Measurable residual disease detection by high-throughput sequencing improves risk stratification for pediatric B-ALL

Abstract

Early response to induction chemotherapy is an important prognostic factor in B-lymphoblastic leukemia (B-ALL). Here, we compare high-throughput sequencing (HTS) of IGH and TRG genes vs flow cytometry (FC) for measurable residual disease (MRD) detection at the end of induction chemotherapy in pediatric patients with newly diagnosed B-ALL. Six hundred nineteen paired pretreatment and end-of-induction bone marrow samples from Children's Oncology Group studies AALL0331 (clinicaltrials.gov #NCT00103285) (standard risk [SR]; with MRD by FC at any level) and AALL0232 (clinicaltrials.gov #NCT00075725) (high risk; with day 29 MRD <0.1% by FC) were evaluated by HTS and FC for event-free (EFS) and overall survival (OS). HTS and FC showed similar 5-year EFS and OS for MRD-positive and -negative patients using an MRD threshold of 0.01%. However, there was a high discordant rate with HTS identifying 55 (38.7%) more patients MRD positive at this threshold. These discrepant patients have worse outcomes than FC MRD-negative patients. In addition, the increased analytic sensitivity of HTS permitted identification of 19.9% of SR patients without MRD at any detectable level who had excellent 5-year EFS (98.1%) and OS (100%). The higher analytic sensitivity and lower false-negative rate of HTS improves upon FC for MRD detection in pediatric B-ALL by identifying a novel subset of patients at end of induction who are essentially cured using current chemotherapy and identifying MRD at 0.01% in up to one-third of patients who are missed at the same threshold by FC.

Figure 1.

EFS and OS for HTS and FC at a threshold of 0.01%. The Kaplan-Meier estimates of EFS (A) and OS (B) at an analytical cutoff of 0.01% for MRD as measured by both HTS and FC. HTS showed similar 5-year EFS and OS to that of FC.

Figure 2.

Discordant HTS and FC MRD have an intermediate prognosis. A Kaplan-Meier estimate of EFS between patients who were negative for MRD as measured by both HTS and FC, positive for HTS and FC, and HTS positive/FC negative at a threshold of 0.01% identified a cohort of 55 patients who were negative for MRD by FC and positive by HTS. These patients had an intermediate prognosis between patients who were negative by both HTS and Flow and patients who were positive by both HTS and Flow. For the Flow-negative group, the outcome was worse for patients who were HTS positive (P = .036).

Figure 3.

Cox proportional hazard ratios at different HTS MRD cutoff thresholds. Hazard ratios were calculated at different thresholds using the Cox Proportional model for the whole cohort (A) and separately for the SR group (B) and the high-risk group (C). The ratios were plotted over the range of HTS MRD levels to determine an optimal cutoff for risk stratification. For the whole group, the hazard ratio was maximal at a level close to 0.01%, and similarly for the high-risk group (C). The SR group showed a broader distribution around a 0.01% cutoff.

Figure 4.

EFS and OS for HTS MRD positive at any level vs negative for the SR group. Kaplan-Meier estimates of EFS (A) and OS (B) for patients in the SR group who were positive for MRD by HTS at any level vs negative demonstrated better EFS for the HTS negative cohort (P = .0226) and a better OS. In the latter group, the OS of HTS-negative patients was 100%.

Figure 5.

EFS and OS for HTS MRD positive at any level vs negative for the high-risk group. Kaplan-Meier estimates of EFS (A) and OS (B) for patients in the high-risk group who were positive for MRD by HTS at any level vs negative demonstrated better EFS and OS for the HTS-negative cohort, although the P values did not demonstrate statistical significance.

Figure 6.

An EFS and OS for patients with or without clonal IGH rearrangements. Kaplan-Meier estimates of EFS (A) and OS (B) for patients from both risk groups combined who did not have a dominant clonal IGH rearrangement by HTS. A subset of 67 patients had no IGH rearrangements and had a worse prognosis than those in whom IGH rearrangements were detected (P = .0110). There was no discernible difference in OS for these patients.