Minimal residual disease by high-throughput sequencing in standard risk–favorable pediatric B-lymphoblastic leukemia

TO THE EDITOR:

Most children with B-lymphoblastic leukemia (B-ALL) are cured by modern chemotherapy immunotherapy, but their burden of treatment-related toxicity underscores a need to further refine risk-adapted therapy. Prognostic advances from frontline incorporation of blinatumomab¹ have stimulated interest in testing whether carefully selected, favorable-risk groups might safely benefit from chemotherapy deescalation. The responsible design of such studies requires a clear understanding of how all prognostic information should inform eligibility.

Minimal/measurable residual disease (MRD) at the end of induction (EOI) remains one of the strongest predictive tools for guiding postinduction treatment intensity for pediatric B-ALL.² High-throughput sequencing (HTS) of clonal immunoglobulin gene rearrangements enhances MRD detection sensitivity below the typical 0.01% threshold of flow cytometry MRD (fcMRD) and provides clinically actionable prognostic information, with the guiding principle that patients who achieve and sustain undetectable HTS-MRD fare better than those with MRD persistence.3, 4, 5, 6 However, because MRD behaves as a continuous variable with unique, subgroup-specific thresholds,^7,8 the clinical relevance of low-level detection in the context of a specific treatment regimen may not be uniform across all patients. In particular, children classified as Standard Risk–Favorable (SR-FAV) by criteria defined in the Children’s Oncology Group (COG) trial AALL1731¹ have consistently outstanding outcomes with chemotherapy alone,^9,10 but the AALL1731 stratification schema, now widely adopted in most of North America, did not incorporate HTS-MRD evaluation for SR-FAV assignment. HTS-MRD has clear prognostic value and is frequently evaluated before final risk group assignment. Thus, defining how EOI HTS-MRD should influence SR-FAV eligibility remains an unresolved gap in current practice.

In the largest published pediatric HTS-MRD analysis to date, Wood et al.³ retrospectively evaluated EOI HTS-MRD using bone marrow samples from a subset of patients from COG AALL0331¹¹ and AALL0232.¹² Of the 296 evaluable National Cancer Institute Standard Risk (age, ≥1 to <10 years; white blood cell count, <50 × 10³/μL) patients,^3,13 53 met AALL1731-defined SR-FAV criteria by retrospective classification (Figure 1A). Day 8 peripheral blood (PB) fcMRD was unavailable for 19 patients; therefore, this parameter was omitted from classification. Four patients lacked trackable sequence(s) at diagnosis. Thus, we analyzed the incidence and prognostic relevance of EOI HTS-MRD in 49 SR-FAV patients (supplemental Table 1).

Figure 1.

Detectable MRD by HTS is prevalent at the EOI in patients with SR-FAV B-ALL. (A) Among the 296 patients from COG AALL0331 evaluated by HTS, 53 (17.9%) met COG AALL1731-defined SR-FAV risk classification criteria. SR-AVG and SR-HIGH classifications encompassed 172 (58.1%) and 68 (23.0%) of the patients, respectively, whereas 3 patients (1%) would have been classified as HR because of CNS3 staging. (B) The incidence of HTS-MRD positivity and (C) level of HTS-MRD detectable in the BM at EOI did not significantly differ between 49 evaluable SR-FAV patients and 166 evaluable SR-AVG patients with available EOI HTS-MRD data. ∗Information about day 8 PB fcMRD status was not consistently available for patients treated on COG AALL0331 and was thus omitted from retrospective classification. BM, bone marrow; HR, high risk; NCI, National Cancer Institute; ns, not significant.

