Risk Prediction Model for Taxane-Induced Peripheral Neuropathy in Early-Stage Cancer

Importance: Taxane-induced peripheral neuropathy (TIPN) affects quality of life and ability to complete cancer treatment and has limited effective interventions for prevention and treatment. Objective: To develop and validate a TIPN risk prediction model. Design, Setting, and Participants: SWOG S1714 was a prospective observational cohort study conducted at sites in the National Cancer Institute National Community Oncology Research Program between March 1, 2019, and November 15, 2021, with 3 years of follow-up. The study included evaluable participants 18 years or older with stage I to III lung, breast, or ovarian, fallopian tube, or primary peritoneal cancer who were starting taxane-based treatment. Statistical analysis was conducted from December 2023 to June 2024. Exposures: Taxane-based regimens including paclitaxel or docetaxel. Main Outcomes and Measures: The primary end point was occurrence of TIPN by 24 weeks. TIPN was assessed using the patient-reported European Organization of Research and Treatment of Cancer Quality of Life Questionnaire-Chemotherapy-Induced Peripheral Neuropathy 20-item scale (CIPN-20) at baseline and weeks 4, 8, 12, and 24. Occurrence of TIPN was defined as an increase of 8 points or more over baseline in the CIPN-20 sensory subscale score. With a 60% random sample of evaluable participants, best-subset selection using logistic regression and k-fold cross-validation identified a best model based on demographic factors, baseline comorbid conditions, and treatment factors. Adverse risk factors were summed, generating a score, split at the median and tested in the remaining 40% of evaluable participants. The target difference was 12% between high-risk vs low-risk groups. Results: A total of 1336 participants enrolled in S1714. Of 1278 evaluable participants (median age, 55.0 years [range, 23.0-84.0 years]; 1264 women [98.9%]; 1164 with breast cancer [91.1%]), 804 (62.9%) experienced TIPN by week 24. Using the training set of 768 participants, a risk prediction model for TIPN was developed that included 5 adverse risk factors: receipt of paclitaxel; stage II or III disease; planned taxane duration of more than 12 weeks; diabetes, autoimmune disease, moderate kidney disease, or a neurologic condition; and self-identified race and ethnicity (Black, Hispanic, Native American, Pacific Islander, multiple races, or unknown race or ethnicity). In the test set of 510 participants, TIPN was more common in high-risk (235 of 345 [68.1%]) vs low-risk (84 of 165 [50.9%]) groups (absolute difference, 17.2%), exceeding the 12% target. Conclusions and Relevance: In this cohort study of participants with early-stage cancer receiving a taxane regimen, a set of baseline risk factors stratified TIPN risk. A risk prediction model may guide treatment decision-making, symptom monitoring, and enrollment in interventional trials for TIPN prevention and treatment.