Construction and preliminary verification of a nomogram for survival of TKI-treated adult patients with newly diagnosed chronic myelogenous leukemia in the chronic phase
10.3760/cma.j.cn115356-20240814-00128
- VernacularTitle:TKI治疗的初诊慢性期慢性粒细胞白血病成年患者生存列线图构建及初步验证
- Author:
Guangling HU
1
;
Haiping LIANG
;
Xuehan MA
;
Bei LIU
Author Information
1. 兰州大学第一临床医学院 兰州大学第一医院血液内科，兰州 730000
- Keywords:
Leukemia, myeloid, chronic-phase;
Tyrosine kinase inhibitor;
Prognosis;
Nomogram
- From:
Journal of Leukemia & Lymphoma
2025;34(1):16-23
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To investigate the related factors affecting the survival of adult patients with newly diagnosed chronic myeloid leukemia in the chronic phase (CML-CP) treated with tyrosine kinase inhibitor (TKI) and the prognostic predictive effect of the nomogram constructed according to them.Methods:A retrospective cohort study was conducted. Clinical general information and laboratory index data of 243 newly diagnosed adult CML-CP patients treated with TKI in the First Hospital of Lanzhou University from December 2008 to June 2023 were collected, and they were divided into a training set (194 patients) and a validation set (49 patients) by complete randomization in the ratio of 8∶2. In the training set, variables affecting poor overall survival (OS) of patients were analyzed using univariate and multivariate Cox proportional hazards models by R4.3.2 software to obtain the independent influences of poor OS, on the basis of which the Cox regression model was constructed and the nomogram predicting the OS rate at 8 and 10 years was plotted. Kaplan-Meier method was applied to analyze the OS in all 243 patients and patients stratified by the screened independent influencing factors for poor OS, and log-rank test was used for comparison between the groups. In the training and validation sets, receiver operating characteristic (ROC) curve was used to analyze the effect of the nomogram on predicting 8- and 10-year OS rates of patients with actual survival as the gold standard; calibration curve was used to assess the accuracy of predictions of the nomogram; decision curve analysis (DCA) was used to assess the clinical utility of the nomogram.Results:The median age of 243 CML-CP patients [ M ( Q1, Q3)] was 46 (35, 58) years old, 150 cases (61.7%) were male and 9 cases (38.3%) were female, 119 cases (49.0%) had comorbidities, and the efficacy of 82 cases (33.7%) reached molecular response (MR) 5.0. Differences in patient compositions for age and gender, levels of major indicators for peripheral blood and bone marrow, spleen size, comorbidities, short-term efficacy, Sokal score, and long-term survival score of European Treatment and Outcomes Study between the training and validation sets were not statistically significant (all P > 0.05). Multivariate Cox regression analysis showed that patients with elevated age ( HR = 1.04, 95% CI: 1.01-1.08, P = 0.041), comorbidities (with vs. without, HR = 3.48, 95% CI: 1.23-9.86, P = 0.019), and those who did not achieve MR5.0 (achieved vs. unachieved, HR = 0.13, 95% CI: 0.02-0.97, P = 0.046) were independent risk factors for poor OS in TKI-treated newly diagnosed adult CML-CP patients. By the last follow-up (December 2023), the median follow-up was 72 months, with the range of 6-180 months. Kaplan-Meier method analysis showed that the 8- and 10-year OS rates of 243 patients were 83.7% and 81.6%, respectively; patients with age ≥46 years compared to <46 years, with comorbidities compared to without comorbidities, and who did not achieve MR5.0 in terms of efficacy compared to who achieved MR5.0, the OS was poorer, and the differences were statistically significant (all P < 0.01). The nomogram of 8- and 10-year OS rates in TKI-treated newly diagnosed adult CML-CP patients was constructed based on the screened independent influencing factors of poor OS. The area under the ROC curve was 0.910 and 0.851 in the training set and 0.778 and 0.764 in the validation set for the predicted 8- and 10-year OS rates based on the nomogram, respectively, and the calibration curve showed that the predicted 8- and 10-year OS rates based on the nomogram were in high agreement with the actual ones in the training and validation sets; the DCA showed that the nomogram within a certain prediction threshold could benefit the clinical decision-making in both the training and validation sets. Conclusions:Having comorbidities, not reaching MR5.0 in efficacy and old age are independent risk factors for poor survival of TKI-treated adult patients with newly diagnosed CML-CP, and the nomogram constructed based on these 3 factors has a good predictive ability for the survival of such patients.
