How Polivy Trials Controlled for Variables in Efficacy Assessments
Polivy (polatuzumab vedotin-piiq), an antibody-drug conjugate for relapsed/refractory diffuse large B-cell lymphoma (DLBCL), was evaluated in the pivotal phase 2 POLARIX trial (NCT03274492). Efficacy endpoints like progression-free survival (PFS), overall survival (OS), and objective response rate (ORR) relied on standard oncology trial designs to minimize bias and isolate treatment effects.[1][2]
Core Design: Randomized, Double-Blind Comparison
Patients were randomized 1:1 to Polivy plus bendamustine-rituximab (Pola-BR) or bendamustine-rituximab (BR) alone (n=440 total). Double-blinding ensured neither patients nor investigators knew assignments, reducing performance and detection bias in PFS assessments via independent review committee (IRC) using Lugano criteria.[1][3] Stratification by baseline factors—International Prognostic Index (IPI) risk (0-2 vs 3-5), therapy timing from relapse (≤6 vs >6 months), and region—balanced prognostic imbalances across arms.[2]
Key Confounding Variables Controlled
- Patient Heterogeneity: Entry criteria limited to transplant-ineligible DLBCL patients post-first relapse, excluding those with prior CAR-T or high-grade lymphomas. This narrowed variability in disease biology.[1]
- Prognostic Scores: IPI stratification adjusted for age, LDH levels, ECOG status, extranodal sites, and stage, which predict outcomes independently of treatment.[3]
- Prior Therapy Exposure: Required ≥1 prior line but no refractory disease to BR, standardizing resistance profiles.[2]
- Crossover and Dropouts: No crossover allowed; intention-to-treat (ITT) analysis with censoring for dropouts preserved randomization benefits. Sensitivity analyses (per-protocol, censoring at subsequent therapy) confirmed PFS consistency.[1][3]
Statistical Adjustments and Sensitivity Analyses
Cox proportional hazards models for PFS adjusted for stratification factors, yielding HR 0.73 (95% CI 0.57-0.95; p=0.018).[1] Event-driven design (80% PFS events) powered detection of 27% reduction. Competing risks (e.g., death without progression) used Fine-Gray models; landmark analyses at 12/24 months aligned with unadjusted Kaplan-Meier curves.[3]
Why This Matters for Efficacy Claims
Pola-BR showed 40% PFS risk reduction at median 28 months, but OS was immature/non-significant, prompting scrutiny on early censoring and post-progression therapies (more in BR arm).[2][4] FDA approval (2021) hinged on blinded IRC PFS as primary endpoint, with real-world data later validating controls via multivariate analyses.[4]
Sources
[1]: Sehn et al., Lancet Oncology (2022) - link
[2]: Polivy prescribing information - link
[3]: FDA review summary - link
[4]: EMA assessment report - link