Purpose Although IKZF1 deletion ( IKZF1del) confers a higher risk of relapse in childhood B-cell acute lymphoblastic leukemia (B-ALL), it is uncertain whether treatment intensification will reverse this risk and improve outcomes. The Malaysia-Singapore ALL 2010 study (MS2010) prospectively upgraded the risk assignment of patients with IKZF1del to the next highest level and added imatinib to the treatment of all patients with BCR- ABL1 fusion. Patients and Methods In total, 823 patients with B-ALL treated in the Malyasia-Singapore ALL 2003 study (MS2003; n = 507) and MS2010 (n = 316) were screened for IKZF1del using the multiplex ligation-dependent probe amplification assay. The impact of IKZF1del on the 5-year cumulative incidence of relapse (CIR) was compared between the two studies. Results Patient characteristics were similar in both cohorts, including IKZF1del frequencies (59 of 410 [14.4%] v 50 of 275 [18.2%]; P = .2). In MS2003, where IKZF1del was not used in risk assignment, IKZF1del conferred a significantly higher 5-year CIR (30.4% v 8.1%; P = 8.7 × 10-7), particularly in the intermediate-risk group who lacked high-risk features (25.0% v 7.5%; P = .01). For patients with BCR-ABL1-negative disease, IKZF1del conferred a higher 5-year CIR (20.5% v 8.0%; P = .01). In MS2010, the 5-year CIR of patients with IKZF1del significantly decreased to 13.5% ( P = .05) and no longer showed a significant difference in patients with BCR-ABL1-negative disease (11.4% v 4.4%; P = .09). The 5-year overall survival for patients with IKZF1del improved from 69.6% in MS2003 to 91.6% in MS2010 ( P = .007). Conclusion Intensifying therapy for childhood B-ALL with IKZF1del significantly reduced the risk of relapse and improved overall survival. Incorporating IKZF1del screening significantly improved treatment outcomes in contemporary ALL therapy.
Identifying patient-specific clonal IGH/TCR junctional sequences is critical for minimal residual disease (MRD) monitoring. Conventionally these junctional sequences are identified using laborious Sanger sequencing of excised heteroduplex bands. We found that the IGH is highly expressed in our diagnostic B-cell acute lymphoblastic leukemia (B-ALL) samples using RNA-Seq. Therefore, we used RNA-Seq to identify IGH disease clone sequences in 258 childhood B-ALL samples for MRD monitoring. The amount of background IGH rearrangements uncovered by RNA-Seq followed the Zipf's law with IGH disease clones easily identified as outliers. Four hundred and ninety-seven IGH disease clones (median 2, range 0-7 clones/patient) are identified in 90.3% of patients. High hyperdiploid patients have the most IGH disease clones (median 3) while DUX4 subtype has the least (median 1) due to the rearrangements involving the IGH locus. In all, 90.8% of IGH disease clones found by Sanger sequencing are also identified by RNA-Seq. In addition, RNA-Seq identified 43% more IGH disease clones. In 69 patients lacking sensitive IGH targets, targeted NGS IGH MRD showed high correlation (R = 0.93; P = 1.3 × 10-14), better relapse prediction than conventional RQ-PCR MRD using non-IGH targets. In conclusion, RNA-Seq can identify patient-specific clonal IGH junctional sequences for MRD monitoring, adding to its usefulness for molecular diagnosis in childhood B-ALL.
Among the recently described subtypes in childhood B-lymphoblastic leukemia (B-ALL) were DUX4- and PAX5-altered (PAX5alt). By using whole transcriptome RNA sequencing in 377 children with B-ALL from the Malaysia-Singapore ALL 2003 (MS2003) and Malaysia-Singapore ALL 2010 (MS2010) studies, we found that, after hyperdiploid and ETV6-RUNX1, the third and fourth most common subtypes were DUX4 (n = 51; 14%) and PAX5alt (n = 36; 10%). DUX4 also formed the largest genetic subtype among patients with poor day-33 minimal residual disease (MRD; n = 12 of 44). But despite the poor MRD, outcome of DUX4 B-ALL was excellent (5-year cumulative risk of relapse [CIR], 8.9%; 95% confidence interval [CI], 2.8%-19.5% and 5-year overall survival, 97.8%; 95% CI, 85.3%-99.7%). In MS2003, 21% of patients with DUX4 B-ALL had poor peripheral blood response to prednisolone at day 8, higher than other subtypes (8%; P = .03). In MS2010, with vincristine at day 1, no day-8 poor peripheral blood response was observed in the DUX4 subtype (P = .03). The PAX5alt group had an intermediate risk of relapse (5-year CIR, 18.1%) but when IKZF1 was not deleted, outcome was excellent with no relapse among 23 patients. Compared with MS2003, outcome of PAX5alt B-ALL with IKZF1 codeletion was improved by treatment intensification in MS2010 (5-year CIR, 80.0% vs 0%; P = .05). In conclusion, despite its poor initial response, DUX4 B-ALL had a favorable overall outcome, and the prognosis of PAX5alt was strongly dependent on IKZF1 codeletion.