*4.3. Strengths and Limitations*

To the best of our knowledge, this is the first network meta-analysis which summarized the direct and indirect evidence on the comparative efficacy of targeted therapies in the treatment of NSCLC. Also, this compiled a large dataset, and the method was valid by checking several assumptions. In addition, this network meta-analysis included clinical trials only, which had a higher level of evidence than observational studies and allowed us to obtain the precise estimates.

Despite the strengths, there are several limitations in the current study. The efficacy of targeted therapies was evaluated through ORR and PFS surrogates only. We did not perform subgroup analyses by different treatment lines and patients of different mutations as well. Also, the potential heterogeneity was observed with approximately 78% for both ORR and PFS outcomes. Finally, among 34,969 subjects included for the analysis of ORR outcome, the small number of patients received cabozatinib (38 subjects, Table S1). Also, a big difference in ORRs between the two arms (10.5% for cabozatinib vs. 2.6% erlotinib) might lead to the large error margins for the comparative effect of cabozantinib and other treatments (Figure 2 and Table 1).
