Table 1. Description of simulation scenarios and simulation results for the eight scenarios.
| Design | Recommend any | Recommend T1 in negative patients | Recommend T1 in B1-positive patients | Recommend T1 in B2-positive patients | Recommend T2 in B1-positive patients |
|---|---|---|---|---|---|
|
Scenario 1: all experimental treatments have the same effect as control in every biomarker subgroup | |||||
| NLB | 0.418 | 0.056 | 0.051 | 0.056 | 0.050 |
| LB | 0.409 | 0.048 | 0.074 | 0.040 | 0.040 |
| PT | 0.498 | 0.069 | 0.069 | 0.071 | 0.065 |
| ER | 0.473 | 0.067 | 0.068 | 0.069 | 0.069 |
|
Scenario 2: as scenario 1, except T1 doubles the probability of response from 30% to 60% in patients who are positive for B1 | |||||
| NLB | 0.856 | 0.051 | 0.767 | 0.058 | 0.057 |
| LB | 0.885 | 0.048 | 0.814 | 0.045 | 0.037 |
| PT | 0.903 | 0.070 | 0.817 | 0.057 | 0.068 |
| ER | 0.800 | 0.066 | 0.665 | 0.065 | 0.067 |
|
Scenario 3: as scenario 1, except T1 doubles the probability of response in patients who are positive for B2 | |||||
| NLB | 0.829 | 0.056 | 0.059 | 0.740 | 0.058 |
| LB | 0.790 | 0.054 | 0.062 | 0.676 | 0.038 |
| PT | 0.672 | 0.066 | 0.070 | 0.407 | 0.073 |
| ER | 0.802 | 0.066 | 0.067 | 0.666 | 0.067 |
|
Scenario 4: T1 provides a moderate benefit in all biomarker groups | |||||
| NLB | 0.916 | 0.606 | 0.403 | 0.430 | 0.057 |
| LB | 0.912 | 0.600 | 0.503 | 0.356 | 0.032 |
| PT | 0.877 | 0.518 | 0.474 | 0.243 | 0.077 |
| ER | 0.869 | 0.526 | 0.367 | 0.362 | 0.070 |
|
Scenario 5: T1 provides a detrimental effect in patients positive for B1 with response rate falling from 0.3 to 0.11 | |||||
| NLB | 0.428 | 0.061 | 0.000 | 0.059 | 0.066 |
| LB | 0.367 | 0.047 | 0.000 | 0.033 | 0.048 |
| PT | 0.475 | 0.066 | 0.000 | 0.056 | 0.068 |
| ER | 0.440 | 0.070 | 0.001 | 0.064 | 0.065 |
|
Scenario 6: as scenario 5, except T1 provides a detrimental effect in patients positive for B2 instead of B1 | |||||
| NLB | 0.404 | 0.047 | 0.050 | 0.001 | 0.053 |
| LB | 0.406 | 0.051 | 0.073 | 0.000 | 0.044 |
| PT | 0.491 | 0.064 | 0.072 | 0.004 | 0.065 |
| ER | 0.450 | 0.070 | 0.070 | 0.001 | 0.069 |
|
Scenario 7: T1 provides large benefit in patients positive for B1 and a detrimental effect in patients positive for B2 | |||||
| NLB | 0.817 | 0.049 | 0.725 | 0.000 | 0.056 |
| LB | 0.878 | 0.057 | 0.806 | 0.000 | 0.033 |
| PT | 0.889 | 0.068 | 0.805 | 0.002 | 0.069 |
| ER | 0.787 | 0.070 | 0.657 | 0.002 | 0.069 |
|
Scenario 8: T1 provides detrimental effect in patients positive for B1 and T2 provides a beneficial effect in those patients | |||||
| NLB | 0.848 | 0.056 | 0.002 | 0.046 | 0.774 |
| LB | 0.820 | 0.044 | 0.002 | 0.028 | 0.724 |
| PT | 0.662 | 0.071 | 0.001 | 0.063 | 0.412 |
| ER | 0.797 | 0.074 | 0.002 | 0.069 | 0.663 |
PT=parallel trials design; NLB=non-linked BAR design; LB=linked-BAR design; ER=equal-randomisation design.
Second to sixth columns give recommendation probabilities.