Ultrasensitive Next-Generation Sequencing-Based Measurable Residual Disease Assessment in Philadelphia Chromosome-Negative Acute Lymphoblastic Leukemia after Frontline Therapy: Correlation with Flow Cytometry and Impact on Clinical Outcomes

Background: Measurable residual disease (MRD) by multiparameter flow cytometry (MFC) is highly prognostic for relapse and overall survival (OS) in patients (pts) with acute lymphoblastic leukemia (ALL). However, many pts with apparent “MRD negativity” by MFC still relapse. These relapses are likely due to residual leukemia that is present below the level of detection of MFC (generally ~1x10^-4). In light of the limitations of conventional MRD assays, we sought to evaluate the clinical impact of a highly sensitive next-generation sequencing (NGS)-based MRD assay that is capable of detecting residual leukemia at a level of 1x10^-6.

Methods: We performed a retrospective study of 67 pts with previously untreated Philadelphia chromosome-negative ALL who received frontline therapy between 2/2012 and 7/2018 with a hyper-CVAD-based regimen (n=44) or a lower-intensity hyper-CVD-based regimen (n=23), which was usually combined with inotuzumab ozogamicin, and achieved complete remission (CR). Pts were selected for this study based on availability of banked samples at baseline and at sequential remission time points up to 4.5 months after start of therapy. Six-color MFC was performed as standard of care with an analytic sensitivity of 0.01% (1x10^-4). Pretreatment genomic DNA derived from bone marrow was evaluated using a quantitative NGS MRD assay (Adaptive Biotechnologies Co., Seattle, WA) to identify dominant rearrangements within the B-cell receptor (IgH [VDJ and DJ], IgK and IgL). High throughput NGS of remission samples was subsequently performed to assess the target clonal rearrangements at an established analytic sensitivity of 0.0001% (1x10^-6).

Results: The baseline characteristics of the study population are shown in Table 1. Twenty pts (30%) underwent allogeneic stem cell transplant (alloSCT) in first remission. In the entire cohort, 14 pts relapsed (21%), including 2 with CNS-only relapses. The median duration of follow-up for the entire cohort was 51 months.

Among the 109 remission samples, 84 were MRD-negative (MRD^neg), 24 were MRD-positive (MRD^pos), and 1 was indeterminate by MFC (Figure 1). All 24 MRD^pos samples by MFC were also MRD^pos by NGS. Among the 84 MRD^neg samples by MFC, 32 (38%) were MRD^pos by NGS. For these discordant cases, the median level of MRD by NGS was below the expected level of detection by MFC (median 0.002% [range 0.00002%-1.15%]).

At the time of CR, 44/67 pts (66%), achieved MRD^neg by MFC and 14/38 (37%) achieved MRD^neg by NGS. The rates of MRD^neg by MFC in pts treated with hyper-CVAD or hyper-CVD regimens were 61% and 74%, respectively, and by NGS were 32% and 46%, respectively. Achieving MRD^neg by NGS at the time of CR better predicted the likelihood of durable remission than did achieving MRD^neg by MFC. The 5-year cumulative incidence of relapse (CIR) of the 44 pts who achieved MRD^neg by MFC and of 14 pts who achieved MRD^neg by NGS was 27% and 8%, respectively. Only 1 out of 14 pts who achieved MRD^neg by NGS at the time of CR subsequently relapsed. In contrast, MRD status by NGS did not predict for relapse when evaluated at post-CR time points.

Among pts treated with a more intensive hyper-CVAD regimen, achievement of MRD^neg by NGS at CR provided additional prognostic information compared to MRD status by MFC at the same time point and identified pts with an excellent long-term outcome. The 5-year OS rate for pts who achieved MRD^neg by NGS at CR was 100% vs. 55% for those who were MRD^pos by NGS (Figure 2A; P<0.05). This superior outcome was driven by the lower CIR rate in the MRD^neg pts (Figure 2B; 5-year CIR: 13% vs. 58% in MRD^pos pts). The impact of achieving MRD^neg by NGS at CR was particularly evident in pts who did not undergo alloSCT in first remission where the 5-year CIR rates were 14% and 81%, respectively. When integrating MRD information by MFC and NGS, the 5-year OS rate for pts who were MRD^neg by both MFC and NGS, MRD^neg by MFC but MRD^pos by NGS, and MRD^pos by both MFC and NGS were 100%, 67%, and 38%, respectively (P=0.02 for trend). Similarly, the 5-year CIR rates were 13%, 57%, and 63%, respectively.

Conclusion: Early assessment of MRD using an ultrasensitive NGS assay can identify pts with ALL who have a very low risk of relapse and excellent long-term survival. Prospective studies are warranted to assess whether de-intensification of therapy is feasible for this favorable-risk group of pts who rapidly achieve MRD^neg remission.