Background and importance Multiple therapeutic alternatives are used in untreated metastatic squamous non-small cell lung cancer (umSNSCLC). Paclitaxel–carboplatin–pembrolizumab combination (PC-pembrolizumab) has recently been authorised for this indication.

Aim and objectives To assess the comparative efficacy among different therapeutic alternatives used in mSNSCLC through a network meta-analysis (NMA).

Material and methods A search was conducted on 19 February 2020 with the following inclusion criteria: phase II/III randomised clinical trials (RCT), including drugs used in umSNSCLC, and overall survival (OS) as the efficacy endpoint. Exclusion criteria: mSNSCLC population with EGFR or ALK mutations and RCTs without a comparator common to the evaluated alternatives. Pooled hazard ratios (HR) were calculated by Bayesian methods, through the combination of direct and indirect evidence by the NMA. Fixed and random effects were evaluated. Deviance information criteria (DIC) statistics were used to compare the models. The agreement of direct and indirect estimations was assessed by node splitting models to evaluate the consistency of NMA. Delta value, maximum acceptable difference as clinical criterion of non-inferiority, was set at 0.70 (and its inverse, 1.43), used to calculate the sample size in the PC-pembrolizumab trial.

Results Nine RCTs were selected. PC was the common comparator. The DIC value for the fixed effects model was more favourable. No statistically significant differences between direct and indirect evidence were found, and therefore NMA was consistent. The PC-pembrolizumab combination was considered as the reference (treatment with the greatest magnitude of effect). HR for OS were: 1.4 (95% CI 0.89 to 2.3) versus carboplatin–gemcitabine; 1.6 (1.2 to 2.1) versus PC; 1.5 (1.1 to 2.1) versus nab–PC-atezolizumab; 1.8 (1.3 to 2.5) versus PC-figitumumab; 1.4 (0.96 to 2.0) versus PC-motesanib; 1.3 (0.66 to 2.5) versus PC-necitumumab; 2.1 (0.86 to 5.0) versus PC-olaratumab; 2.9 (1.7 to 4.8) versus PC-sorafenib and 1.2 (0.82 to 1.7) versus pembrolizumab monotherapy. Carboplatin–gemcitabine, PC-motesanib, PC-necitumumab, PC-olaratumab and pembrolizumab did not present statistically significant differences compared with PC-pembrolizumab. Statistically significant benefit was observed for PC-pembrolizumab over PC, nab–PC-atezolizumab, PC-figitumumab and PC-sorafenib. According to the delta values, there could be clinically relevant differences among them.

Conclusion and relevance NMA showed no significant differences in OS between PC-pembrolizumab and carboplatin–gemcitabine, PC-motesanib, PC-necitumumab, PC-olaratumab and pembrolizumab in umSNSCLC, but there could be clinically relevant differences. PC, nab–PC-atezolizumab, PC-figitumumab and PC-sorafenib were inferior to PC-pembrolizumab, with possible clinically relevant differences.

Conflict of interest No conflict of interest