*3.4. Assumption Checking*

Figures S2 and S3 show a heat map, which provides visual inconsistency between direct and indirect comparisons in the frequentist approach. There was a big difference between inconsistency before and after the detachment in some treatment comparisons. However, no inconsistency was observed in the Bayesian approach (Figures S4 and S5).

Substantial heterogeneity was detected in both ORR and PFS, with the global I2 = 78% for both outcomes as well as for either a pairwise pooled effect or a consistency effect (Table S2).

The width of every line reflects the number of studies. The size of the circles is proportional to the number of study participants. A dummy group is a placebo or a control group without additional treatment.

### *3.5. Comparative E*ffi*cacy*

Compared to chemotherapy, afatinib, alectinib, ceritinib, and crizotinib were found to have a higher ORR with RRs ranging between 2.26 (95% CI, 1.34–3.82) for crizotinib and 3.75 (95% CI, 1.80–7.94) for ceritinib (Figure 2). Also, cabozantinib, gefitinib, and osimertinib vs. chemotherapy were found to improve PFS with HRs ranging from 0.17 (95% CI, 0.10–0.29) for alectinib to 0.78 (0.67–0.91) for gefitinib (Figure 3).

**Figure 2.** Response ratio for overall response rate of each targeted therapy vs. chemotherapy.

**Figure 3.** Hazard ratio for progression-free survival of each targeted therapy vs. chemotherapy.

Tables 1 and 2 show the league tables representing the comparative efficacy of targeted therapies for ORR and PFS in the network meta-analysis based on the Bayesian approach.

Among EGFR inhibitors, ORR was found to be significantly higher in afatinib treatment, compared to cetuximab (RR, 2.46; 95% CI, 1.25–4.90), erlotinib (RR, 2.64; 95% CI, 1.54–4.58), and gefitinib (RR, 2.08; 95% CI, 1.18-3.68) (Table 1). Also, afatinib had a significantly longer PFS, compared to cetuximab (HR, 0.49; 95% CI, 0.33–0.71), erlotinib (HR, 0.59; 95% CI 0.44–0.80), and gefitinib (HR, 0.69; 95% CI, 0.50–0.95) (Table 2). Osimertinib was found to improve PFS, compared to cetuximab (HR, 0.27; 95% CI, 0.14–0.55), erlotinib (HR, 0.33; 95% CI, 0.17–0.64), and gefitinib (HR, 0.39; 95% CI, 0.20–0.75) (Table 2). Gefitinib showed a better PFS compared to cetuximab (HR, 0.70; 95% CI, 0.53–0.94) (Table 2).

Regarding ALK/ROS1/MET targeted drugs, there were no significant differences in ORR between each pair of crizotinib, ceritinib, and alectinib (Table 1). However, alectinib showed a superior efficacy compared to either crizotinib (HR, 0.40; 95% CI, 0.25–0.64) or ceritinib (HR, 0.33; 0.17–0.67) for PFS (Table 2).

As for VEGF pathway (bevacizumab and ramucizumab) and RET targeted therapy (cabozantinib and vandetanib), only cabozantinib was found to improve PFS compared to vandetanib (HR, 0.36; 95% CI, 0.20–0.66) (Table 2).
