Table 4.
Part 2 rwRR analysis results in specific clinical settings corresponding with registrational trials
| Benchmark trial | Clinical setting/treatment | rwRR, % (95% CI) | |||
|---|---|---|---|---|---|
| N | Unweighteda | Weightedb | Weighted, confirmedc | ||
| KEYNOTE-024 | 1L IO, PD-L1 + / pembrolizumab | 72 | 51.4 (40.1, 62.6) | 54.0 (41.9, 65.7) | 42.5 (31.1, 54.8) |
| ALEX | 1L targeted, ALK rearrang/ alectinib | 60 | 81.7 (70.1, 89.4) | 81.2 (69.1, 89.3) | 67.6 (54.5, 78.4) |
| 1L targeted, ALK rearrang/ crizotinib | 145 | 64.1 (56.1, 71.5) | 65.8 (51.5, 77.7) | 48.1 (34.5, 62.0) | |
| KEYNOTE-021 | 1L IO + chemo/ pembrolizumab + carboplatin + pemetrexed | 121 | 66.1 (57.3, 73.9) | 68.8 (59.0, 77.1) | 43.7 (34.3, 53.5) |
| 1L chemo/ carboplatin + pemetrexed | 83 | 63.9 (53.1, 73.4) | 63.3 (51.8, 73.5) | 37.6 (27.3, 49.2) | |
| CheckMate 057 | 2L + IO, non-squam/ nivolumab | 83 | 27.7 (19.2, 38.2) | 27.6 (18.4, 39.3) | 17.6 (10.4, 28.4) |
| 2L + chemo, non-squam/ docetaxel | 97 | 23.7 (16.4, 33.1) | 25.8 (17.4, 36.5) | 10.9 (5.7, 19.8) | |
| CheckMate 017 | 2L + IO, squam/ nivolumab | 86 | 44.2 (34.2, 54.7) | 45.7 (35.1, 56.7) | 29.5 (20.6, 40.4) |
| 2L + chemo, squam/ docetaxel | 53 | 24.5 (14.9, 37.6) | 18.2 (9.7, 31.5) | 9.8 (4.1, 21.6) | |
| OAK | 2L + IO/ atezolizumab | 58 | 19.0 (10.9, 30.9) | 25.3 (15.1, 39.1) | 11.1 (4.9, 23.0) |
| 2L + chemo/ docetaxel | 117 | 22.2 (15.6, 30.6) | 21.8 (15.1, 30.4) | 12.3 (7.4, 19.8) | |
| AURA-3 | 2L + targeted, EGFRmt/ osimertinib | 97 | 56.7 (46.8, 66.1) | 62.2 (47.7, 74.8) | 52.1 (38.0, 65.9) |
1L first line, 2L second line, ALK anaplastic lymphoma kinase, CI confidence interval, EGFR epidermal growth factor receptor, IO immuno-oncology, PD-L1 programmed cell death ligand 1, rwRR real-world response rate
aUnweighted rwRR refers to analyses based on cohorts generated by only applying corresponding trial criteria, as feasible
bWeighted rwRR refers to the analyses performed on the real-world cohorts after inverse odds weights (based on published summary baseline characteristics of the trial populations) were applied in order to maximize the relevance of these cohorts to the intended comparison. Those were the cohorts used for the “weighted” analysis
cWeighted, confirmed rwRR refers to the analysis based on “confirmed responses” (as opposed to all responses) observed in the cohorts after weights were used