Changes in L-phenylalanine concentration is a risk factor for NSCLC patients who received PD-1 Ab treatment: a case-control study

Abstract

Background Chemotherapy combined with checkpoint blockade antibodies that target programmed cell death protein (PD-1) has achieved remarkable success in non-small cell lung cancer (NSCLC). However, only a small percentage of NSCLC patients experience long-term benefits. We aimed to design and validate a clinical predictive indicator based on serum metabolism for chemotherapy in combination with PD-1 treatment. Methods Here, we collected serial blood samples from 49 patients with NSCLC who underwent combined chemotherapy and PD-1 treatment and from 34 patients who received only chemotherapy. Samples were collected before treatment initiation (baseline) and after treatment. Non-targeted metabolomics was used to explore the different metabolites in patients. Results L-phenylalanine was identified as a predictor in patients with NSCLC during chemotherapy combined with PD-1, which was not found in patients receiving chemotherapy alone. An elevated ratio of L-phenylalanine concentration (two cycles after completion of treatment VS the initiation of treatment) was associated with improved progression-free survival [hazard ratio [HR] = 1.8000, 95% CI, 0.8566–3.7820, p < 0.0001] and overall survival (HR = 1.583, 95% CI, 0.7416–3.3800, p < 0.005). Furthermore, we recruited two validation cohorts (cohort 1:40 patients; cohort 2:30 patients) with blood samples taken at baseline and after one cycle of treatment to validate the sensitivity and specificity of L-phenylalanine prediction. The area under the curve (AUC) values of the L-phenylalanine concentration ratios in validation cohorts 1 and 2 were 0.8650 and 0.8400, respectively. Our results demonstrate that L-phenylalanine is a potential and novel predictive biomarker for chemotherapy combined with PD-1 in patients with NSCLC. Conclusions We constructed a serum metabolite prediction model for combined treatment by combining two independent predictors of patient response. This model can assess the risk of response to treatment in patients with NSCLC in the early stages of treatment and may help stratify and optimize clinical decisions.