Adaptive Phase II/III Trial — Anti-Fibrotic Agent in NASH, Riyadh
Simulated Case Study
This case study uses simulated data to demonstrate the statistical methodology, analysis workflow, and reporting standards we apply to real client projects. No actual patient or institutional data is represented.
Bayesian Adaptive Design · SSR · Proportional Odds · DSMB
Confidentiality Notice
All data presented here is simulated and replicates the statistical structure of the original trial datasets. The sponsor name is anonymized. No identifiable patient data, biopsy results, or proprietary trial information is disclosed.
Trial Phase
Seamless Phase II/III Adaptive
Indication
NASH / NAFLD (non-alcoholic)
Sites
3 centers, Riyadh
Max sample size
240 (adaptive)
Primary endpoint
NAS score improvement ≥ 2
Design feature
Bayesian SSR at 30% info fraction
Project Overview
A Saudi-based sponsor engaged Naggar Analytics to provide end-to-end statistical support for a seamless adaptive Phase II/III trial evaluating an anti-fibrotic agent versus placebo in adult NAFLD/NASH patients with fibrosis stage F1–F3 across 3 hepatology centers in Riyadh. The adaptive design allowed a blinded sample size re-estimation at 30% information fraction based on the observed treatment effect variance.
Bayesian predictive probability modeling (using RBesT) defined Go/No-Go boundaries for the interim look, calibrated through 10,000 simulated trials in Cytel EAST 6. The primary endpoint — NAS score improvement ≥ 2 points at week 48 — was analyzed using a proportional odds model. All deliverables complied with ICH E9(R1) estimand framework and CDISC SDTM/ADaM data standards.
Statistical Deliverables
- ▹Adaptive design simulation — 10,000 iterations (Cytel EAST 6) calibrating Go/No-Go decision boundaries
- ▹Bayesian predictive probability — posterior beta-binomial model for interim success prediction
- ▹Blinded sample size re-estimation (SSR) at 30% information fraction; max n = 240
- ▹Randomization: 1:1:1 (two doses vs placebo), stratified by fibrosis stage (F1–F2 vs F3)
- ▹Primary endpoint analysis — proportional odds model for NAS improvement ≥ 2 points
- ▹Histological adjudication — dual-reader agreement (κ), consensus protocol for biopsy endpoints
- ▹CDISC SDTM/ADaM datasets — biopsy-derived LB and MB domains
- ▹DSMB charter — stopping rules, cumulative AE review at each interim look
- ▹Sensitivity analysis — responder analysis, tipping point, pattern-mixture model
- ▹Regulatory package — ICH E9(R1) estimand framework, final TFLs to ICH E3 structure