Table 6.

Statistical power for testing the treatment effect under the null and alternative hypotheses for erlotinib resistant and naïve patients by marker status in the BATTLE-2 adaptive randomization design using the Bayesian logistic regression. Cut-off values for posterior probability were chosen to control the type I error rates for comparing TX 2, 3, 4 to TX 1 under the null hypothesis to 10%.

Prior	Scenario	Cut-off	TX	Margin	Naïve. Margin	Resist. Margin	Naïve. MK(−,−)	Naïve. MK(+,−)	Naïve. MK(−,+)	Resist. MK(−,−)	Resist. MK(+,−)	Resist. MK(−,+)	Overall	All TX
NI	Null	0.982	2	0.004	0.040	0.000	0.035	0.046	0.014	0.000	0.000	0.000	0.098	0.184
			3	0.004	0.034	0.000	0.034	0.052	0.011	0.001	0.001	0.001	0.093
			4	0.004	0.018	0.000	0.033	0.010	0.002	0.000	0.000	0.000	0.044
	Alt.		2	0.302	0.463	0.003	0.348	0.293	0.083	0.004	0.004	0.000	0.632	0.865
			3	0.316	0.442	0.001	0.358	0.263	0.098	0.003	0.009	0.000	0.592
			4	0.375	0.440	0.002	0.348	0.102	0.244	0.003	0.004	0.011	0.622
Beta5	Null	0.984	2	0.003	0.038	0.000	0.033	0.041	0.013	0.000	0.001	0.000	0.091	0.174
			3	0.003	0.033	0.010	0.030	0.049	0.009	0.000	0.001	0.001	0.095
			4	0.004	0.016	0.000	0.030	0.009	0.001	0.000	0.000	0.001	0.039
	Alt.		2	0.291	0.447	0.277	0.336	0.277	0.071	0.271	0.211	0.039	0.787	0.987
			3	0.299	0.432	0.579	0.338	0.243	0.087	0.299	0.230	0.050	0.861
			4	0.357	0.423	0.040	0.337	0.096	0.225	0.311	0.208	0.207	0.801