Pan-tumor prognostic value of multiple immune protein expressions.

Abstract

2618 Background: Using multiple immune-checkpoint proteins (ICP) screening in clinical routine could improve the evaluation of patients’ prognosis and ultimately tailor their treatment choice. We have evaluated this hypothesis in the context of early drug clinical trials. Methods: Patients included in MOSCATO-02 trial had refractory cancers and were candidate for phase 1 study. They were proposed to have a biopsy on an accessible tumor site for the analysis of four proteins by immunohistochemistry (IHC) and RNAseq: PD-L1, CD3, CD8 and FOXP3. Quantification of IHC staining was separated between intratumoral, intersitial and stromal by semi-quantitative method. Their relations to prognosis have been evaluated by survival Random Forest and compared to classical prognosis clinical variables, such as age and RMH score (calculated by the number of metastatic sites, lactate dehydrogenase (LDH) and serum albumin). Results: From April 2016 to September 2017, 228 patients included in MOSCATO-02 had a successful biopsy procedure with available IHC expression analysis. The main tumor subtypes were gastro-intestinal, urological, head and neck, breast and lung. RNAseq analyzes were performed for two thirds of the patients (N=170). Median overall survival was 8.1 months (CI95% 7.79 – 10, 65). We found that, in a cohort of phase I patients, RMH score was the most important variable used to estimate prognosis. Prognosis value of immune proteins were considerably inferior compared to clinical criteria. Among those proteins, the percentage of PD-L1 low score (1+) and average staining intensity of CD3 were the most valuables for prognosis evaluation. Variables with very few importance to prognosis estimation were CD8 and FOXP3 IHC scores, biopsy site and cancer types, subsequent treatments by immunotherapies or targeted therapies. Conclusions: In this cohort of patients with refractory cancers, the RMH score is confirmed as highly prognosis. Immune proteins could be used as a support to guide patient’s selection but does not constitute effective prognosis criteria.