Rapid identification of BCR/ABL1-like acute lymphoblastic leukaemia patients using a predictive statistical model based on quantitative real time-polymerase chain reaction: clinical, prognostic and therapeutic implications.

ABSTRACT

BCR/ABL1-like acute lymphoblastic leukaemia (ALL) is a subgroup of B-lineage acute lymphoblastic leukaemia that occurs within cases without recurrent molecular rearrangements. Gene expression profiling (GEP) can identify these cases but it is expensive and not widely available. Using GEP, we identified 10 genes specifically overexpressed by BCR/ABL1-like ALL cases and used their expression values - assessed by quantitative real time-polymerase chain reaction (Q-RT-PCR) in 26 BCR/ABL1-like and 26 non-BCR/ABL1-like cases to build a statistical "BCR/ABL1-like predictor", for the identification of BCR/ABL1-like cases. By screening 142 B-lineage ALL patients with the "BCR/ABL1-like predictor", we identified 28/142 BCR/ABL1-like patients (19·7%). Overall, BCR/ABL1-like cases were enriched in JAK/STAT mutations (P < 0·001), IKZF1 deletions (P < 0·001) and rearrangements involving cytokine receptors and tyrosine kinases (P = 0·001), thus corroborating the validity of the prediction. Clinically, the BCR/ABL1-like cases identified by the BCR/ABL1-like predictor achieved a lower rate of complete remission (P = 0·014) and a worse event-free survival (P = 0·0009) compared to non-BCR/ABL1-like ALL. Consistently, primary cells from BCR/ABL1-like cases responded in vitro to ponatinib. We propose a simple tool based on Q-RT-PCR and a statistical model that is capable of easily, quickly and reliably identifying BCR/ABL1-like ALL cases at diagnosis.