Adaptive immune system and immune checkpoints are associated with response to pertuzumab (P) and trastuzumab (H) in the NeoSphere study.

Background: Trastuzumab and pertuzumab have been shown to work by inhibiting the intracellular signaling linked to HER2 receptor activation, and cytotoxic immune mechanisms, including Fc-dependent immune cell activation. ER+ and ER-/HER2+ breast cancers (BC) are considered molecularly distinct entities (Bianchini G, ASCO 2011; Iwamoto T, ESMO 2012). Regardless of the ER status, immune gene signatures are prognostic and predictive of chemotherapy response (Iwamoto T, ESMO 2012). The gene expression profiles of pre-treatment tumor samples from patients of the NeoSphere trial of neoadjuvant pertuzumab+trastuzumab ± docetaxel (T) (Gianni L, Lancet Oncol 2012) were characterized. We assessed the association of pre-selected adaptive immune functions and key immune regulatory genes with the likelihood of achieving a pathological complete response (pCR) in NeoSphere. Methods: Baseline core biopsies were collected from 387/417 patients. Gene expression profiles (Affymetrix U133 Plus 2.0) were obtained for 367 samples (88% of all patients) that were evenly distributed among arms A (TH, n=90), B (THP, n=95), C (HP, n=98), and D (TP, n=84). ER status was defined by IHC. The primary endpoint was pCR in breast. We assessed metagenes (average expression of the associated genes) corresponding to: immunoglobulins (IgG), CD8A, MHC type I and type II (MHC1, MHC2), interferon inducible genes (I-IG) and STAT1-related genes. We also assessed the individual genes PD1, PD-L1, PD-L2, CTLA4, IFNG. Results: ER+ and ER-- tumors had differential mRNA expression for: CD8A, IgG, PD-L2, PD1, IFNG (overexpressed in ER--), and I-IG (overexpressed in ER+). Positive correlations were observed between most of these biomarkers. Only PD1 was inversely correlated with STAT1, interferon and MHC2. In logistic univariate regression analyses some biomarkers showed moderate association with pCR or residual disease without consistent patterns between treatment arms. However, a multivariate logistic regression model constructed for all the selected biomarkers revealed that high expression of PD-L1 was consistently associated with lower pCR rate in all chemotherapy containing arms (A, B and D). A similar trend was present also in Arm C, the arm with antibody treatment alone. In all arms, high expression of IFNG and/or STAT1 were associated with higher pCR rate. In multivariate analysis PD-L1, PD1 and STAT1 were associated with pCR irrespective of the ER status in combined arms A, C and D. In arm B, both CTLA4 and PD-L1 were independently associated with lower pCR in ER-- tumors. Conclusions: The association between pCR and selected immune biomarkers in patients treated with pertuzumab, trastuzumab, or both, with or without chemotherapy, supports our understanding of the key role of the immune system in contributing to HER2-targeted antibody therapy on top of signaling inhibition. High PD-L1 expression emerged for its consistent association with lower pCR. These results provide a rationale for combining HER2-targeted treatments with immune-modulating agents and may allow the prediction of treatment benefit. Validation of these results with different assays is ongoing. [ABSTRACT FROM AUTHOR]