumors by dedicated pathologists at the participating centers as per national standards. Estrogen receptor and PgR status were determined at each center by IHC according to the local standards. Positivity was considered at a cut-off of ≥ 1%. Generation of drug-resistant cell lines.
Drug-resistant cells were obtained by continuous exposure of HER2 + BT474 and SkBr3 cell lines, obtained from the American Type Culture Collection (ATCC), to 20 µg/ml trastuzumab or pertuzumab, or 10µ g/ml trastuzumab + 10µ g/ml pertuzumab, for two months, followed by 7 months of culture in medium supplemented with 50 µg/ml trastuzumab or pertuzumab, or 25 µg/ml trastuzumab + 25 µg/ml pertuzumab. Further details are reported in the Supplementary materials and methods and are available online Statistical Analysis.
Within the overall cohort of the 555 patients, the associations of interest were also evaluated in light of: a.
T-DM1 administration in second-or subsequent line, and b. molecular subgroups, with these latter being set based on the results of IHC analysis and according to the criteria fully reported in the patients' selection paragraph. The covariates used in the Cox uni/multivariate models included the following categorical variables: " rst-line pertuzumab" (yes vs. no) and "treatment sequence", which both concurred to de ne the ve categories described in detail in the methods section; age; IHC subtype; Ki-67 (> 20% vs ≤ 20%); metastasis at diagnosis (yes vs no); number of metastatic sites (> 1 vs 1); visceral metastasis (yes vs no); brain metastasis (yes vs no); bone-only metastasis (yes vs no), lenght of PFS1 and disease free interval (DFI). This latter was calculated from the time of surgery to the time of metastatic disease diagnosis. Variables testing signi cant in univariate analyses were further tested in multivariate models.
The level of statistical signi cance was set at p ≤ 0.05, with a 95% con dence interval (95%CI). The SPSS software (SPSS version 21.0, SPSS Inc., Chicago, IL) was used for all statistical evaluations.
Regarding the experiments in cell lines, all data were reported as mean +/-standard deviation.
Differences were considered statistically signi cant when p ≤ 0.05, with a 95%CI. Student's t test was performed for the comparison of results from all different tests (*p < 0.05, **p < 0.001, ***p < 0.0001).

Generation of trastuzumab, pertuzumab, and trastuzumab + pertuzumab resistant SkBr3 and BT474 cells in vitro
To investigate the mechanisms underlying lower T-DM1 e cacy following trastuzumab/pertuzumab in ABC patients, we established trastuzumab (T), pertuzumab (P), and trastuzumab + pertuzumab (T + P)resistant HER2 + SkBr3 and BT474 cell lines (Fig. 1A). In agreement with previously reported data, Western Blot (WB) analysis revealed that short-term trastuzumab treatment inhibits the phosphorylation of HER family members and AKT, but not of ERKs (Fig. 1B, leftpanels) (20). We observed reduced HER2 activity and expression and marked induction of ERKs phosphorylation in T + P cells as compared to parental and both T and P cells, whereas HER3 and EGFR were upregulated despite the reduced phosphorylation, in both BT474 and SkBr3 (Fig. 1B, right most panels). Resistant cell lines exhibited a higher proliferation rate ( Fig. 1C) and invasive capability (Fig. 1D) compared to control cells.The number of invading T + P cells was signi cantly higher compared to T cells in both cell lines.
Dual HER2 blockade is associated with reduced T-DM1 e cacy due to HER2 downregulation Following T-DM1 treatment, the percentage of responsive T + P cells was signi cantly lower compared to T cells in BT474 (p < 0.0001), and in SkBr3, although at a lower level of statistical signi cance (p < 0.05) ( Fig. 2A).
The reduction of total and phosphorylated HER2 was further con rmed in T-DM1-low-responders T + P cells (Fig. 2B). T-DM1 induced a marked down-regulation of HER2 as well as HER3 and EGFR, in parental, T and P BT474 cells, whereas T + P BT474 cells maintained unmodi ed levels of these receptors (Fig. 2C).

