Prediction for sustained deep molecular response of BCR-ABL1 levels in patients with chronic myeloid leukemia in chronic phase

BACKGROUND: The achievement of a sustained deep molecular response is a goal of increasing relevance because it opens the possibility of treatment discontinuation. The objective of this analysis was to develop a prediction model for a sustained molecular response with BCR-ABL1 level <0.0032% on the international scale (MR ^4.5 ) for at least 2 years according to BCR-ABL1 levels achieved within the first 12 months of therapy. METHODS: Data for 603 patients with newly diagnosed chronic myeloid leukemia in chronic phase in consecutive prospective clinical trials were analyzed. The best fit average molecular response was defined by robust linear regression models, with which the average molecular levels were defined. The minimum acceptable molecular response was defined by quantile regression for the 95th percentile, with which the worst 5% BCR-ABL1 levels were identified. RESULTS: In 603 patients with a median follow-up of 103 months, 2002 BCR-ABL1–level data points within 1 year of tyrosine kinase inhibitors were identified. The regression equation for the best fit average levels for a sustained MR ^4.5 was Log ₁₀ (PCR) = −0.1424 × (Months) – 0.8668, and the regression equation for minimum acceptable levels was Log ₁₀ (PCR) = −0.1403 × (Months) + 0.6142 (where PCR indicates polymerase chain reaction). To achieve a sustained MR ^4.5 , the best fit average levels were 0.051%, 0.019%, 0.007%, and 0.003% at 3, 6, 9, and 12 months, respectively; the minimum acceptable levels were 1.561%, 0.592%, 0.225%, and 0.085% at 3, 6, 9, and 12 months, respectively. CONCLUSIONS: This model proposes optimal values that predict the highest probability of reaching such a goal. These values can be used to guide therapy when a sustained MR ^4.5 is the objective. Cancer 2018;124:1160-8.