Table 1.

Estimated power for the logrank test from 10,000 simulated trials generated under the 3 component cure rate model. Total sample size (N) and number of events (D) for full information was calculated for 80% power when each of 3 alternative hypothesis models was assumed to be true. Estimated power was the percent of trials that reject the null hypothesis using the N and D from each alternative model

	Alternative hypothesis model
	3 component cure rate	Proportional hazards cure rate	Exponential model
Total sample size (N)	409	228	209
Total number of deaths (D)	354	196	198
Estimated power with 2 interim looks	78.7%	52.5%	45.0%
Estimated power without interim looks	80.1%	55.4%	46.5%
% of trials stopped early for futility	6.6%	22.3%	20.8%