1884P - Development and validation of a prediction tool for severe treatment-related toxicities in older cancer patients on systemic treatment (TR-TRM)

Background

Older cancer patients are at a higher risk of treatment-related toxicities (TRT) due to various reasons. Existing chemotherapy toxicity calculators have limitations. This study aimed to develop and validate a prediction tool for severe TRT in older cancer patients on systemic treatment.

Methods

Consecutive cancer patients aged ≥65 undergoing systemic anti-cancer treatment were recruited from three large oncology centers in Hong Kong (Queen Mary Hospital, Queen Elizabeth Hospital and Gleneagles Hospital) from March 2019 to May 2022. Pre-treatment assessments captured clinical, tumour/treatment, laboratory, and geriatric assessment variables. Patients were followed throughout treatment or 6 months for grade 3-5 TRT based on CTCAE version 5. Univariate and multivariable logistic regression identified predictive factors, and a weighted scoring system was used to develop the prediction model. Model performance was evaluated using area under the ROC curve (AUC) and goodness-of-fit statistics, with internal and external validation.

Results

Among the 500 patients (400 in development cohort, 100 in validation cohort) with median age 71, 304 (60.8%) developed grade 3-5 TRT. Ten independent predictors associated with TRT were identified. The predictive model categorized patients into low (0-3 points: incidence of TRT in development cohort: 35.3%, validation cohort: 29.0%), intermediate (4-8 points: 59.7% and 61.7%), and high-risk (9-26 points: 82.8% vs. 77.3%) groups. The AUC was 0.718 (95% CI: 0.667-0.769) in the development cohort and 0.717 (95% CI:0.616-0.818) in the validation cohort. Table: 1884P

Treatment-related toxicity risk model (TR-TRM)