Predictive value of HER2 expression, early response and tumor infiltrating lymphocytes (TILs) on efficacy of de-escalated pertuzumab+trastuzumab in the neoadjuvant WSG-ADAPT-HER2+/HR- trial

PURPOSE: Only limited data are available on predictive markers for use of chemotherapy-free, anti-HER2 combinations in HER2-positive early breast cancer (eBC), in particular within the molecularly distinct HER2+/HR- subgroup. BACKGROUND: In the ADAPT HER2+/HR- trial, a promising pCR of about 43-45% was found in patients treated by 4 cycles of pertuzumab and trastuzumab with either defined early response (low tumor cellularity or relative Ki-67 decrease ≥30%) or indeterminate early response (e.g. no visible tumor by ultrasound), compared to under 10% in early non-responders. In addition to early response, HER2 protein expression and stromal TIL (s-TIL) dynamics after one cycle of therapy are promising tools for identification of patients with high likelihood of pathological complete response (pCR) after therapy with double anti-HER2 blockade (e.g. by lapatinib+trastuzumab) or T-DM1. METHODS: Patients with cT1-cT4c, cN0-3 early HER2+/HR- BC (n=134) were treated with 4 cycles of P+T +/- paclitaxel d1,8,15 q3w. Primary endpoint of the study was pCR (ypT0/is, ypN0). All tumors were HR-negative (ER and PR<1%) and locally HER2 positive, i.e., 2+ with positive FISH or 3+ by immunohistochemistry (IHC). HR and HER2 status were confirmed by central pathology prior to randomization. Tumors were classified as “HER2-low” if HER2-1+ or HER2-2+ by either local or central IHC assessment, otherwise “HER2-high”. s-TILs were measured semi-quantitively according to current international consensus in triplicate at baseline and on-treatment (at cycle 2); the median of the three measurements was taken to define the quantities TIL-0 and TIL-3, respectively. “Lymphocyte-dominant subtype” at baseline and cycle 2 were defined as TIL-0≥40% and TIL-3≥40%, respectively. The present analysis characterizes the predictive impact of early response HER2 protein expression, and s-TILs on pCR under dual anti-HER2 therapy for the first time within a prospective neoadjuvant trial specific to the HER2+/HR- BC subtype. RESULTS: As previously reported, pCR was 34.4% without chemotherapy and 90.5% with chemotherapy. TIL-0 and TIL-3 were available in n=119/134 and n=103/134 patients, respectively. TIL-0 and TIL-3 were not significantly associated with pCR in the whole cohort or within the T+P arm, either modeled as a continuous variable or in terms of binary variables representing lymphocyte-dominant subtype or defined by the respective population medians. HER2-high expression was found in 87% of patients. In the T+P arm, pCR was much higher in HER2-high than HER2-low patients (40.3% vs. 0%, p=.003). In the T+P arm, 24 and 38 of 92 patients were classified as non- and responders, unclassified early response was observed in 30 of 92 patients. pCR in these groups were 8.3% vs. 44.7% vs. 42.9% respectively. Higher baseline TILs were positively associated with early response. Clinically meaningful pCR of 49% after only 4 cycles of chemotherapy-free P+T was seen in those patients with early response after one cycle of therapy and HER2-high classification, compared to 11.8% in the HER2-high/non-responder group). CONCLUSIONS: At present, a combination of baseline high HER2 expression with low cellularity after one cycle of neoadjuvant therapy – rather than s-TIL determination (at baseline or in response to therapy) – appears to be a simple and feasible tool for identification of candidates for de-escalated treatment in HER2+/HR- disease. Further research on high-precision determination of HER2-high expression (by immunohistochemistry vs. mRNA-based tools) is strongly needed for optimal patient selection for future chemotherapy de-escalation trials.