Machine learning reveals a PD-L1-independent prediction of response to immunotherapy of non-small cell lung cancer by gene expression context.

Current predictive biomarkers for PD-1 (programmed cell death protein 1)/PD-L1 (programmed death-ligand 1)-directed immunotherapy in non-small cell lung cancer (NSCLC) mostly focus on features of tumour cells. However, the tumour microenvironment and immune context are expected to play major roles in governing therapy response. Against this background, we set out to apply context-sensitive feature selection and machine learning approaches on expression profiles of immune-related genes in diagnostic biopsies of patients with stage IV NSCLC. RNA expression levels were determined using the NanoString nCounter platform in formalin-fixed paraffin-embedded tumour biopsies obtained during the diagnostic workup of stage IV NSCLC from two thoracic oncology centres. A 770-gene panel covering immune-related genes and control genes was used. We applied supervised machine learning methods for feature selection and generation of predictive models. Feature selection and model creation were based on a training cohort of 55 patients with recurrent NSCLC treated with PD-1/PD-L1 antibody therapy. Resulting models identified patients with superior outcomes to immunotherapy, as validated in two subsequently recruited, separate patient cohorts (n = 67, hazard ratio = 0.46, p = 0.035). The predictive information obtained from these models was orthogonal to PD-L1 expression as per immunohistochemistry: Selecting by PD-L1 positivity at immunohistochemistry plus model prediction identified patients with highly favourable outcomes. Independence of PD-L1 positivity and model predictions were confirmed in multivariate analysis. Visualisation of the models revealed the predictive superiority of the entire 7-gene context over any single gene. Using context-sensitive assays and bioinformatics capturing the tumour immune context allows precise prediction of response to PD-1/PD-L1-directed immunotherapy in NSCLC.

Informations de copyright

Copyright © 2020 Elsevier Ltd. All rights reserved.

Déclaration de conflit d'intérêts

Conflict of interest statement M.W. reports honoraria from Boehringer Ingelheim, Novartis, Roche and Takeda and research funding from Bristol Myers Squibb and Takeda. F.M. reports research funding from Bristol Myers Squibb. Henning Reis reports a consulting and advisory role for Bristol Myers Squibb; honoraria from Roche and Bristol Myers Squibb; travel support from Philips, Roche and Bristol Myers Squibb; research funding from Bristol Myers Squibb and share ownership from Bayer. M.G. reports travel support from MSD Sharp & Dohme. J.K. reports a consulting and advisory role without personal honoraria for Roche, Boehringer Ingelheim, Bristol Myers Squibb, MSD and Takeda. T.H. reports honoraria from Bristol Myers Squibb, Chugai, MSD Sharp & Dohme and Roche and a consulting and advisory role for Bristol Myers Squibb and Chugai. M.M. reports honoraria from Roche and Boehringer Ingelheim. W.E.E.E. reports honoraria from Eli Lilly, Boehringer Ingelheim, Pfizer, Novartis, Roche, Merck, Bristol Myers Squibb, Amgen, GlaxoSmithKline, Astellas, Bayer, Teva, Merck Serono, Daiichi Sankyo and Hexal; a consulting or advisory role for Eli Lilly, Boehringer Ingelheim, Novartis, Pfizer, Roche, Merck, Bristol Myers Squibb, Astellas, Bayer, Teva and Daiichi Sankyo and research funding from Eli Lilly (institutional). C.A. reports research funding from Bristol Myers Squibb. K.D. reports a consultancy/advisory role for Boehringer Ingelheim and Novartis and honoraria from Boehringer Ingelheim. M.S. reports consultancy for AstraZeneca, Boehringer Ingelheim, Bristol Myers Squibb, Celgene, Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen (IQWiG), Lilly and Novartis; honoraria for CME presentations from Alexion, Boehringer Ingelheim, Celgene, GlaxoSmithKline, Lilly and Novartis; research funding to the institution from Boehringer Ingelheim, Bristol Myers Squibb and Novartis and other support from Universität Duisburg-Essen (patents). All the remaining authors declared no conflicts of interest.