Among these 49 patients, 12 (24.5%) had undetectable HTS-MRD, 33 (67.3%) had detectable HTS-MRD below the 0.01% fcMRD threshold, and 4 (8.2%) showed discordant results (HTS, ≥0.01%; fcMRD, <0.01%). Neither the likelihood of having detectable EOI HTS-MRD nor the level of HTS-MRD differed between SR-FAV and SR-Average (SR-AVG) patients who did not otherwise meet SR-FAV or SR-HIGH criteria (Figure 1B-C). Importantly, among 37 SR-FAV patients with HTS-MRD⁺ status, 28 (75.7%) had double trisomy (DT) cytogenetics, which is associated with favorable outcomes despite slower disease clearance kinetics relative to other subtypes.¹⁴

Five-year event-free survival and overall survival did not differ by EOI HTS-MRD status for SR-FAV patients (Figure 2A-B), in contrast to the inferior 5-year event-free survival observed in SR-AVG patients with HTS-MRD⁺ status (supplemental Figure 1). There were 4 SR-FAV events, all in patients with DT cytogenetics. Three relapses in evaluable patients included (1) late (>36 months from diagnosis) combined medullary/central nervous system relapse in a patient with undetectable HTS-MRD, (2) late medullary relapse in a patient with HTS-MRD of 0.000042% (<1 cell per million), and (3) late medullary relapse in a patient with HTS-MRD of 0.0062% (62 cells per million). None of the 4 patients with HTS/fcMRD discordance experienced relapse; all had DT cytogenetics and HTS-MRD <0.1%.

Figure 2.

Clinical impact of detectable EOI HTS-MRD among retrospectively classified SR-FAV patients from COG AALL0331. (A) 5-year event-free survival and (B) overall survival are shown for 49 patients with available EOI HTS-MRD data who met SR-FAV risk classification criteria. Because of incomplete information about day 8 PB fcMRD status for patients treated on COG AALL0331, this parameter was omitted from classification.

Since its US Food and Drug Administration clearance in 2018, the clinical HTS-MRD assay (clonoSEQ; Adaptive Biotechnologies)¹⁵ has become a broadly implemented tool for guiding B-ALL treatment decisions. Extensive pediatric and adult data support its general prognostic value, including at EOI.3, 4, 5, 6^,16 To our knowledge, this analysis is the first focused assessment of EOI HTS-MRD in SR-FAV B-ALL and provides context for how HTS-MRD data should influence classification for this historically favorable-risk cohort. In this analysis, the absence of an observed survival impact from low-level EOI HTS-MRD persistence suggests that other defining features of SR-FAV B-ALL, namely, favorable cytogenetics (DT, with its known slower early disease clearance,¹⁴ or ETV6::RUNX1) and EOI bone marrow fcMRD clearance (at least below the 0.01% threshold), remain primary drivers of excellent SR-FAV outcomes. These findings align with prior evidence. For instance, immunoglobulin/T-cell receptor polymerase chain reaction MRD data from UKALL2003 demonstrated variable prognostic MRD thresholds across cytogenetic groups, highlighting the importance of considering subtype-specific treatment response kinetics in classifying risk.⁷ As with other prognostic variables,⁸ HTS-MRD functions as a continuous, rather than binary, predictor of relapse influenced by disease biology, sampling variability, and the unique treatment response dynamics of each patient. Nevertheless, if future trials are to explore treatment deescalation in favorable subgroups, they will require clear, evidence-based eligibility parameters that encompass all relevant biomarkers. COG AALL1731¹ systematically incorporated HTS-MRD testing for SR-AVG patients but not SR-FAV. Our data (interpreted strictly in the context of a COG AALL0331 chemotherapy backbone without deescalation) support maintaining SR-FAV classification even in the presence of low-level EOI HTS-MRD.

