Chronic Myeloid Leukemia (CML): BCR-ABL Transcript Levels Halving Time and Relative Reduction Ratio at Months 3 As New Predictors of Outcome. an Argentine Pilot Study

BCR-ABL transcript levels (IS%BCR-ABL) in CML patients (pts) allow to define molecular response (MR) to TKIs. MR at 3 months (M3>10%) is the strongest predictor of bad outcome and a point of divergence for changing therapy. Emerging biomarkers are being proposed in order to optimize prognosis in addition to initial MR and other clinical markers. The rate of BCR-ABL decline measured as halving time (HT) for transcript level reduction (RT^0.5) at month 3 in 1GTKI (imatinib) treated pts,  and the fractional clearance as a relative BCR-ABL ratio (RR3) during the first 3 months in 2GTKI (dasatinib) treated pts were recently proposed as predictive markers of outcome.  Optimal responses (OR) were defined upon %ISBCR-ABL values at 3, 6 and 12 months :  M3 ≤10%, M6 ≤1% y M12 ≤0,1%. Early time point prognostication could allow better outcome predictions and rapid therapeutic interventions.

Aim: to determine whether the rate of BCR-ABL decline as HT and the BCR-ABL transcripts reduction as relative ratio at M3 from baseline, have predictive value in CML pts treated with 2GTKIs.

Method: BCR-ABL transcript level was measured by Q-PCR according international recommendations and results were  reported as BCR-ABL/ABL on the international scale (IS BCR-ABL %).  Molecular monitoring was performed prior treatment (M0) and at months 3 (M3), 6 (M6), 12 (M12) and every 6 months thereafter. RT^0.5 from baseline to M3 was calculated as the time to halve by considering an exponential model of reduction during the period.  The median RT^0.5 and quartiles, were calculated to define rapid declines as those HT in the lowest quartile. RR3 was determined as a relative ratio of transcript clearance at M3 from baseline. The median RR3 and quartiles, were considered to define high BCR-ABL clearance as those in the highest quartile. Fisher statistical test was used to determine whether the proportions of good responses were statistically different depending on the decline rate.

Results: A total of 77 CML pts were included, first line treatment :  Dasatinib (D)  28%, Nilotinib (N) 32% and  Imatinib (I) 40%. Sokal risk score was available in 70% pts: low (L) 67%, intermediate (In) 22% and high (H) 11%.  Molecular assessment: prior therapy M0 (90%), M3 (60%), M6 (79%), M12 (78%) and M0+M3+M12 (39%). Deep responses (MR4.0) were considered as IS%BCR-ABL<0,01 or no detectable transcripts in ≥10000 ABL copies.  OR according to first line TKI (I vs D+N), were: M3≤10%: 75% vs 90% and M12 ≤0,1%: 50% vs 65%.  RT^0.5 median values for I vs D+N groups were 32d vs 16d and RR3 median values were 79% vs 98% . RT^0.5 ≤13d and RR3 ≥99% were cut off values for fast responses and high clearance in pts treated with 2GTKI(83%). Milestone molecular result M3≤10% showed 20% pts with M12>0,1% and without MR^4.0. All of these pts showed RT^0.5 >13d and RR3 <99%. From total pts with OR, 80% pts achieved MR^4.0 and all showed RR3≥99% meanwhile the remaining 20% pts who never achieved MR^4.0 had RR3<99%. Higher frequencies of OR were observed in low Sokal risk pts, in RT^0.5≤13d pts and in pts with RR3 ≥99%. High Sokal risk pts had poorer results at milestones molecular responses but those with rapid initial decline (RT^0.5 ≤13d) and high BCR-ABL clearance (RR3 ≥99%) achieved molecular responses similar to those observed in the low risk sokal group.

Conclusions: Faster BCR-ABL decline and high clearance were related to OR and MR^4.0 achievement in 2GTKIs patients. No direct relation between Sokal Index  and M12≤ 0,1% was observed. Patients with RT^0.5 ≤13d showed higher frequencies of good responses. Statistical differences were not found probably due to low sample numbers. Our data suggests that RT^0.5 in 2GTKIs (N+D) patients could have a similar predictive value as reported by Branford in 2014, even though it is known that BCR-ABL measurements tend to be less satisfactory when control gene is ABL, owing to possible non-linear ratios. Whilst there were no statistical correlations probably due to the low sample number, pts with OR but no MR^4.0 had low reduction rate RR3<99%. RR3>99% could provide a stronger predictive value for milestone optimal responses achievement in high risk pts. It will be necessary to extend this study including a higher number of patients treated with 2G TKIs.