Bone marrow next generation sequencing minimal residual disease (NGS-MRD) positivity pre-allogeneic hematopoietic cell transplantation (HCT) predicts outcomes for patients with B-cell acute lymphoblastic leukemia (B-ALL): A planned analysis from the ptctc ONC1401 trial

Abstract

Abstract HCT is a well-established standard therapy for children and young adults (CYA) with very high-risk/relapsed B-ALL. Recent studies have shown pre-HCT positive MRD by multichannel flow cytometry (MFC) leads to poorer outcomes, namely lower relapse-free and overall survival, with many centers delaying HCT if any detectable MRD by MFC is present. NGS has emerged as a methodology to detect MRD 2-3 logs lower than MFC (as low as 10-7 vs. 10-4). A subset analysis of the Children's Oncology Group ASCT0431 trial showed improved predictive power of NGS for relapse, event-free survival (EFS), and overall survival (OS) compared to MFC pre- and post-HCT (Pulsipher, 2015). Based upon these findings, we performed a prospective correlative biology study as part of the NCI-funded PTCTC ONC1401 trial. Samples from 9 US centers with patients transplanted between 2015 to 2021 were obtained for BM and PB NGS testing pre- and post-HCT at pre-specified time points. We hypothesized that BM NGS pre- and post-HCT would be highly predictive of relapse, relapse-free survival (RFS), and non-relapse mortality (NRM) in patients undergoing HCT for B-ALL and would be superior to BM MFC. A secondary hypothesis was that PB NGS would outperform BM MFC and may be a reasonable post-HCT relapse monitoring approach. Our cohort consisted of 121 CYA with B-ALL and a median age of 12 (range <1-27.0) years; 29.8%, 41.3%, and 14.9% were CR1, CR2, CR3+, respectively. Over half (59.5%) were male and 46.3% identified as Hispanic/Latino. Few (8.3%) received matched sibling BM, with the majority (50.4%) receiving haploidentical grafts. NGS MRD values were centrally performed using Clonoseqassay (NGS study results were not available to clinical teams). Pre-HCT BM and PB NGS results were available for 61.2% and 71.9% of patients, respectively, and multiple imputations were completed for missing values. Only 9.1% of patients had positive BM MFC MRD pre-HCT while 27.3% were positive by NGS BM MRD, showing higher sensitivity of NGS. Univariable and multivariable Cox proportional hazards models were used to assess RFS, with the multivariable model adjusting for age, conditioning type, and donor type. Univariable and multivariable Fine and Gray competing risk models were applied to evaluate relapse and NRM, treating mortality and relapse as competing events, respectively. Model predictive performance was assessed using the concordance index (C-index). All MRD methods predicted relapse on univariable analysis, but subdistribution hazard ratios (SHR) for NGS BM were much more predictive for relapse (3.46, 95% confidence interval: 1.45, 8.28) and RFS (hazard ratio (HR) 2.20 (1.19, 4.06)) compared to NGS PB and BM MFC. NGS PB was more predictive than BM MFC (SHR 2.12 (0.87, 5.15) vs. 1.63 (0.46, 5.81) for relapse and HR 1.94 (0.97, 3.86) vs. 0.77 (0.24, 2.51) for RFS), but statistical power was low, likely due to fewer patients who had positive MRD pre-HCT using PB NGS and MFC (13.2% and 9.1%, respectively). In addition, current practice is to intervene with a variety of approaches post-HCT for pre-HCT MFC MRD+ patients to prevent relapse. These interventions likely led to better than expected RFS for those who were MFC MRD+ pre-HCT. Using multivariable models, NGS BM clearly predicted a higher risk for relapse and inferior RFS than NGS PB and BM MFC (SHR 3.58 (1.33, 9.60) vs. 1.83 (0.65, 5.20) vs. 1.58 (0.44, 5.71) for relapse, respectively, and HR 2.00 (1.03, 3.87) vs. 1.67 (0.79, 3.55) vs. 0.68 (0.21, 2.24) for RFS). This was also demonstrated by higher C-index for NGS BM compared to the other methods (0.64 vs. 0.58 vs. 0.54 for relapse and 0.60 vs. 0.56 vs. 0.51 for RFS). A total of 24 patients relapsed post-HCT. Pre-HCT flow was positive for 3/24 (12.5%) of those who relapsed, while PB NGS was positive for 7/24 (29.2%), and BM NGS was positive for 17/24 (70.8%). NGS MRD from the BM predicted relapse 5.7x more often than BM MFC and 2.4x more often than PB NGS. The high level of sensitivity of BM NGS MRD at defining relapse risk could facilitate planned post-HCT interventions to prevent relapse. Ongoing analyses to be presented at the meeting include assessment of the predictive power of post-HCT BM and PB NGS versus BM MFC MRD and an assessment of the effect of graft-versus-host disease (GvHD) on outcomes; notably, preliminary analyses showed a key role for acute GvHD in mitigating relapse in MRD+ patients, which will be further defined through ongoing landmark analyses.