Predictive value of oncogenic driver subtype, programmed death-1 ligand (PD-L1) score, and smoking status on the efficacy of PD-1/PD-L1 inhibitors in patients with oncogene-driven non-small cell lung cancer

Background This multicenter, retrospective study explored the value of oncogene driver subtype, programmed death-1 ligand (PD-L1) status, and smoking status for predicting which patients with oncogene-driven non-small cell lung cancer (NSCLC) would benefit from treatment with programmed death-1 (PD-1)/PD-L1 inhibitors. Methods The clinical features, PD-L1 tumor proportion scores, and PD-1/PD-L1 inhibitor (PDi) outcomes (objective response rate and progression-free survival) of patients who had advanced NSCLC with Kirsten rat sarcoma 2 viral oncogene homolog (KRAS) mutations or common, actionable oncogenic drivers were captured. Results In total, 189 oncogene-positive patients were analyzed. Of these, 104 received a PDi, and 108 had undergone prior PD-L1 testing. The frequency of PD-L1 positivity (>= 1%) was higher in patients who had KRAS mutations (P = .031), smokers (P = .006), and non-Asian patients (P = .002). Multivariable analysis indicated that smoking status (P < .001) was the only factor associated significantly with KRAS mutation. The objective response rate to PDi treatment was 16.9% (11 of 65 patients) among smokers (17.3% in the KRAS-mutant and 15.4% in the non-KRAS-mutant smoker subgroups), which was significantly higher than the 0% rate (0 of 26 patients; P = .019) among never-smokers. In subgroup analyses, progression-free survival was influenced by KRAS mutation status (median, 4.57 vs 1.63 months; P = .004), smoking status (4.07 vs 1.73 months; P = .004), PD-L1 positivity (3.8 vs 1.2 months; P = .040), and non-Asian race (3.0 vs 1.97 months; P = .046). In multivariable analysis, only smoking status (P = .008) remained a significant predictor when a PD-L1 level >= 1% was used. However, both smoking status (P = .001) and PD-L1 status (P = .028) were independent predictors when a PD-L1 level >= 50% was used. Conclusions Among associated clinical features among patients who have NSCLC with oncogenic drivers, smoking status potentially was the most important, easily available predictor of single PDi efficacy.