BCR::ABL1/GUSB halving times predict molecular response to asciminib at 18 months

Abstract The initial rate of reduction of BCR::ABL1 mRNA on TKI therapy, usually measured as the halving time over the first 3 months of treatment, has prognostic value for both early and late responses as well as the likelihood of achieving TFR. Standardized measurement of pretreatment disease levels by RT-qPCR, however, is problematic due to the different properties of the 3 established reference genes (ABL1, GUSB, BCR) used for CML MRD analysis. These differences are not relevant for standard MRD assessments, but the use of ABL1 or BCR results in reference gene-specific distortions at high disease burdens that limit their utility for assessment of baseline disease levels. GUSB is completely independent of BCR::ABL1 and is thus theoretically the best reference gene to determine halving times. To date, however, no comparative analysis of the utility of different reference genes to assess early response kinetics has been performed. We have evaluated the utility of halving times derived using GUSB and ABL1 as reference genes in FASCINATION, a multicenter, prospective, open-label, interventional phase II trial designed to evaluate the efficacy and tolerability of asciminib following <6 weeks treatment with other TKIs or <4 weeks of hydroxyurea. Halving times in days were calculated by comparing BCR::ABL1/ABL1 and BCR::ABL1/GUSB levels at trial entry (63% and 23%, respectively, at the start of asciminib) to the corresponding measurements at 3 months on asciminib. Halving times and IS levels at 3 months were compared to molecular outcomes at 18 months. Of 103 cases with available data, 45 achieved DMR at 18 months (MR4 or better; median 0.0015% IS) and 58 did not achieve DMR (median 0.052% IS). Median halving times were shorter for patients who achieved DMR using both ABL1 (13.5 vs 20.9, P<0.001) and GUSB (11.2 vs 18.1, P<0.001) as reference genes. Similarly, IS values at 3 months were significantly lower for cases who achieved DMR at 18 months (0.23% vs 1.6%; P<0.001). Using ROC analysis, the predictive value of GUSB halving times (AUC GUSB = 0.792) for achievement of DMR at 18 months was superior to both ABL1 halving times (AUC ABL1 = 0.746) and IS values at 3 months (AUC = 0.776). Similarly, we found that GUSB halving times correlated better (r=0.58, Spearman's rank correlation) with month 18 IS levels considered as a continuous variable compared to both ABL1 halving time (r=0.47) and month 3 IS levels (r=0.51). Focusing on GUSB, 19/51 (37%) cases had a halving time shorter than the median for all cases (13.9 days) but were not in DMR at 18 months. 12 of these 19 cases also had lower than the median IS levels at 3 months (0.73%). Conversely 13/52 cases (25%) had a long halving time but achieved DMR at 18 months. 8/13 also had higher than median IS levels at 3 months. Of the 45 cases who achieved DMR at 18 months, 11 (24%) had IS levels at 3 months that were higher (median = 3.97%, range 0.77%-35.3%) than the median for all cases (0.73%). Only 1/11 had a shorter than median GUSB halving time. Conversely, 17/58 (29%) of cases who were not in DMR at 18 months had <0.73% IS at 3 months (median = 0.40; range 0.05-0.70); only 5/17 had long GUSB halving times. These data indicate that combining halving times with 3 month IS data adds limited additional predictive value. Since GUSB measurements remain largely unstandardized, we assessed the performance of the AcroMetrix™ BCR-ABL Panel (Thermo Fisher), recently calibrated to GUSB and BCR as well as ABL1, to standardize MRD results using GUSB as a reference gene in two centers. Established GUSB conversion factors (CF) derived by sample exchange were essentially indistinguishable from those derived using the panel [Lab 1: existing CF (0.99) / panel CF (1.02) = 0.97; Lab 2: existing CF (1.57) / panel CF (1.40) = 1.12, thus validating the panel to standardize GUSB measurements. We conclude that halving times over the first 3 months of asciminib treatment using GUSB or ABL1 as well as IS levels at 3 months are all predictive of molecular response at 18 months, but the predictive value of GUSB halving times are greatest. We therefore recommend that standardized baseline BCR::ABL1/GUSB measurements are incorporated into future studies to enable comprehensive assessment of the value of BCR::ABL1 halving times for routine management.