Clinical Utility of Next-Generation Sequencing-Based Minimal Residual Disease in Childhood Acute Lymphoblastic Leukemia

Abstract

PurposeNext-generation sequencing (NGS)-based monitoring of minimal residual disease (MRD) was developed to increase the sensitivity and specificity of standard MRD detection methods. However, few published studies have tested the clinical utility of this novel technique. We assessed the clinical utility of NGS-MRD in a uniformly treated cohort of patients with pediatric B-cell precursor acute lymphoblastic leukemia (BCP-ALL).PATIENTS AND METHODSWe enrolled 79 unselected patients with pediatric BCP-ALL. Bone marrow samples were collected at the time of diagnosis, on days 33 and 80, at pre- and post-maintenance therapy time points (4-5 and 24 months, respectively), and upon relapse. Genomic DNA was extracted from frozen bone marrow mononuclear cells at each time point. We used diagnostic samples to define the immunoglobulin heavy chain (IGH), complementarity-determining region 3 (CDR3), and T-cell receptor gamma chain (TCRG) loci. From these samples, we detected leukemia-specific CDR3 sequences in >5.0% of all sequence reads. In addition, we performed a multiplex polymerase chain reaction (PCR) to determine the IGH, CDR3, and TCRG loci and subsequently assessed MRD using NGS. The result was considered positive for NGS-MRD if the leukemia-specific CDR3 sequence was detected. The resulting positive MRD values were categorized as "low positive" (<10−4) or "high positive" (≥10−4).RESULTSWe detected leukemia-specific CDR3 sequences in 72 of 79 patients (91%). MRD was measured in 232 samples and we obtained positive results in 59 samples. MRD was detected in 51% (28/55) samples on day 33, and the frequencies of positive MRD decreased to 25% (16/65), 19% (11/58), and 7.4% (4/54) samples at day 80, 4-5 months, and 24 months, respectively. Each of the four patients with a positive MRD at 24 months relapsed shortly after detection. In a univariate analysis, the MRD values at day 80 {risk ratio [RR; 95% confidence interval (CI)] = 7.438 (2.561-21.6), p < 0.001}, 4-5 months [RR (95% CI) = 10.24 (3.374-31.06), p < 0.001], and 24 months [RR (95% CI) = 19.26 (4.974-74.59), p < 0.001] showed a statistically significant association with inferior leukemia-free survival (LFS).The classification of patients as either low or high positive for NGS-MRD at day 80 was a significant risk factor for poor LFS [low positive, RR (95% CI) = 6.63 (2.01-21.82), p = 0.002; high positive, RR (95% CI) = 9.40 (2.32-38.17), p = 0.002]. Furthermore, both low and high positivity for MRD at 4-5 months was also a significant risk factor for poor LFS [low positive, RR (95% CI) = 10.32 (3.07-34.70), p < 0.001; high positive, RR (95% CI) = 10.04 (2.00-50.34), p = 0.005]. In an assessment of three multivariate Cox proportional hazard models, we found that both low and high positive NGS-MRD values at day 80 [low positive, RR (95% CI) = 6.05 (1.80-20.39), p = 0.0037; high positive, RR (95% CI) = 8.20 (1.92-35.07), p = 0.002] and at 4-5 months [low positive, RR (95% CI) = 12.98 (3.49-48.28), p < 0.001; high positive, RR (95% CI) = 23.16 (3.28-163.7), p < 0.001] were independent covariates predictive of poor LFS.CONCLUSIONWe detected leukemia-specific CDR3 rearrangements in 91% of our cohorts, which was comparable with the frequencies detected using sensitive real-time quantitative (RQ)-PCR methods. In both univariate and multivariate analyses, low and high positive NGS-MRD results were significantly associated with poor LFS. In addition, we found that MRD positivity at later time points (4-5 and 24 months) was predictive of a high incidence of relapse and poor LFS. Therefore, NGS-MRD can identify a greater number of patients who are at a high risk of relapse and candidates for intensified chemotherapy or allogeneic HSCT. Our study demonstrates the potential superiority of NGS over the current standard method of MRD monitoring. However, standardization, quality control, and validation of this new technology are warranted prior to its use in routine practice. DisclosuresNo relevant conflicts of interest to declare.