Validation of the Cancer and Aging Research Group (CARG) model to identify risks of chemotherapy toxicities in Asian older adults.

Abstract

12043 Background: Older adults are at an increased risk of chemotherapy related toxicities. Identification of risk factors can facilitate oncologists in tailoring treatment, potentially mitigate the risk of chemotherapy toxicity and improve social economic outcomes. Tools like the Karnofsky Performance Status (KPS) and Geriatric Assessment (GA) are used in clinical practice to guide treatment decisions for older adults with cancer. The CARG model incorporating the GA and clinical variables was developed and validated to predict for grade 3-5 toxicities, although this has not yet been validated in Asians. Methods: Patients ≥70 years old with solid malignancies receiving chemotherapy at the National University Cancer Institute, Singapore were recruited between June 2017 and January 2018. The study aims to verify the application of the CARG model, KPS, GA and oncologists’ estimate of toxicity in a multi-ethnic Asian population. The risks of chemotherapy toxicity were calculated using the CARG model (low risk: 0-5, medium risk: 6-9, high risk: ≥10). A GA including a physician rated KPS score (low risk: 90-100, medium risk:80, high risk: ≤70) and a timed up and go (TUG) test (low risk: ≤12s, high risk: > 12s) were performed for all patients. The attending oncologist (blinded to the CARG score) was asked to give an estimated likelihood (low/medium/high) of chemotoxicity. Chemotherapy related toxicities were captured by chart review based on the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. The area under the receiver operating characteristic (ROC) curve and a univariate analysis of predictors of toxicity were calculated. Results: A total of 200 patients were included. Median age was 74 (range: 70-89). The ethnic makeup of the population consisted of Chinese (N = 177; 88.5%), Malay (N = 17; 8.5%), Indians (N = 4; 2%) and other ethnicities. Majority of patients were males (N = 110; 55%), had metastatic disease at diagnosis (N = 114; 57%) and had gastrointestinal cancers (N = 80; 40%). 75% (N = 150) had polychemotherapy and 55% (N = 110) received full dose chemotherapy. More than half of patients (68.5%) experienced G3-5 toxicities, with 65.5% requiring hospitalization. The CARG model predicted 22.5% of patients as low, 50.5% as medium and 27% as high risk of chemotoxicity. The area under ROC for the CARG model was 0.74 (95% CI, 0.67-0.82). A one-point increase in the CARG score was associated with a 15% increase in the odds of G3-5 toxicities. 84.5% of patients needed > 12s for the TUG test. The TUG score, KPS and oncologist’s prediction of toxicity was predictive of an increased risk of G3-5 toxicities (p < 0.01). Conclusions: This study confirms the validity of the CARG predictive model in Asians and can streamline healthcare delivery in older adults. Development of a more robust predictive model incorporating the KPS and TUG into the CARG model could be considered.