Original Research
Sunday, October 13, 2024
12:45 PM–02:15 PM
Abstract
Introduction:
Approximately 30% of patients receiving tyrosine kinase inhibitors (TKI) might occur QTc prolongation.
Research Question or Hypothesis: What are the best set of factors used to prediction the risk probability of QTc prolongation for cancer patients treated with oral TKIs in clinical practice?
Study Design: The retrospective cohort study was conducted using the data retrieved from electronic medical records (EMR) for cancer patients newly treated with oral TKIs in a tertiary medical center in Taiwan.
Methods: The occurrence of QTC prolongation was defined as = 450 millisecond for male and = 470 millisecond for female using Bazett's formula. Other than performing the statistical (backward logistic regression [LR]) and supervised machine learning (ML) approaches to identify the candidate factors and further to train the best prediction models, the standardized multivariate LR and Receiver Operating Characteristics Curves analysis were performed to explore the impact of the identified factors and the cut-off points of risk probability prediction.
Results: The statistical and ML approaches identified the two different sets of 12 factors in the corresponding best models, where the statistically driven model showed excellent model performance and fitting. The cut-off prediction of high-risk probability were around 0.4. Given approximately 0.13 chance occurred QTc prolongation without considering the other factors, the two most important factors were prolongation in the baseline and diagnosed with the other cardiovascular diseases (excluding arrythmia, cardiomyopathy or so) in the two best models. Although the ranking of risk probability predictions due to the other individual factors were various, the factors in the 12-parameter statistically driven model revealed better clinical meaning.
Conclusion: The identified statistically driven model with 12 easily-accessible variables from EMR performed better than the other ML-driven models. Those patients with prior experience of QTc prolongation and existing cardiovascular diseases should be monitored intensively to prevent QTc prolongation for cancer survivors newly treated with oral TKIs in the future.
Presenting Author
Hsiang-Wen Lin RPh., M.S., Ph.D.China Medical University
Authors
Yu-Chieh Chen RPh., MS, PhD
Chien-Ning Hsu RPh., M.S., Ph.D.
Kaohsiung Medical University Hospital
Chen-Yang Lin M.D., Ph.D.,
China Medical University
Tien-Chao Lin RPh., M.S.
China Medical University Hospital
Liang-Chih Liu MD, PhD
Division of Breast Surgery, China Medical University Hospital
Tzu-Pei Yeh MS, PhD
School of Nursing and Graduate Institute, College of Health Care, China Medical University