PG-SGA SF in nutrition assessment and survival prediction for elderly patients with cancer

Background: This study was sought to report the prevalence of malnutrition in elderly patients with cancer. Validate the predictive value of the nutritional assessment tool (Patient-Generated Subjective Global Assessment Short Form, PG-SGA SF) for clinical outcomes and assist the therapeutic decision. Methods: This is a secondary analysis of a multicentric, observational cohort study. Elderly patients with cancer older than 65 years were enrolled after the first admission. Nutritional status was identified using the PG-SGA SF. Results: Of the 2724 elderly patients included in the analysis, 65.27% of patients were male (n = 1778); the mean age was 71.00 ± 5.36 years. 31.5% of patients were considered malnourished according to PG-SGA SF. In multivariate analysis, malnutrition(PG-SGA SF > 5) was significantly associated with worse OS (HR: 1.47,95%CI:1.29–1.68), affects the quality of life, and was related to more frequent nutrition impact symptoms. During a median follow-up of 4.5 years, 1176 death occurred. The mortality risk was 41.10% for malnutrition during the first 12 months and led to a rate of 323.98 events per-1000-patient-years. All nutritional assessment tools were correlated with each other (PG-SGA SF vs. PG-SGA: r = 0.98; PG-SGA SF vs. GLIM[Global Leadership Initiative on Malnutrition]: r = 0.48, all P < 0.05). PG-SGA SF and PG-SGA performed similarly to predict mortality but better than GLIM. PG-SGA SF improves the predictive ability of the TNM classification system for mortality in elderly patients with cancer, including distinguishing patients’ prognoses and directing immunotherapy. Conclusions: The nutritional status as measured by PG-SGA SF which is a prognostic factor for OS in elderly cancer patients and could improve the prognostic model of TNM.