RESUMO
BACKGROUND: The 90-day BCR-ABL1 (breakpoint cluster region-Abelson 1) level has been one of the accepted milestones for predicting the molecular response in patients with chronic myeloid leukemia (CML). The rate of decline in BCR-ABL1 has been considered a better predictor of the response but has not been uniformly accepted. A paucity of evidence is available to predict the accuracy of the rate of decline in the Indian context. Therefore, we tested the accuracy of the rate of decline of BCR-ABL1 in predicting the molecular response compared with the single 90-day values in a retrospective cohort study of selected cancer centers in south India. METHODS AND MATERIALS: Patients with chronic-phase CML diagnosed from January 2013 to December 2018, the serial BCR-ABL1 levels were estimated at 0, 45, and 90 days, 6 months, and 1 year. Data on patient demographics, risk stratification assessed using the Sokal and EUTOS (European Treatment and Outcome Study) scores were extracted using a mobile-based data capture tool from the medical records of the enrolled patients. The halving time, determined by log reduction, was compared with the 90-day BCR-ABL1 values using the receiver operating characteristic curve for the major and complete molecular response at 6 months and 1 year as standards. Accuracy was determined from the area under the curve. The cutoff for the halving time was chosen to balance the sensitivity and specificity. RESULTS: The rate of decline had more predictive accuracy compared with the 90-day BCR-ABL1 values (area under the curve for rate of decline, 0.83; 90-day, 0.80). A halving time of < 20 days identified 95% of the patients who had achieved major molecular response at 12 months compared with 80% using the single 90-day BCR-ABL1 response. CONCLUSIONS: The halving time of BCR-ABL1 appears promising as a predictor of the outcomes for patients with CML.