Immune-related histologic phenotype in pretreatment tumor biopsy predicts pathologic response to neoadjuvant anti-PD-1 treatment in squamous lung cancer.

Abstract

e20540 Background: The neoadjuvant platform affords a rich and valuable resource for understanding the responses to therapy and carrying out reverse translation, including pathologic morphology. We aimed to develop a pretreatment histologic scoring system reflecting the preexisting immune response to predict the efficacy of neoadjuvant immunotherapy based on the morphological changes we mastered in the pathologic assessment after neoadjuvant immunotherapy. Methods: Surgical specimens from the 31 squamous cell lung cancer patients recruited in a phase Ib study of neoadjuvant anti-PD-1 therapy and eligible paired pretreatment biopsies from 15 of them were included in this study. The posttreatment surgical specimens were assessed according to the immune-related pathologic response criteria. Immune-related histologic phenotype assessment criteria (irHPC) were developed based on the pathologic features identified after neoadjuvant anti-PD-1 treatment. Three pathologists trained for irHPC independently scored the HE slides of the 15 pretreatment tumor biopsies according to irHPC. Results: Whether necrosis was included in the calculation of percent of residual viable tumor (%RVT) or not had almost no effect on the consistency of pathologic assessment ( P= 0.811) and the histological response grouping. The inter-pathologist variability of assessing %RVT with immune-activated phenotype was not statistically significant ( P= 0.480). Four immune-related features of pretreatment biopsies were included for calculating the predictive score, including three positive features (tumor-infiltrating lymphocytes, tumor-infiltrating eosinophils and dense plasma cells in stroma) and one negative feature (tumor-infiltrating neutrophils) according to the developed irHPC scoring system. The trained pathologist accurately predicted 6 out of 8 patients in the cPR/MPR group and 5 out of 7 patients in the non-cPR/MPR group according to irHPC. For inter-observer reproducibility using “2 points” as the cut-off point, the overall percent agreement (OPA) was 77.8%. The reliability between pathologists for a binary tumor evaluation showed “moderate” agreement (κ = 0.54). Conclusions: The irHPC scoring system reflecting the preexisting immune response could be used to predict the pathologic response of neoadjuvant immunotherapy, but still needs the larger trails to verify.