Energy expenditure profiles and the risk of early limiting toxicity in older patients with cancer: The ELCAPA-25 prospective cohort survey

Background & aims: Predicting the risk of early limiting toxicity (ELT) is major challenge for the clinician seeking an effective, safe treatment for older patients with cancer. The Cancer and Aging Research Group (CARG) and CRASH (Chemotherapy Risk Assessment Scale for High-Age Patients) toxicity scores were designed to predict chemotherapy-related toxicity. Elevated resting energy expenditure (REE) may predispose to cachexia and increase ELT and mortality in older patients with cancer. The primary objective was to assess the association between elevated REE and ELT in older patients with cancer. The secondary objectives were to assess the discriminant ability of a predictive model including REE (relative to the CARG and CRASH scores) and the prognostic value of elevated REE.

Methods: We assessed patients aged 70 or over included in the prospective ELCAPA cohort between 2014 and 2018. The inclusion criteria were a solid tumour, a measurement of REE at baseline (mREE, by indirect calorimetry), and a geriatric assessment prior to cancer treatment in a teaching hospital (Paris, France). The mREE was compared with the predicted REE (pREE), as defined by the Harris-Benedict equation. Depending on the mREE/pREE ratio, study participants were classified as hypermetabolic, hypometabolic or normometabolic. The primary endpoint was 3-month ELT, defined as any unplanned hospitalization or any event leading to dose reduction, a treatment delay of more than 7 days, or treatment discontinuation within 3 months of initiation. The secondary endpoint was the 3-month mortality rate.

Results: A total of 179 patients were included. The median age was 80 [interquartile range: 76-84] years, 37% of the patients were female, 81.8% had metastatic disease, 67.6% received chemotherapy, 20.7% received hormone therapy, and 11.7% received targeted therapies. According to the mREE/pREE ratio, 85 patients (47%) were hypermetabolic, 63 (35%) were normometabolic, and 31 (18%) were hypometabolic. Sixty patients (33.5%; 95% confidence interval (CI): 26.7-40.9) experienced ELT. The discriminant ability (as assessed by the C-index) of a multivariate model including REE and adjustment factors was 0.82 [95%CI: 0.73-0.91]. In comparison, the discriminant ability of the CARG and CRASH models was 0.57 [0.45-0.68] and 0.51 [0.40-0.62], respectively. In our model, hypermetabolism was an independent risk factor for ELT (adjusted odds ratio = 2.44; 95%CI: 1.02-5.80). Other risk factors were the cancer type and stage, the treatment protocol, a clinical diagnosis of depression, the presence of grade 3 or 4 comorbidities, and the serum lactate dehydrogenase level.

Conclusion: Hypermetabolism status is an independent predictor of ELT in older patients with cancer, relative to normometabolic status. Baseline REE measurement might improve the ELT risk assessment and decision-making process.