- Design
- Frequentist network meta-analysis of phase 3 randomised trials with P-score ranking, sensitivity and carcinoma in situ-stratified analyses
- Population
- Three trials (ALBAN, CREST, POTOMAC) of a checkpoint inhibitor plus BCG versus BCG alone in BCG-naive high-risk non-muscle-invasive bladder cancer
- Primary outcome
- Event-based disease control, pooling event-free and disease-free survival, with BCG as reference
- Effect
- Durvalumab HR 0.68 (95% CI 0.50-0.93), sasanlimab 0.68 (0.49-0.94), atezolizumab 0.98 (0.71-1.36); grade 3 or worse treatment-related events 21-29% vs 3.8-8.8% with BCG alone
Three phase 3 trials have tested a checkpoint inhibitor added to BCG in BCG-naive high-risk non-muscle-invasive bladder cancer, and they did not agree. This network meta-analysis puts them side by side: ALBAN (atezolizumab), CREST (subcutaneous sasanlimab) and POTOMAC (durvalumab), with BCG alone as the reference and event-free or disease-free survival as a joint endpoint.
Durvalumab plus BCG (HR 0.68, 95% CI 0.50-0.93) and sasanlimab plus BCG (0.68, 0.49-0.94) both significantly improved event-based disease control. Atezolizumab plus BCG did not (0.98, 0.71-1.36) - essentially identical to BCG alone, and its P-score of 0.22 sat next to BCG's 0.16 while the other two ranked at 0.81.
The toxicity price is substantial and should be quoted precisely: grade 3 or worse treatment-related adverse events rose from 3.8-8.8% with BCG alone to 21-29% with the combinations. CREST and POTOMAC reported no clinically meaningful decline in quality of life, which is reassuring but is a different measure from serious toxicity.
Carcinoma in situ stratification produced different rankings - durvalumab highest without it, sasanlimab highest with it - and the authors label this hypothesis-generating, which it is: three trials cannot support subgroup rankings.
What this changes is how to read the class. PD-1 and PD-L1 inhibitors added to BCG are not interchangeable here, and a negative trial for one agent does not condemn the approach any more than a positive one endorses the class. For Indian practice, where BCG supply itself is the recurring problem, the relevant question is still whether adequate BCG is being delivered before any intensification is discussed.
- Do not treat these agents as interchangeable - atezolizumab did not separate from BCG alone
- Quote the toxicity: grade 3 or worse events roughly tripled to quadrupled with combination therapy
- Ensure adequate BCG induction and maintenance is actually deliverable before discussing intensification
- Treat the carcinoma in situ subgroup rankings as hypothesis-generating, not as a selection rule
- Event-free and disease-free survival were pooled as one endpoint; neither is overall survival
The statistics, in plain English
Two hazard ratios of 0.68 with intervals just excluding 1.0 (0.50-0.93 and 0.49-0.94) are real but not dramatic effects, and both rest on single trials rather than replicated evidence. Atezolizumab's 0.98 with an interval of 0.71-1.36 does not show the drug is useless - the interval still includes a 29% reduction - but it provides no evidence of benefit. P-scores rank treatments on the probability of being best and are easily over-read: a P-score of 0.81 in a three-treatment network is not a strong statement. Pooling event-free and disease-free survival because their definitions were comparable is a reasonable choice that nonetheless adds heterogeneity.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for urology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free