T-DM1 inhibited ERKs phosphorylation in control cells, T and P cells, whereas it considerably induced
ERKs activation in T + P cells, in both cell lines (Fig. 2C, data available upon request). The downregulation of HER2 induced by trastuzumab/pertuzumab combination was con rmed in vivo, by the immunohistochemical assessment of bioptic specimens from four trastuzumab/pertuzumab-treated ABC patients prior to and following exposure to double-block (Table 1). Representative images of eosinhematossilin and HER2 staining from two patients are shown in Fig. 2D. The combination trastuzumab+pertuzumab induces HER2 nuclear translocation When administered in vitro as short-term treatments, trastuzumab and pertuzumab alone barely downregulate HER2 (17)(18)(19), while their combination induces a stronger HER2 downregulation (18,19). To investigate the long-term effects of trastuzumab and pertuzumab on HER2 subcellular distribution, we performed immuno uorescence experiments. As shown in Fig. 3A, in control cells, T and PBT474 and SkBr3 cells, HER2 was mainly localized at the plasma membrane level, and concentrated on cellular protrusions (20). A diffuse cytoplasmic signal was also present. Conversely, T + P cells lost membrane-HER2 and retained HER2 cytoplasmic distribution (Fig. 3A). Western Blot analysis of the cytoplasmic and nuclear fractions of control and resistant cells showed a marked translocation of HER2 to the nucleus in T + P BT474 and SkBr3 cells (Fig. 3B). Nuclear HER2 was phosphorylated, suggesting its active involvement in transcriptional control mechanisms (21,22). A short trastuzumab/pertuzumab pre-exposure of parental BT474 did not affect T-DM1 e cacy compared to control cells, and no HER2 nuclear translocation was observed ( Fig. 3C and 3D). Hence, prolonged exposure to trastuzumab/pertuzumab induces the loss of membrane HER2 and its nuclear traslocation, that represent a mechanism of acquired resistance, which in turn reduces the T-DM1 targeting potential and e cacy.
Patients' cohort description Overall, 555 patients were treated with T-DM1 for ABC.      No differences by trastuzumab/pertuzumab-pretreatment emerged for mOS, mPFS1 and mPFS2 in patients who received T-DM1 in third-line or beyond (Table 4) .
When the overall cohort of 555 patients was strati ed by IHC subtype, a lower mOS in pertuzumab pretreated patients was con rmed only in hormonal receptors (HRs)-negative patients (p < 0.0001), while a shorter mPFS2 was observed in both TP (p = 0.04) and HRs-negative patients (p = 0.003). However, multivariate analysis showed no effect of the IHC-de ned subtype on mOS, mPFS1 or mPFS2.

Discussion
Resistance to HER2 targeting agents is a challenging topic in BC. Several pre-clinical studies explored potential resistance mechanisms to T-DM1, involving reduction of lysosomal proteolytic activity (23) Nuclear HER2 has been reported to act as a transcriptional regulator and represents an independent prognostic factor of poor clinical outcome (21,22). In preclinical trastuzumab-resistant BC models, the inhibition of nuclear HER2 suppresses cell growth, indicating that nuclear HER2 is the major proliferation driver in trastuzumab-resistant BC. In this context, the inability of trastuzumab to disrupt the neregulinβ1induced assembly of a nuclear HER2/HER3/STAT3 transcriptional complex has been demonstrated (28).
The lack of HER2 nuclear translocation following exposure to trastuzumab or pertuzumab alone highlights the synergy emerging from the combination of these two drugs. The issue of a possible decrease in T-DM1 e cacy if given immediately after the double pertuzumabbased HER2 double-block has not been exhaustively addressed in previous studies (8,9). The EMILIA (8) and TH3RESA(9) trials were the two pivotal randomized phase III clinical studies that brought T-DM1 as a standard of care in second-line or beyond for patients with HER2 + advanced BC that progressed to standard treatments. TH3RESA trial showed clinical advantage by using T-DM1 compared to treatment of choice by the clinician also in patients that had received lapatinib and capecitabine, while EMILIA trial showed superiority of T-DM1 in second-line even when compared head to head with the lapatinib plus capecitabine regimen. Unfortunately, none of the patients included in the EMILIA and TH3RESA trials had received pertuzumab before being treated with T-DM1. Data from prospective studies is lacking in this context. Evidence on the topic under debate comes from three recent retrospective studies showing lower response rate and shorter survival outcomes in patients treated with T-DM1 following pertuzumab administration (15,16,29) In a previous our retrospective multicentric study involving 250 pretreated HER2 positive BC patients we showed lower e cacy of T-DM1 in the pertuzumab-pretreated cohort in comparison with the trastuzumab pretreated group (15). In the Dzimitrowicz paper, authors showed results that are congruent with our ndings (16). In this study, including only patients that had received a pertuzumab-trastuzumab-based rst-line, a lower e cacy of T-DM1 in terms of response rate was reported when compared to the response rates observed in the randomized clinical trials, where patients were only trastuzumab resistant. In the Noda-Narita et al retrospective study enrolling 42 advanced HER2 positive patients, median PFS and objective responses were lower in the group pretreated with pertuzumab/trastuzumab in comparison to the trastuzumab subgroup (29).
The main limitation of the observational section of the present study is its retrospective, multicenter design, which per se represents a considerable source of data heterogeneity. Moreover, the lack of central assessment on IHC features of primary and metastatic lesions deserves mentioning, although quality controls routinely performed at the pathology labs of the institutions involved increase our con dence in data quality. At the same time, the involvement of a relevant number of cancer centres/oncologic divisions, i.e., N: 45, has allowed to collect and analyze the largest amount of data ever made available to investigate the e cacy of T-DM1 following trastuzumab/pertuzumab-based treatment. Beyond its intrinsic limitations and bias, this approach allowed us to con rm the reduced T-DM1 e cacy following dual HER2 blockade by trastuzumab/pertuzumab in HER2 + ABC patients. The main strength of the present study is that, to our knowledge, we rst reported on HER2 downregulation as a key mechanism underlying lower T-DM1 e cacy observed in the clinical setting when this drug is administered as second-line therapy in trastuzumab/pertuzumab-pretreated HER2 + ABC patients. Results from the experiments in bioptic specimens of trastuzumab/pertuzumab pretreated ABC patients were further con rmative.
Although hypothesis-generating, data from the study herein presented, as well as from prior similar studies within this same research pipeline, are limited in nature. Still, they provided an appropriate ground in terms of preliminary evidence, on which we designed a randomized clinical trial investigating the optimal treatment sequence in HER2-positive ABC patients. In more detail, our team at the IRCCS Regina Elena National Cancer Institute is the coordinating center of the STEP trial, an active randomized multicenter prospective trial exploring the optimal Sequence TrEatment in HER2 + Pertuzumab-pretreated ABC patients. The STEP trial was granted formal approval and nancial support by the Italian Ministry of Health (project code: GR-2018-12367431). Evidencefrom the STEP and similar ad hoc, prospective randomized trials are eagerly awaited to delineate the optimal treatment sequence in HER2 + ABC patients, in order to gain more favorable treatment outcomes in this patients' population.

