qPCR MRD Monitoring in Peripheral Blood May Predict Hematological Relapse in Pediatric Acute Myeloid Leukemia

Background: Long-term patient survival in childhood acute myeloid leukemia (AML) has improved dramatically the last decades. However, 30% of children with AML die from the disease with hematological relapse being the most common event and cause of treatment failure. Early detection of imminent hematological relapse during follow-up may facilitate preparation of upcoming therapy including hematopoietic stem cell transplantation (HSCT) or an intervention e.g. preemptive therapy. Increasing levels of minimal residual disease (MRD) in peripheral blood (PB) may predict impending relapse in patients in first complete remission (CR1). Quantitative PCR (qPCR) is a highly sensitive assay suitable for MRD quantification in PB but absence and instability of valid qPCR MRD markers limit molecular detection of disease recurrence in a large proportion of patients. The Wilms tumor 1 gene (WT1) is frequently overexpressed at diagnosis in hematological malignancies but the predictive value of WT1 expression levels in childhood AML during follow-up remain uncertain.

We investigated the applicability of various fusion transcripts and WT1 expression as putative MRD targets in PB disease monitoring.

Methods: From January 2004 to August 2014, 135 pediatric patients were diagnosed with de novo AML and treated on two consecutive protocols, NOPHO-AML 2004 and NOPHO-DBH AML 2012, in Iceland, Finland and Denmark. Potential qPCR MRD targets were identified in 69% (93/135) of these patients. qPCR MRD were quantified in PB at scheduled time points during therapy followed by sampling every second month from end of therapy until 2 years after diagnosis or hematological relapse. qPCR MRD data were available from at least 5 time points (median samples/patient: 13, range 5-19) after induction therapy in 42 patients; RUNX1-RUNX1T1 (8), CBFβ/MYH11 (4), MLLT3/MLL (7), MLL-ENL (2), and WT1 overexpression (21). WT1 was considered overexpressed if WT1/ABL ratio was higher than 1/16 at diagnosis. Patients with both fusion transcript and WT1 overexpression were monitored using the fusion transcript as MRD target.

Molecular remission (mCR) for fusion transcripts was defined as an MRD level in PB < 5x10^-4. Molecular relapse was defined as a 10-fold increase in MRD to a level of at least 5x10^-4.

A remission baseline WT1 expression was defined as the mean of the last three WT1 levels during consolidation therapy. Molecular relapse in patients with WT1 overexpression was defined as a 10-fold increase in MRD compared to baseline WT1 expression and above normal WT1 expression range.

Results: Median follow-up time in patients with fusion transcripts was 74 weeks from diagnosis (range: 20-169). Seven of 21 patients with fusion transcripts experienced molecular relapse during follow-up and all patients subsequently experienced hematological relapse. Median interval between molecular and hematological relapse was 3.4 weeks (range: 0.1-18.1). Patients in continuous mCR remained in CR1.

Median follow-up time in patients with WT1 overexpression was 116 weeks from diagnosis (range: 40-169). Five of 21 patients developed hematological relapse. In all cases relapse was preceded by molecular relapse. Median interval between molecular and hematological relapse was 2.7 weeks (range: 2.0-5.9). A similar increase in WT1 expression was not detected in any long-term CR1 patients.

Conclusion: Introduction of WT1 facilitates molecular disease monitoring in the vast majority of pediatric AML patients in CR1. Although numbers are small, our study indicates that both fusion transcripts and WT1 may serve as sensitive MRD targets. All patients in continuous CR1 maintained mCR throughout follow-up. All overt hematological relapses were preceded by molecular relapse though in 50% of cases the interval between molecular and hematological relapse was 3 weeks or less. Interval from molecular to hematological relapse showed great diversity in patients with fusion transcripts indicating heterogeneous relapse kinetics according to the underlying genetic lesion. A shorter interval between monitoring sampling may allow earlier detection of molecular relapse and a potential window for preemptive therapy.