Association of Paclitaxel-Induced Peripheral Neuropathy with ABCB1 Gene Single Nucleotide Polymorphisms in Patients with Breast Cancer

Abstract

Paclitaxel-induced peripheral neuropathy is a serious adverse event in Egyptian breast cancer patients which negatively impacts their performance status and quality of life and frequently requires medical intervention. Paclitaxel-induced peripheral neuropathy is difficult to predict and treat and patients usually require dose reductions. Single nucleotide polymorphisms in ABCB1 are frequently associated with increased severity of PIPN. However, the association and the validity of ABCB1 SNP markers to predict the incidence of PIPN are unclear. In this study, four logistic regression models including phenotypes and ABCB1 genotypes of the G1236A and the G3435A SNPs evaluated in four genetic models -additive, dominant, overdominant, and recessive- were fitted. A machine learning algorithm -repeated k-fold cross-validation- was applied on the four models to select the model with the highest prediction accuracy -denoting the highest internal validity-. Finally, model diagnostics and receiver-operating characteristics analysis were performed for the model with the highest classification accuracy, which was the additive model. Model diagnostics demonstrated convenient prediction accuracy, sensitivity, and specificity. In the additive model, the PIPN grade was higher in patients with increased BSA, with pre-existing diabetes before paclitaxel therapy, and in patients who had the ABCB1 G1236A variant AA. The severity of PIPN was independent of patients’ age, menopausal status, and BMI.