Serum metabolomic signatures predict clinical outcomes in advanced non-small cell lung cancer treated with pembrolizumab plus platinum-based chemotherapy

Abstract

Background: Predicting response to pembrolizumab plus chemotherapy in advanced non-small cell lung cancer (NSCLC) remains challenging. Serum metabolomics offers a promising approach to identify biomarkers capturing host-tumor metabolic interactions.

Methods: We conducted nuclear magnetic resonance (NMR) spectroscopy-based metabolomic analysis on 216 longitudinal serum samples from 36 patients with advanced NSCLC receiving first-line pembrolizumab plus chemotherapy. We examined how baseline and dynamic metabolite profiles related to survival and impending disease progression, applying multivariate analyses and Random Forest (RF) modelling.

Results: Lower serum levels of branched-chain amino acids (BCAAs) valine and isoleucine were associated with disease progression within 60 days. Overall survival was linked to distinct metabolomic signatures: long-term survivors showed higher serum levels of various lipids, including total phospholipids, sphingomyelin, and apolipoproteins A1 and A2. In contrast, patients who died during follow-up had higher inflammatory markers, including glycoprotein acetyls and mannose. An RF model predicted survival status with high accuracy (AUC = 0.93), with sphingomyelin, apolipoprotein A2, and glycoprotein acetyls B among the top contributors.

Conclusion: Serum metabolomic profiles are closely linked to clinical outcomes in advanced NSCLC treated with pembrolizumab plus chemotherapy. Key metabolites - particularly BCAAs, lipids, and inflammatory markers - emerge as promising non-invasive biomarkers for predicting progression and survival.

Keywords: NSCLC; biomarker; branched-chain amino acids; immunotherapy; metabolomics; phospholipids.