Prediction performance comparison of biomarkers for response to immune checkpoint inhibitors in advanced non-small cell lung cancer

BackgroundThe aim of the present study was to compare the predictive accuracy of PD-L1 immunohistochemistry (IHC), tissue or blood tumor mutation burden (tTMB, bTMB), gene expression profile (GEP), driver gene mutation, and combined biomarkers for immunotherapy response of advanced non-small cell lung cancer (NSCLC).MethodsIn part 1, clinical trials involved with predictive biomarker exploration for immunotherapy in advanced NSCLC were included. The area under the curve (AUC) of the summary receiver operating characteristic (SROC), sensitivity, specificity, likelihood ratio and predictive value of the biomarkers were evaluated. In part 2, public datasets of immune checkpoint inhibitor (ICI)-treated NSCLC involved with biomarkers were curated (N = 871). Odds ratio (OR) of the positive versus negative biomarker group for objective response rate (ORR) was measured.ResultsIn part 1, the AUC of combined biomarkers (0.75) was higher than PD-L1 (0.64), tTMB (0.64), bTMB (0.68), GEP (0.67), and driver gene mutation (0.51). Combined biomarkers also had higher specificity, positive likelihood ratio and positive predictive value than single biomarkers. In part 2, the OR of combined biomarkers of PD-L1 plus TMB (PD-L1 cutoff 1%, 0.14; cutoff 50% 0.13) was lower than that of PD-L1 (cutoff 1%, 0.33; cutoff 50% 0.24), tTMB (0.28), bTMB (0.48), EGFR mutation (0.17) and KRAS mutation (0.47), for distinguishing ORR of patients after immunotherapy. Furthermore, positive PD-L1, tTMB-high, wild-type EGFR, and positive PD-L1 plus TMB were associated with prolonged progression-free survival (PFS).ConclusionCombined biomarkers have superior predictive accuracy than single biomarkers for immunotherapy response of NSCLC. Further investigation is warranted to select optimal biomarkers for various clinical settings. In current clinical trials for non-small cell lung cancer (NSCLC), the predictive roles of PD-L1, tissue or blood tumor mutation burden (tTMB, bTMB), gene expression profile (GEP), and driver gene mutations such as EGFR and KRAS have been explored. We are particularly interested in determining which biomarker exhibits higher predictive efficacy to facilitate its application in clinical practice and enhance healthcare cost-effectiveness. This study demonstrates that combined biomarkers are more efficient than single biomarkers for predicting the success of immunotherapy, and bTMB is a promising liquid biopsy biomarker. In addition, immunotherapy is effective for EGFR wild-type but KRAS mutant-type patients, which is warranted for further investigation across multiple clinical settings. We believe that our study provides valuable insights into the predictive efficacy of various biomarkers for immunotherapy response in NSCLC. image