Table 2.
Study design and methods used to evaluate the surrogacy of progression-free survival for overall survival.
| Meta-analysis on aggregated data (N = 65) | Meta-analysis on individual patient data (N = 18) | Trial (N = 8) | All (N = 91) | |
|---|---|---|---|---|
| Evaluation of correlation at patient level | NA | 14 (78%) | 8 (100%) | 22 (24%) |
| Coefficient of correlation | NA | 11 (61%) | 7 (88%) | 18 (20%) |
| Coefficient of determination R² | NA | 3 (17%) | 5 (63%) | 8 (9%) |
| Evaluation of correlation at trial level | 64 (98%) | 15 (83%) | NA | 79 (87%) |
| Evaluation on aggregated measuresa | 37 (57%) | 2 (11%) | NA | 39 (43%) |
| Coefficient of correlation | 31 (48%) | 0 (0%) | NA | 31 (34%) |
| Coefficient of determination R² | 14 (22%) | 1 (6%) | NA | 15 (16%) |
| Tau de Kendall | 0 (0%) | 1 (6%) | NA | 1 (1%) |
| Evaluation on treatment effect | 50 (77%) | 14 (78%) | NA | 64 (70%) |
| ∙ Hazard ratio | 36 (55%) | 14 (78%) | NA | 50 (55%) |
| ∙ Difference of median | 13 (20%) | 0 (0%) | NA | 13 (14%) |
| ∙ Ratio of median | 1 (2%) | 0 (0%) | NA | 1 (1%) |
| Coefficient of correlation | 29 (45%) | 1 (6%) | NA | 30 (33%) |
| Coefficient of determination R² | 36 (55%) | 13 (72%) | NA | 49 (54%) |
| Surrogate threshold effect | 11 (17%) | 8 (44%) | NA | 19 (21%) |
NA not applicable.
aMedian progression-free survival, the progression-free survival rate at a clinically meaningful timepoint.