Table 2.
Meta-analysis (univariate and multivariate) using random effects model according to trial characteristics (including personalized vs non-personalized drugs) in randomized registration trials
| Characteristic | Relative response rate ratio | PFS | OS | |||
|---|---|---|---|---|---|---|
| N | RRR (95% CI) | N | HR (95% CI) | N | HR (95% CI) | |
| Personalized status | ||||||
| Personalized | 14 | 3.82 (2.51 to 5.82) | 13 | 0.41 (0.33 to 0.51) | 13 | 0.71 (0.61 to 0.83) |
| Non-personalized | 37 | 2.08 (1.76 to 2.47) | 38 | 0.59 (0.53 to 0.65) | 33 | 0.81 (0.77 to 0.85) |
| P (univariate) | .009 | .004 | .11 | |||
| P (meta-regression)*,† | .03 | <.001 | .07 | |||
| Control arm | ||||||
| Placebo | 16 | 7.03 (4.27 to 11.58) | 15 | 0.42 (0.37 to 0.48) | 15 | 0.81 (0.74 to 0.90) |
| Active treatment | 35 | 2.08 (1.75 to 2.47) | 36 | 0.60 (0.54 to 0.67) | 31 | 0.78 (0.74 to 0.83) |
| P (univariate) † | <.001 | <.001 | .51 | |||
| P (meta-regression) *,† | <.001 | <.001 | NA | |||
| Tumor type | ||||||
| Solid | 43 | 2.27 (1.92 to 2.69) | 44 | 0.56 (0.50 to 0.62) | 41 | 0.81 (0.77 to 0.85) |
| Hematologic | 8 | 3.30 (1.93 to 5.63) | 7 | 0.42 (0.33 to 0.54) | 5 | 0.60 (0.49 to 0.73) |
| P (univariate) † | .2 | .04 | .004 | |||
| P (meta-regression) *,† | NA | .004 | .006 | |||
| Treatment status | ||||||
| Naïve | 22 | 2.23 (1.76 to 2.83) | 23 | 0.54 (0.46 to 0.63) | 20 | 0.77 (0.71 to 0.84) |
| Pretreated | 29 | 2.70 (2.09 to 3.50) | 28 | 0.54 (0.47 to 0.61) | 26 | 0.81 (0.76 to 0.87) |
| P (univariate) † | .29 | .99 | .37 | |||
| Cross-over allowed | ||||||
| Yes | 23 | 3.67 (2.86 to 4.69) | 22 | 0.42 (0.37 to 0.47) | 21 | 0.76 (0.68 to 0.85) |
| No | 28 | 1.68 (1.45 to 1.95) | 29 | 0.65 (0.58 to 0.72) | 25 | 0.80 (0.76 to 0.84) |
| P (univariate) † | <.001 | <.001 | .48 | |||
| P (meta-regression)*,† | NA | NA | .32 | |||
* Meta-regression analysis was performed including only factors considered statistically significant in univariate analysis. In the case of allowance for crossover for RRR and PFS analysis, meta-regression was not performed because crossover occurred after progression and could therefore not confound responses or PFS. HR = hazard ratio; N = number of trials included; NA = not applicable; OS = overall survival; PFS = progression-free survival; RRR = relative response rate ratio.
† P values were two-sided.