Serum cytokine analysis in a cohort of advanced non-small cell lung cancer treated with PD-1 inhibitors reveals predictive markers of CXCL12

BACKGROUND: The circulating predictive factors for the outcomes of advanced non-small cell lung cancer (NSCLC) patients receiving immune checkpoint inhibitors (ICIs) remain elusive. We aimed to assess the predictive value of circulating cytokines for outcomes. METHODS: Serum samples of 102 advanced-stage NSCLC patients who underwent immunotherapy were collected at baseline. The relative levels of 37 cytokines were detected. PD-L1 expression was also analyzed. RESULTS: Higher serum CXCL12 levels (top 33%) were a poor predictive biomarker for durable clinical benefit (DCB) (23.5% vs. 72.1%, p<0.001), progression-free survival (PFS) (3.76 vs. 14.40 months; p<0.001) and overall survival (OS) (12.20 vs. 44.84 months; p=0.008). Compared with PD-L1-negative patients, PD-L1-positive patients had a significantly higher objective response rate (ORR) (70.0% vs. 28.8%, p<0.001) and a prolonged mPFS (25.35 vs. 4.64 months, p=0.003) and tended to have an increased mOS (44.84 vs. 20.42 months, p=0.087). A signature comprising PD-L1<1% and the top 33% CXCL12 level was associated with the lowest ORR (27.3% vs. 73.7%, p<0.001) and DCB (27.3% vs. 73.7%, p<0.001) and the worst mPFS (2.44 vs. 25.35 months, p<0.001) and mOS (11.97 vs. 44.84 months, p=0.007). Area under the curve (AUC) analyses of PD-L1 expression, CXCL12 level and PD-L1 expression plus CXCL12 level to predict DCB or no durable benefit (NDB) showed AUC values of 0.680, 0.719 and 0.794, respectively. CONCLUSION: Our findings suggest that serum cytokine CXCL12 levels can predict the outcomes of patients with NSCLC receiving ICI. Moreover, the combination of CXCL12 levels and PD-L1 status can predict outcomes with a significantly improved discriminatory power.