MRD Monitoring Using Minor-BCR-ABL1 Genomic Breakpoint in Childhood ALL Identifies a Subgroup with Distinct Biology and a Very Poor Prognosis

Background:

The exact role of minimal residual disease (MRD) testing in childhood BCR-ABL1+ ALL is still unclear and MRD levels are used rather for stem cell transplantation (SCT) indication than for frontline treatment stratification. There are two standard targets used for MRD monitoring in these ALLs: Ig/TR rearrangements and the fusion transcript quantification. However, despite similar sensitivity of the methods, we previously showed a poor correlation between them in some patients. Overall, 20% of samples negative by Ig/TR MRD were BCR-ABL1 positive (even at high levels). By demonstrating presence of BCR-ABL1 within the non-lymphoid (Ig/TR-negative) population in the non-correlating samples we showed that multilineage involvement is at least partly responsible for the discrepancy (Zaliova, Leukemia 2009). Here we established MRD monitoring based on the quantification of the patient-specific genomic (intronic) BCR-ABL1 breakpoint sequence, compared MRD levels obtained from different approaches and analyzed their prognostic significance.

Methods:

Our cohort totals 35 children diagnosed with minor-BCR-ABL1+ ALL. The BCR-ABL1 genomic breakpoint was found using multiplex long distance DNA PCR. The patient-specific intronic fusion sequences were used for MRD quantification by qPCR. We quantified MRD levels in 386 bone marrow (BM) samples. The results of genomic BCR-ABL1 quantification were compared with Ig/TR and BCR-ABL1 transcript levels. Samples with >1 log difference in MRD levels were considered non-correlating. For correlation analysis of methods, double-negative (DN) samples were excluded. Overall survival (OS) rates were calculated according to Kaplan-Meier and compared by log-rank test.

Results:

Analysis of MRD in BM samples confirmed poor correlation between Ig/TR and BCR-ABL1 mRNA quantification (Spearman correlation coefficient R=0.70; n=166 non-DN samples). Moreover, we also saw poor correlation when comparing the two DNA methods (Ig/TR vs. BCR-ABL1 DNA, R=0.69; n=254 non-DN samples) with 21% of samples positive by BCR-ABL1 (at levels up to 10e-1) while negative by Ig/TR. While in some patients the correlation of the DNA-based methods was very good, others had several subsequent follow-up samples with poor correlation (> 1 log difference) suggesting distinct biology of the disease. MRD levels determined by BCR-ABL1 DNA quantification had the best predictive value for overall survival (the best separation of survival curves was at week 12 of treatment with 10e-3 cut-off: BCR-ABL1 DNA, p=0.0003; Ig/TR, p=0.017; BCR-ABL1 mRNA, not significant).

The proportion of samples with BCR-ABL1 DNA levels higher than Ig/TR by more than 1 log increased with time on treatment (day 15: 10% (2/20); day 33: 26% (7/27); week 12: 39% (7/18); week 22: 50% (7/14)). Importantly, presence of samples with this poor correlation was prognostically relevant as patients with BCR-ABL1 levels higher than Ig/TR by more than 1 log at the end of induction treatment (day 33) had significantly worse outcome (2 years OS 84+/-9% (n=21) vs. 28+/-23% (n=6) for correlating and non-correlating patients, respectively; p=0.004).

Conclusion:

The BCR-ABL1 quantification at genomic level probably provides the most accurate measurement of leukaemic burden. Due to multi-lineage involvement in some Ph+ ALL as well as occasional presence of subclones with different Ig/TR markers, quantification of the BCR-ABL1 gene breakpoint is more reliable than Ig/TR monitoring. Moreover, in contrast to the BCR-ABL1 transcript, measurement of the gene is not influenced by possible variability in expression between cell subtypes or during concurrent treatment. Importantly, the MRD levels determined by BCR-ABL1 DNA-based testing have higher predictive value than Ig/TR DNA or BCR-ABL1 transcript quantification. The poorer outcome observed in the subset of patients in whom BCR-ABL1 DNA-based MRD levels were higher than Ig/TR-based MRD detection possibly reflects multilineage involvement, suggesting, that some cases with minor-BCR-ABL1+ ALL are in fact CML cases identified in a lymphoid blast crisis. In view of their poor outcome, these cases might be candidates for early SCT or alternative treatments. The most reliable method for MRD detection in BCR-ABL1+ ALL needs to be further investigated within a large therapeutic protocol.

Support: GAUK 554214; MH CZ-DRO UH Motol 00064203; NHMRC Australia APP1057746.