This analysis is limited by its relatively small sample size, retrospective nature, and low event rate, which may preclude the ability to detect small but potentially clinically meaningful differences in relapse risk between EOI HTS-MRD detectable vs undetectable patients. Accordingly, we cannot exclude the possibility that low-level HTS-MRD may be associated with an increase in relapse risk that this analysis is underpowered to detect. Despite these limitations, to our knowledge, this is the largest HTS-MRD analysis in patients with SR-FAV B-ALL treated with a COG standard-intensity chemotherapy backbone. Similar patient populations over several generations of COG trials, none of which incorporated HTS-MRD, have demonstrated outstanding outcomes.^9,10 Given the overall low event rate in this patient population and the high prevalence of low-level HTS-MRD detection, any such increase in risk is expected to be modest at most. Indeed, given that two-thirds of SR-FAV patients in this analysis had low-level EOI HTS-MRD, a larger dataset is unlikely to reveal a large or clinically actionable difference in outcomes. We also acknowledge the potential for inaccurate retrospective SR-FAV classification in our analysis because of (1) incomplete day 8 PB MRD data and (2) AALL0331 not yet encompassing expanded cytomolecular evaluation of co-occurring lesions, which can impact prognosis.¹⁷ However, these omissions would bias our cohort toward worse outcomes, which we did not observe. Likewise, subgroup analysis of the 25 patients with known day 8 PB MRD <1% recapitulated that low-level EOI HTS-MRD did not impair prognosis (supplemental Figure 2). Finally, restricting our analysis to retrospective HTS data encompassing only immunoglobulin heavy chain and T-cell receptor rearrangements,^3,13 although this assay now includes trackable immunoglobulin heavy chain and immunoglobulin light chain rearrangements, could modestly affect results, even though this is unlikely to have substantially biased our conclusions, considering only 4 patients (7.5%) lacked a trackable rearrangement.

In summary, HTS frequently detects low-level EOI MRD persistence in pediatric patients with SR-FAV B-ALL below the threshold of fcMRD detection. In the context of an intact SR-FAV chemotherapy backbone, our findings do not identify an association between low-level EOI HTS-MRD and outcome in this otherwise favorable risk group and thus do not support introducing an arbitrarily rigid HTS-MRD cutoff that would artificially dichotomize risk. Rather, our data suggest that low-level HTS-MRD at a single time point (EOI) should not, in isolation, affect eligibility for SR-FAV classification or trial eligibility. Though limited sample size and very low event rate restrict the statistical power of this analysis, our findings provide a framework for interpreting low-level HTS-MRD detection in the context of an otherwise established favorable-risk subgroup for which prospective HTS-MRD data are unlikely to be readily available. Specifically, our findings do not support that low-level HTS-MRD itself necessitates augmentation of therapy for SR-FAV patients. Confirmation of favorable outcomes in the larger cohort of SR-FAV patients from AALL1731, for whom HTS-MRD was not assessed, may aid in further validating this approach. In an era in which blinatumomab has improved outcomes across heterogeneous B-ALL subgroups,^1,18, 19, 20 clinical decision-making should integrate HTS-MRD results into the broader clinical and genomic context, encompassing all relevant variables to decide optimal treatment and trial suitability for each patient.

Conflict-of-interest disclosure: L.J. serves on a data and safety monitoring committee board for Pfizer. M.L.L. serves on advisory boards for Jazz Pharmaceuticals and Illumina Inc and is a member of the Amgen Speakers Bureau. L.W.L. is employed by Adaptive Biotechnologies. I.K. is employed by Adaptive Biotechnologies and has an equity interest in the company. D.T.T. serves on unpaid advisory boards for AbbVie, AstraZeneca, Amgen, Autolus, Beam Therapeutics, BeOne, Jazz Pharmaceuticals, Johnson & Johnson Innovation, Merck, Novartis, Pfizer, Servier, Sobi, and Syndax; has received laboratory research funding from Beam Therapeutics and NeoImmune Tech; and is an inventor on chimeric antigen receptor (CAR) T-cell-related patents or pending patents. S.G. has received honoraria and travel support from Amgen, honoraria from BeOne, and consulting fees from Jazz Pharmaceuticals. B.L.W. has served as a consultant for Amgen. R.E.R. has received honoraria and consulting fees from Jazz Pharmaceuticals and honoraria and travel support from Amgen. The remaining authors declare no competing financial interests.