Predicting chemotherapy toxicity in older adults: Comparing the predictive value of the CARG Toxicity Score with oncologists' estimates of toxicity based on clinical judgement.

Abstract

The Cancer and Aging Research Group's (CARG) Toxicity Score was designed to predict grade ≥3 chemotherapy-related toxicity in adults aged ≥65 yrs. commencing chemotherapy for a solid organ cancer. We aimed to evaluate the CARG Score and compare it to oncologists' estimates for predicting severe chemotherapy toxicity in older adults. Patients aged ≥65 yrs. starting chemotherapy for a solid organ cancer had their CARG Score (range 0-23) calculated. Their treating oncologist, blinded to these results, independently estimated each patient's risk of severe chemotherapy toxicity (0-100%). Toxicities were captured prospectively. The predictive value of the CARG Score and oncologists' estimates was estimated using logistic regression and in terms of Area Under the Receiver Operating Characteristic curve (AU-ROC). 126 patients from ten oncologists at two sites participated. The median age was 72 yrs. (range 65-84). The median CARG Score was 7 (range 0-17); the median oncologist estimate of risk was 30% (range 3-80%), and these measures were not correlated (r = -0.01). 64 patients (52%) experienced grade ≥ 3 toxicity. Rates of severe toxicity in low-, intermediate-, and high-risk groups by CARG Score were 58%, 47%, and 58% respectively, and 63%, 44%, and 67% by oncologist estimate. Severe chemotherapy toxicity was not predicted by the CARG Score (OR 1.04, 95%CI 0.92-1.18, p = .54, AU-ROC 0.52), or oncologists' estimates (OR 1.00, 95%CI 0.98-1.02, p = .82, AU-ROC 0.52). Neither the CARG Score, nor oncologists' estimates based on clinical judgement, predicted severe chemotherapy-related toxicity in our population of older adults with cancer. Methods to improve risk prediction are needed.