P685: late phase bcr-abl decline predicts the duration of tki (imatinib) for successful treatment free remission in chronic myeloid leukemia.

Background: Goal of treatment in chronic myeloid leukemia is to achieve TKI (tyrosine Kinase Inhibitor) free remission (TFR); however nearly half of the patients relapse post discontinuation of TKI. Across all TKI-discontinuation studies duration of TKI has consistently been a predictor of TFR. Additionally, in mathematical models of clinical trial data, late phase BCR-ABL kinetics is a predictor of successful TFR as well. Aims: In this study, we sought to predict the duration of TKI for successful TFR based on late phase BCR-ABL decline (halving-time). Methods: CML patients who were in stable MMR4.5 for at-least two years and atleast three years of TKI and discontinued TKI were analyzed for the predictors of sustained TFR (no molecular relapse (MR3.0) by twelve months). Half live of BCR-ABL was calculated using the formula =ln(2)*x/ln(y/z). Number of half-lives of TKI treatment was calculated by dividing total of duration of TKI from the half-life. Univariate and multivariate analysis was done using cox proportional hazards regression to identify the predictive factors. Decision tree analysis was done to categorize the patients as low, intermediate and high risk based on number of half-lives of TKI treatment. One-year mRFS rates between the groups were compared by Kruskall-Wallis test. Median mRFS was calculated by using Kaplan-Meir method. Censoring was done at twelve months. Results were validated by five-fold cross validation. Results: Median age of the patients were 33yrs (IQR:28-41). 35.2%, 34.4% and 30.4% of the patients were in ELTS low, intermediate and high-risk category. Median time to achieve cytogenetic remission (CCyR) was 14 months (IQR:8-13); Median duration of TKI was 81 months (IQR: 67-116). Median late phase BCR-ABL halving time was 219 days (IQR: 139-311) and median number of half-lives of TKI was 13.2(IQR:8.0-19.3). Median follow up after discontinuing TKI was 13.5 months (IQR:6.0-36.0). One year relapse rate was 40.4%. In the univariate analysis time to CCyR (HR:1.18, CI: 1.1-1.25, p=0.001), duration of TKI (HR: 0.98, CI: 0.89-0.99, p=0.001), number of half-lives (HR: 0.72, CI: 0.65-0.78, p=0.001), and halving time (HR:1.1, CI: 1.04-1.17, p=0.001) were statistically significant. In the multivariate model; time to CCyR and number of half-lives were included because of multicollinearity between halving time, TKI duration and number of half-lives Number of half-lives was a significant predictor of successful TFR (HR:0.72, CI: 0.65-0.79, p=0.001). HR for time to CCyR was 1.055 (CI: 0.98-1.12, p= 1.0). Partitioning analysis that included time to CCyR and number of half-lives classified patients based on number of half-lives into High (<9, n=37), intermediate (9-13.2, n=27) and low risk category (>13.2 n=62). One-year mRFS rate for high, intermediate and low risk category was 3%, 52% and 68% (p=0.0001). mRFS for high, intermediate were 4 months and 23 months respectively. No relapses were seen in low-risk group (log rank p=0.00001). Error rate calculated for the model was 14% and the average error rate of five-fold cross validation was 10%. Summary/Conclusion: Late phase BCR-ABL kinetics is a strong predictor of successful TFR in CML. TKI duration based on half-live of BCR-ABL can be used to improve the probability of TFR in CML. However, our findings need to be validated in prospectively conducted studies. Figure-1. K-M curves for 12 month molecular relapse free survival (mRFS)Keywords: BCR-ABL, treatment-free remission, Imatinib, Chronic myeloid leukemia