Prediction of moderate and severe toxicities of chemotherapy in older patients with cancer: a propensity weighted analysis of ELCAPA cohort.

Abstract

Currently available predictive models for chemotherapy-related toxicity are not sufficiently discriminative in older patients with cancer and do not consider moderate toxicities. The purpose of this study was to identify factors associated with moderate and severe chemotherapy toxicities in older patients with cancer. Patients aged 70+ recruited in the prospective ELCAPA cohort were analyzed. A total of 837 patients with data on toxicities had received chemotherapy without other systemic treatment and were included between 2015 and 2022. To adjust for any imbalances in the distribution of covariates between patients receiving single-agent chemotherapy vs combination chemotherapy, we applied overlap weighting (a propensity-score-based technique). We used multinomial logistic regression. Median (interquartile range) age was 81 (77-84). Forty-one percent experienced moderate toxicity, and 33% experienced severe toxicity. Hematologic toxicities accounted for 53% of severe toxicities and 66% of moderate toxicities. Age <80 years, cancer type, metastatic status, Eastern Cooperative Oncology Group performance status (ECOG-PS) >1, no cognitive impairment were associated with combination chemotherapy decision. In a univariate analysis with overlap weighting, no factors were associated with moderate toxicity. Hemoglobin < 0g/dL and a CIRS-G score >12 were associated with severe toxicity. In a multivariate analysis, only hemoglobin < 10g/dL was independently associated with severe toxicity, adjusted OR 2.96 (95% CI, 1.20-7.29). By addressing indication bias for combination chemotherapy decision, only anemia and not cancer type, combination chemotherapy was predicting for severe chemotherapy-related toxicity in older patients with cancer. We did not find any predictors of moderate chemotherapy-related toxicity.