Table 2.
Actual and alternative biomarker-stratified trial design of lapatinib trial (5)
| Patient population | Actual designa |
Biomarker-stratified designb |
||||
|---|---|---|---|---|---|---|
| Actual sample size | Target HR (power, %) | Formally tested | Target sample size | Target HR (power, %) | Formally tested | |
| Biomarker-positive: HER2-positive | 218 | 0.645 (80) | Yes | 218 | 0.645 (80) | Yes |
| Biomarker-negative: HER2-negative | — | — | No | 430 | 0.700 (90) | Yes |
| Overall | 1280 | 0.769 (90) | Yes | 648 | — | No |
Sequential biomarker-positive/overall design; sequential testing (α = 0.025): first test the biomarker-positive subgroup at the statistical significance threshold level α (with the treatment declared ineffective if the test is not statistically significant). If the biomarker-positive subgroup test is statistically significant then test the overall population at the same statistical significance level α; the treatment is recommended for both biomarker-positive and biomarker-negative subgroups if the overall test is statistically significant and only for the biomarker-positive subgroup otherwise. (This procedure controls the overall type I error of the design at level α). CPS = PD-L1 combined positive score; HR = hazard ratio.
Sequential biomarker-stratified design; sequential testing (α = 0.025): first test the biomarker-positive subgroup at the statistical significance threshold level α (with the treatment declared ineffective if the test is not statistically significant). If the biomarker-positive subgroup test is statistically significant, then test the biomarker-negative subgroup at the same statistical significance level α; the treatment is recommended for both biomarker-positive and biomarker-negative subgroups if the biomarker-negative test is statistically significant and only for the biomarker-positive subgroup otherwise. (This procedure controls the overall type I error of the design at level α).