Evaluation of a multi-arm multi-stage Bayesian design for phase II drug selection trials – an example in hemato-oncology

Table 1 MDS results – Sensitivity analyses

Prior	ESS			Posterior mean			Decision criteria
Arm	A	B	C	A	B	C	A	B	C	B	C	B	C
							Rule 1	Rule 1	Rule 1	Rule 2	Rule 2	Rule 3	Rule 3
MLE	0	0	0	0.3750	0.3250	0.4000
Non informative	1	1	1	0.3780	0.3293	0.4024	0.1505	0.3576	0.0863	0.3198	0.5906	0.0286	0.1197
Non informative	2	2	2	0.3810	0.3333	0.4048	0.1384	0.3346	0.0789	0.3223	0.5894	0.0281	0.1161
Sk eptical	10	1	1	0.3600	0.3244	0.3976	0.1900	0.3833	0.0971	0.3575	0.6437	0.0310	0.1340
	10	5	1	0.3600	0.3222	0.3976	0.1900	0.3885	0.0971	0.3465	0.6437	0.0262	0.1340
	10	1	5	0.3600	0.3244	0.3889	0.1900	0.3833	0.1074	0.3575	0.6148	0.0310	0.1099
Enthusiastic	10	1	1	0.3600	0.3222	0.3889	0.1900	0.3885	0.1074	0.3465	0.6148	0.0262	0.1099
	10	5	1	0.3600	0.3280	0.4012	0.1900	0.3640	0.0889	0.3716	0.6570	0.0338	0.1422
	10	1	5	0.3600	0.3389	0.4012	0.1900	0.2996	0.0889	0.4128	0.6570	0.0393	0.1422

The first line refers to the maximum likelihood estimate of response probability of each treatment arm, while the other lines refer to Bayes posterior estimates with computed decision criteria based on different combinations of the priors
MLE maximum likelihood estimate, ESS effective sample size, Decision criteria use p₀=0.3, Δ=0, δ*=0.15: p₀ refers to the minimum required treatment response rate of the first rule (Eq. 1); Δ to the targeted difference of the second rule (Eq. 2), and δ* to the sufficient treatment response rate of the third rule (Eq. 3)

ISSN: 1471-2288