P721: PREDICTIVE SCORING SYSTEMS FOR MOLECULAR RESPONSES IN PERSONS WITH CHRONIC MYELOID LEUKEMIA RECEIVING INITIAL IMATINIB THERAPY
X. Zhang,Zhijun Li,M. Zhang,P. G. Robert,Xiao‐Jun Huang,Qian Jiang
DOI: https://doi.org/10.1097/01.hs9.0000845768.41886.03
2022-01-01
HemaSphere
Abstract:Background: Nowadays more patients have benefited from tyrosine kinase inhibitors (TKIs) therapy to achieve a stable major molecular response (MMR) and even deep molecular responses (DMR, at least MR4), which prevented disease progression and allows for pursuing treatment-free remission (TFR). However, there were still lacking of robust prediction models for molecular responses including MMR or DMR to TKI-therapy. Aims: Develop and validate predictive scoring systems for MMR and MR4 in persons with newly-diagnosed chronic phase chronic myeloid leukemia (CML) receiving imatinib. Methods: Data from 1364 consecutive subjects were randomly-assigned 2:1 to training (n=909) and validation (n=455) datasets. Co-variates significantly associated with MMR and MR4 in multi-variable analyses in the training dataset were used to develop a predictive scoring system based on standardized regression coefficients (beta [β]) and the model tested in the validation dataset. Time-dependent area under the receiver-operator characteristic curves (AUROC), calibration plots and decision curve analyses (DCAs) were used to evaluate predictive accuracy. Results: In the training dataset, male sex (β=-0.350, HR=0.7 [95%CI: 0.6, 0.8]), WBC ≥ 130 ×10E+9/L (β=-0.538, HR=0.6 [0.5, 0.7]), hemoglobin < 110 g/L (β=-0.490, HR=0.6 [0.5, 0.8]) and EUTOS long-term survival (ELTS) risk score (intermediate-risk: β=-0.464, HR=0.6 [0.5, 0.8]; high-risk: β=-0.743, HR =0.5 [0.3, 0.7]) were significantly-associated with lower cumulative incidence of MMR. Each co-variate was assigned 1 point in the predictive scoring system for MMR except WBC ≥ 130 ×10E+9/L and ELTS high-risk assigned 2 points. Using the predictive scoring system, subjects were classified into (1) low- (score=0-1; n=429, 47%), (2) intermediate- (score=2-4; n=340, 38%); and (3) high- (score=5-6; n=140, 15%) risk subgroups. 5-year cumulative incidences of MMR were 86% (84, 89%), 61% (55, 65%) and 34% (27, 43%) (p for trend <0.001; Figure A). Comparable 5-year cumulative incidences in validation dataset were 86% (82, 90%), 69% (64, 74%) and 31% (22, 40%) (p for trend<0.001; Figure B). AUROC values were 0.77 and 0.74 in the training and validation datasets. Similar analyses for MR4 in training dataset identified male sex (β=-0.282, HR=0.8 [0.6, 0.9]), WBC ≥ 120 ×10E+9/L (β=-0.747, HR=0.5 [0.4, 0.6]), hemoglobin<110 g/L (β=-0.496, HR=0.6 [0.5, 0.8]) and ELTS risk score (intermediate-risk: β=-0.509, HR=0.6 [0.4, 0.8]; high-risk: β=-1.212, HR=0.3 [0.1, 0.6]) were significantly-associated with lower cumulative incidence of MR4. Male sex was assigned 1 point; ELTS intermediate-risk and hemoglobin < 110 g/L, 2 points; WBC ≥ 120 ×10E+9/L, 3 points; and ELTS high-risk, 4 points in the predictive scoring system for MR4. Subjects in the training dataset were classified into (1) low- (score=0-2; n=416, 46%), (2) intermediate- (score=3-7; n=348, 38%); and (3) high- (score=8-10; n=145, 16%) risk cohorts. 5-year cumulative incidences of MR4 were 71% (70, 73%), 35% (31, 39%) and 16% (10, 22%) (p for trend<0.001, Figure C). Similar 5-year cumulative incidences in validation dataset were 66% (61, 71%), 45% (38, 52%) and 21% (12, 30%) (p for trend<0.001, Figure D). AUROC values were 0.76 and 0.72. Calibration plots and DCAs showed good performance for predicting MMR and MR4. Image:Summary/Conclusion: We developed the predictive scoring systems for MMR and MR4 in persons with chronic phase CML receiving imatinib. These data may help physicians make an appropriate therapy decision.