End-of-Consolidation MRD Predicts Survival in Pediatric Core-Binding Factor AML, Taiwan Group Reports

The induction timepoint carried no signal

At the end of induction, children with RUNX1::RUNX1T1 leukemia who cleared a 2-log reduction had 72.7% five-year relapse-free survival and those who did not had 76.2%, a difference with a P value of 0.862. Overall survival at that timepoint split 77.9% against 72.7%, P = 0.640. Push the assessment two cycles later and the curves come apart: a 3-log reduction or better at the end of the second consolidation gave 82.8% relapse-free survival against 44.5% (P = 0.007), and a 4-log reduction at the end of consolidation gave 96.2% (95% CI 88.7 to 100.0) against 57.1% (31.2 to 82.9), P = 0.001, with overall survival at 95.7% against 67.5%, P = 0.013. Only 55% of the cohort cleared a 2-log reduction at end of induction, which the authors attribute to their lighter induction backbone of three days of idarubicin and seven of cytarabine under the TPOG-AML-97A protocol. In multivariate Cox modeling, failure to reach hematological complete remission after one induction cycle and MRD below 4-log reduction at end of consolidation were the independent adverse predictors of overall survival, and the MRD variable independently predicted relapse-free survival as well.

KIT D816, and the MRD result that overrides it

Sequencing of KIT exons 8, 11 and 17 turned up variants in 30 of 75 tested RUNX1::RUNX1T1 patients (40%) and eight of 28 with CBFB::MYH11 (28.5%). Carrying any KIT variant did not move survival in the RUNX1::RUNX1T1 group. The D816 substitution did, lowering five-year relapse-free survival with a P value of 0.017, and in the CBFB::MYH11 group any KIT variant lowered five-year overall survival (P = 0.019). The authors read the discrepancy as biological and statistical at once, since D816 sits in the activation loop and drives ligand-independent signaling while many non-D816 changes have uncertain function, and pooling the two dilutes the effect. Only 34% of D816 carriers reached a 4-log reduction at end of consolidation, and those who did fared better than those who did not. Their conclusion is that "achieving MRD ≥ 4-LR at EOC could overcome the adverse effect of KIT D816 and that clinical decisions should integrate the variant status and MRD kinetics rather than relying on the variant status alone."

Twenty-eight patients is not enough for the CBFB arm

The CBFB::MYH11 findings come with numbers small enough that the authors flag them as hypothesis-generating. A 2-log reduction at end of induction went with 100% five-year overall survival in 15 patients against 50% in four (P = 0.003), and a 4-log reduction at end of consolidation with 100% relapse-free survival in seven against 50% in five (P = 0.044). Every CBFB::MYH11 patient who had an end-of-consolidation measurement stayed in continuous complete remission through follow-up, which shut down further comparisons in that subgroup. Two other limits are stated plainly: not every child had sequential MRD drawn, and the 23-year accrual window spans real changes in chemotherapy and supportive care.

Log reduction and copy number sometimes disagreed

No previous series has reported log reduction and normalized copy number for these two fusion transcripts side by side, and the two measures sometimes disagreed. Children whose end-of-consolidation MRD fell below a 4-log reduction while their copy number stayed above 500 per 100,000 ABL1 transcripts had, in the authors' words, extremely poor survival, worse than any other combination. The paper asks laboratories to report both.

Why This Matters to the APO|APE Reader

A laboratory running flow cytometric AML MRD cannot generate the number this study made prognostic: the separation came from quantitative fusion-transcript PCR normalized to ABL1, with only samples above 100,000 ABL1 copies counted as assessable and a defined copy-number threshold at end of consolidation. That distinction has already produced conflicting literature, and the authors point to it, noting a Japan Children's Cancer Group report in which KIT exon 17 carriers did poorly even with negative flow-based MRD at the end of induction. The disagreement traces to timing and platform, so a center that reports KIT status from a myeloid panel without a matched molecular MRD schedule hands the pediatric oncologist half of the risk assignment this paper describes.