Conclusion
Overall, our ndings suggested HER2 downregulation following dual HER2 blockade by trastuzumab/pertuzumab as a key mechanism underlying lower T-DM1 e cacy as second-line therapy in HER2 + ABC patients. Results from our retrospective study showed indeed lower T-DM1 e cacy in terms of mOS and mPFS in 177 patients who received trastuzumab/pertuzumab in rst-line, as compared to 194 pertuzumab-naïve patients. In addition, we showed HER2 nuclear translocation in trastuzumab + pertuzumab-resistant HER2 + BC cell lines in vitro.

Page 19/27
The design of prospective randomized trials may lead to delineate the optimal treatment sequence in HER2 + ABC, and to more favorable outcomes in these patients' population.    Trastuzumab+pertuzumab combination is associated with HER2 downregulation. Cell viability of CTR, T, P, and T+P SkBr3 and BT474 cells treated with 0.1μg/ml and 1μg/ml T-DM1, respectively, for 48 hours was evaluated by Crystal Violet Assay (A). The results are expressed as percentage of T-DM1-responsive cells relative to control cells, as mean +/-standard deviation. Total cell lysates from control, T, P, and T+P cell lines were analysed by WB for expression of phosphorylated and total HER2 (B).The expression of phosphorylated and total HER2, HER3, EGFR, AKT and ERKs was evaluated by WB following exposure to 1μg/ml T-DM1 for 48 hours (C). Representative pre-and post-therapy sections from 2 ABC patients, stained by immunohistochemistry with eosin-hematossilin and anti-HER2 antibody (D).

Figure 3
Prolonged trastuzumab+pertuzumab induces HER2 nuclear translocation. Control, T, P, and T+P cell lines were plated on poly-l lysine coated slides, and stained 24 hous later with anti-HER2 (green signal) (A).
These cells were counterstained with Hoechst to highlight nuclei. Red arrows indicate HER2 localization on cellular protrusions. Cytoplasmic and nuclear fractions extracted from control, T, P, and T+P cells were analysed by WB for the expression of phosphorylated and total HER2. Lamin A and α-tubulin were used to validate purity of nuclear and cytoplasmic extracts respectively (B). Following pre-treatment with 5μg/ml trastuzumab + 5μg/ml pertuzumab for 72 hours, cell viability of control cells, pre-treated and T+P BT474 exposed to1 g/ml T-DM1 for 72 hours was evaluated by Crystal Violet Assay (C).Cytoplasmicand nuclear fractions of control, pre-treated and T+P BT474 cells were analysed by WB for the expression of total HER2 (D).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.