Metabolomics as predictive biomarker of response to immune checkpoint inhibitors in advanced non-small cell lung cancer (NSCLC)

Metabolomics as predictive biomarker of response to immune checkpoint inhibitors in advanced non-small cell lung cancer (NSCLC) Background: Nivolumab and Pembrolizumab, a human antibodies against the checkpoint programmed cell death protein 1 (PD-1), have demonstrated to sustain overall survival in advanced NSCLC. However not all patients benefit. No marker is currently considered as predictive of response. In oncology field metabolomics has shown immense potential for early diagnosis, prognosis, individual monitoring and drug therapy design. Nuclear Magnetic Resonance (NMR)-metabolomics is an efficient and highly reproducible platform for the analysis of biofluids that can be collected non-invasively; serum profiles are used to underline signature both before and after a give drug therapy, to inform on treatment outcomes and to predict the nivolumab immunotherapy response. Methods: 33 patients were treated with Nivolumab as second line therapy after chemotherapy and 11 patients were treated with pembrolizumab as first line therapy. Serum sample from 48 patients with an advance form of NSCLC were collected before the treatment and efficacy was clinically and radiologically evaluated during the whole duration of the therapy. H-NMR spectra were acquired for each serum sample at a 600 Mhz spectrometer, optimized for metabolomics. The obtained NMR data were analyzed using multivariate statistical analyses. Results: The best result we obtained is the discrimination between responders and non-responders according to the number of cycles of therapy. We decided to take into accounts the number of cycles performed as surrogate method of response evaluation. The general idea is that the higher is the number of cycles performed, the higher is the improvement in the life-spun of the patients, thus the better is the outcome of the treatment. Thus, all the subjects that performed at least 14 cycles (11 subjects) were considered as responders, less than 10 cycles (17 subjects) were considered as non-responders. Using this classification, the discrimination power ones significantly improved, obtaining a discrimination accuracy of 84% Conclusions: These results show that metabolomic profiling of serum samples collected before the beginning of nivolumab and pemrbolizumab can inform on treatment outcomes and can be successfully used to predict response to treatment. The fulfilment of this project will pave the way for a prospective larger confirmatory study.