Determinants of response to checkpoint inhibition in ovarian cancer.

Abstract

e17538 Background: Despite evidence that ovarian cancer (OC) is an immunogenic disease, single-agent checkpoint inhibitors have shown limited response. A clinical trial was conducted evaluating dose-dense paclitaxel and pembrolizumab in platinum-resistant OC. Our aim was to evaluate for biomarkers of response to this combination. Methods: Immunohistochemistry was performed to identify PD–L1 and tumor infiltrating lymphocytes (TIL) in pretreatment formalin fixed paraffin embedded tissue. We performed flow cytometry on matched blood to identify presence of immunosuppressive myeloid-derived suppressor cells (MDSCs) in circulation. We performed T-cell receptor (TCR) sequencing on tumor and matched blood to evaluate whether an oligoclonal or polyclonal immune response would be associated with outcomes. On selected patients we assessed peripheral T-cell responses to treatment using an interferon-ɣ enzyme-linked immunospot (IFN-ɣ ELISpot) assay between excellent and poor responders. Cox proportional hazards model was used to estimate hazards ratios (HR) for response in relation to each biomarker. Kaplan-Meier analysis and log-rank test were used to evaluate associations between progression free survival (PFS), overall survival (OS) and each biomarker. Results: Analysis was limited to 34 patients with tumor available and one RECIST response evaluation. Odds of response trended higher (OR = 4.1) in patients with PD-L1 (+) than patients with PD-L1 (-) tumors (p = 0.061). When restricting analysis to best response of CR/PR versus PD, the odds was 13.0 (p = 0.041). Patients with PD-L1(+) tumors had a HR for progression 0.56 (0.26-1.22; p = 0.144). Presence of TILs was classified as 0-3 (0-1 low, 2-3 high). High TIL presence was not associated with response, but was associated with improved OS and trend toward improved PFS; HR 0.29 (0.09-0.90, p = 0.022) and HR 0.47 (0.21-1.05, p = 0.06). Pre- and post-treatment levels of MDSCs were not associated with response or survival, though a trend was noted; a one unit increase in pretreatment MDSCs leads to 5% decrease in the odds of being responder. Immunoseq analysis of TCR repertoire pre- and post-treatment suggested that in increase in productive clonality was associated with increased odds of being a responder; this was not statistically significant. Notably, IFN-ɣ ELISpot assay demonstrated a significant T-cell response against neoantigens post-treatment in patients who had a partial response and long overall survival. Post treatment T-cell response showed a significantly mixed response to neoantigens in the patient analyzed who had progressive disease. Conclusions: Finding biomarkers predicting benefit from checkpoint inhibitors has been challenging. Though inconclusive, these data suggest biomarkers that may be associated with this combination. T-cell response to neoantigens in patients with response to treatment suggests neoantigens as a potential target for therapeutic direction.