A review of studies examining the relationship between progression ...

Table 3: Subgroups consideredPaper Factors analysedHackshaw et Before/after 1990 when second line therapies not commonly usedal 24 Death included in surrogate time-to-event outcome (i.e PFS not TTP)Sherrill et al 11 Treatment class (hormonal, anthracyclines, first-line, non first-line) Only HER2+ patients Study size (>100 per arm) TTP >6 mths in control arm Reported HRsMiksad et al 23 Strict PFS definition, Year last patient recruited First / subsequent line treatmentHotta et al 26 Year of trial Old agents used Cisplatin used Carboplation used Full publication or abstract Description of sample size calculation Definition of primary endpoint Description of TTP definition Description of OS definition Description of definition for both TTP and OS Sample size5.4. RESULTS5.4.1 Regression parametersIn the studies using aggregate data from multiple trials, the R 2 from linear regression varied from0.19 to 0.56 for the treatment effect in TTP and OS, and 0.35 to 0.65 for the treatment effect in PFSand OS. For studies defining the treatment effect in terms of absolute change in survival betweentreatments arms, the R 2 varied from 0.19 to 0.65. For studies defining the treatment effect using thehazard ratio, the R 2 varied between 0.30 and 0.59. When considering the R 2 for treatment effect bytumour type, the values were fairly consistent for three studies in breast cancer (0.30 to 0.56), andsimilarly for three studies in colorectal cancer (0.32 to 0.65).The two highest R 2 values were reported by the studies using IPD. Foster et al 17 reported an R 2 of0.79 between PFS and OS when using log HRs as the treatment outcome in extensive small-cell lungcancer. Buyse et al 14 reported an R value much higher than the values reported in other studies (R =0.99), but they gave a 95% CI spanning unity, which suggests that this estimate should be treatedwith caution.27