Event-free survival (EFS) in patients with high-risk neuroblastoma (HRNB) receiving eflornithine (DFMO) maintenance treatment based on a retrospective, blinded independent central review (BICR) of imaging during long-term follow-up.

Abstract

e22007 Background: BICR methodology can evaluate investigator bias in open label studies. Here (BICR W2218), we address potential investigator bias in the open-label evaluation of EFS in a subset of NMTRC003B patients selected for comparison to patients from ANBL0032 (COG trial; NCT00026312) using imaging and clinical assessments. BICR results were then compared to the propensity-score (PS)-matched external control (NO DFMO) patients from ANBL0032. Methods: NMTRC003B DFMO-treated patients (≤2 years, 750 ±250 mg/m2 BID) on (n = 87) or per (n = 5) ANBL0032 and in remission post-dinutuximab completion were selected for inclusion in the BICR. Imaging evaluations (incl. anatomical [CT/MRI] and nuclear [MIBG and/or PET]) were determined at end of immunotherapy (baseline; to confirm remission); at 3, 6, 9, 12, 18, and 24 months; and, per Standard of Care after 24 months. While only required for the 2-year on-therapy period, imaging data were collected during follow up and also evaluated by BICR. All available imaging data (per patient at each visit; baseline via long-term follow up) from NMTRC003B patients were evaluated by two independent radiologists (IR) with adjudication by a third independent radiologist. BICR data from DFMO patients (N = 92) were used to evaluate BICR-EFS vs. EFS in DFMO-treated patients (N = 92) meeting selection criteria (demographic and risk characteristics) for PS-matched comparison (1:3 ratio for DFMO:NO-DFMO) to ANBL0032 (N = 270). ANBL0032 had enrolled N = 1328 HNRB patients assigned to dinutuximab and follow up (≤10 years)1 and required similar imaging evaluations that were unavailable for the BICR. Concordance was evaluated with Cohen’s Kappa coefficient analysis, and EFS using Kaplan-Meier and Cox regression analyses. Results: All 92 patients had adequate imaging at baseline and were included in the BICR. N = 21/92 (22.8%) cases were adjudicated with a final outcome of unequivocal EFS event in N = 15. Despite discordance in n = 2 patients, concordance was very high during the on-treatment (2 years) and full follow-up (≤7) periods with near perfect agreement. In line with previously reported data1, DFMO treatment was associated with significant improvement in relapse events in BICR-DFMO cases vs. PS-matched, external NO DFMO patients over the full BICR period [hazard ratio, HR 0.54 (95% CI: 0.31, 0.95), p = 0.0337] with 2- and 4-year BICR-EFS for DFMO vs. NO DFMO of 88% vs. 80%, and 84% vs. 73%, respectively. Conclusions: This BICR indicates minimal, if any, bias in imaging evaluations in the original NMTRC003B study, providing confidence in the finding that DFMO treatment in HNRB patients in remission post upfront dinutuximab approximately halves the risk of relapse compared to matched control patients.