Comparison of the prognostic value of eight nutrition-related tools in older patients with cancer: A prospective study

Objectives The primary objective of the present study was to evaluate and compare the ability of eight nutrition-related tools to predict 1-year mortality in older patients with cancer. Design, setting and participants We studied older patients with cancer from the ELCAPA cohort and who had been referred for a geriatric assessment at one of 14 participating geriatric oncology clinics in the greater Paris area of France between 2007 and 2018. Measurements The studied nutrition-related tools/markers were the body mass index (BMI), weight loss (WL) in the previous 6 months, the Mini Nutritional Assessment, the Geriatric Nutritional Risk Index (GNRI), the Prognostic Nutritional Index, the Glasgow Prognostic Score (GPS), the modified GPS, and the C-reactive protein/albumin ratio. Results A total of 1361 patients (median age: 81; males: 51%; metastatic cancer: 49%) were included in the analysis. Most of the tools showed a progressively increase in the mortality risk as the nutrition-related risk category worsened (overall p-values <0.02 for all) after adjustment for age, outpatient status, functional status, severe comorbidities, cognition, mood, cancer treatment strategy, tumour site, and tumour metastasis. All the models were discriminant, with a C-index ranging from 0.748 (for the BMI) to 0.762 (for the GPS). The concordance probability estimate ranged from 0.764 (WL) to 0.773 (GNRI and GPS)). Conclusion After adjustment for relevant prognostic factors, all eight nutrition-related tools/markers were independently associated with 1-year mortality in older patients with cancer. Depending on the time or context of the GA, physicians do not always have the time or means to perform and assess all the tools/markers compared here. However, even when some information is missing, each nutritional tool/marker has prognostic value and can be used in the evaluation.