Adaptive Enrichment - Workbook Tables

Operating Characteristics

**Table X - Simulated Operating Characteristics for Adaptive Enrichment Design**

Nmax = 300, interims at n = 100, 200. Model: BLRM logit(p) = alpha + beta*B1. Priors: N(0, 100). 1000 simulations per scenario. 

Scenario	True.ORR.(B1.<.50)	True.ORR.(B1.>=.50)	P(Enrich)	Median.B1.Cutoff	N.in.Enriched.Region	Go.Rate.(overall).[1]	Go.Rate.|.Enriched.[2]	Go.Rate.|.Not.Enriched.[3]	E[ORR].Final.Pop
Null - Inactive (flat ORR = 0.15)	15.0%	15.0%	20.3%	85.0	216	0.0%	0.0%	0.0%	15.1%
Null - Boundary (flat ORR = 0.25)	25.0%	25.0%	50.6%	65.0	175	34.7%	48.6%	20.4%	26.0%
Null - Inverted B1 effect (high ORR in low B1)	25.0%	10.0%	1.4%	80.0	216	0.0%	0.0%	0.0%	17.4%
Active - Global effect (flat ORR = 0.35)	35.0%	35.0%	18.8%	15.0	256	100.0%	100.0%	100.0%	35.3%
Active - Strong B1 effect (step at 50)	10.0%	35.0%	99.9%	65.0	201	99.6%	99.7%	0.0%	35.4%
Active - Moderate B1 effect (step at 50)	15.0%	30.0%	95.9%	75.0	162	86.5%	90.2%	0.0%	30.3%
Active - Weak B1 effect (shallow gradient)	15.8%	25.4%	90.9%	85.0	147	71.7%	78.9%	0.0%	28.9%

Notes

- Go/No-Go decision rule: At each stage, the lowest B1 cutoff c where P(ORR(c) >= 0.25 | data) >= stage threshold is selected. The cutoff can only increase across stages (ratchet). At the final analysis (Stage 3), 'Go' if P(mean ORR(B1 >= c) >= 0.25 | data) >= 80%.
- Enrichment ratchet: Thresholds 30%/50%/80% at Stages 1/2/3. Cutoff = 0 means all-comers is the active region. Cutoff > 0 means enrolment is restricted to B1 >= cutoff. The cutoff can increase at subsequent stages but never decrease.
- [1] Go Rate (overall) = proportion of simulated trials concluding 'Go'. For null scenarios, this is the Type I error rate.
- [2] Go Rate | Enriched = Go rate among trials where cutoff > 0 (population narrowed from all-comers).
- [3] Go Rate | Not Enriched = Go rate among trials where cutoff = 0 (all-comers identified as active region).

Simulation Settings

Parameter	Value
N_MAX	300
INTERIM_N	100, 200
N_SIMS	1000
TARGET_ORR	0.25
NULL_ORR	0.15
ENRICH_THRESHOLDS	0.3, 0.5, 0.8
PRIOR_MEAN	0, 0
PRIOR_VAR	100, 100
N_POST_SAMPLES	4000
B1_CUTOFF_GRID (range)	0 to 95 (by 5)
CUTOFF_STRATEGY	lowest
N_GRID_POINTS	200
SEED	42
Simulation Duration (min)	1.1

Enrichment Details

Scenario	N_Trials_Enriched	Pct_Enriched_Stage1	Pct_Enriched_Stage2	Cutoff_Q25	Cutoff_Median	Cutoff_Q75	Mean_N_Post_Enrich
Null - Inactive (flat ORR = 0.15)	203	99.5%	0.5%	75.0	85.0	90.0	216
Null - Boundary (flat ORR = 0.25)	506	99.4%	0.6%	40.0	65.0	85.0	175
Null - Inverted B1 effect (high ORR in low B1)	14	100.0%	0.0%	75.0	80.0	85.0	216
Active - Global effect (flat ORR = 0.35)	188	100.0%	0.0%	10.0	15.0	30.0	256
Active - Strong B1 effect (step at 50)	999	98.9%	0.9%	60.0	65.0	75.0	201
Active - Moderate B1 effect (step at 50)	959	97.6%	2.2%	65.0	75.0	85.0	162
Active - Weak B1 effect (shallow gradient)	909	96.5%	3.1%	75.0	85.0	90.0	147

Diagnostics

Scenario	Post_Alpha_Mean	Post_Alpha_SD	Post_Beta_Mean	Post_Beta_SD	Final_Prob_Mean	Final_Prob_SD	Obs_ORR_Mean	Obs_ORR_SD	N_Final_Pop_Mean	N_Final_Pop_SD
Null - Inactive (flat ORR = 0.15)	-1.761	0.181	-0.015	0.283	0.003	0.019	0.151	0.022	283	34.7
Null - Boundary (flat ORR = 0.25)	-1.141	0.174	0.053	0.272	0.590	0.302	0.260	0.034	237	77.2
Null - Inverted B1 effect (high ORR in low B1)	-1.640	0.158	-0.835	0.281	0.013	0.045	0.174	0.023	299	10.1
Active - Global effect (flat ORR = 0.35)	-0.628	0.134	0.007	0.230	0.997	0.012	0.353	0.028	292	21.5
Active - Strong B1 effect (step at 50)	-1.282	0.255	0.968	0.309	0.994	0.041	0.354	0.035	201	32.1
Active - Moderate B1 effect (step at 50)	-1.285	0.208	0.625	0.278	0.886	0.221	0.303	0.043	168	53.8
Active - Weak B1 effect (shallow gradient)	-1.453	0.215	0.668	0.305	0.784	0.302	0.289	0.049	161	65.6