BACKGROUND & AIMS: The Global Leadership Initiative on Malnutrition (GLIM) criteria have been recommended for the diagnosis of malnutrition. It requires that the patient meets at least one phenotypic criterion and at least one aetiological criterion. For the latter, the patient must either demonstrate reduced food intake or have evidence of systemic inflammation. As both are common in advanced cancer, the aim of the present study was to determine, in patients who met the GLIM phenotypical criteria, which GLIM aetiological criteria (reduced food intake or systemic inflammation) is most useful in predicting overall survival (OS). METHODS: Data from two cancer biobanks were combined. Inclusion criteria were: ≥18 years, advanced cancer (stage III or IV) and ability to provide written consent. Weight loss (WL) was selected as the phenotypic criterion of choice, as preliminary analysis demonstrated it to be a superior predictor of OS compared to body mass index. Malnutrition type 1 was defined as >
5% WL over 6 months and a C reactive protein (CRP) ≥3mg/l. Further analysis was performed with a CRP >
10mg/l cut-off. Malnutrition type 2 was defined as >
5% WL over 6 months and reduced food intake, as reported in the Patient Generated Subjective Global Assessment. Survival was assessed using Kaplan-Meir methodology, log-rank tests and cox proportional hazard models, with hazard ratios (HR) and confidence intervals (CI) reported. RESULTS: In total, 176 patients were studied, with 147 events observed. The 3-month mortality rate was 32.4% (CI: 25.1 to 39.0) and the 1-year mortality rate was 71.8% (CI: 63.8 to 78.0). Malnutrition type 1 and malnutrition type 2 were observed in 37.8% (HR: 2.27 [CI: 1.54 to 3.33], p<
0.001) and 26.3% (HR: 1.74 [CI: 1.19 to 2.54], p=0.005) of patients respectively, with both significantly increasing the risk of death. Following adjustment for relevant confounders both malnutrition type 1 (HR: 1.92 [CI: 1.25 to 2.94], p=0.003) and malnutrition type 2 (HR: 1.61 [CI: 1.09 to 2.38], p=0.019) remained significant predictors of OS. Median survival for patients with malnutrition type 1 was 2.14 (CI: 1.74 to 4.90) months compared to 9.5 (6.94 to 13.64) months for those without (p<
0.001). For malnutrition type 2, this was 2.37 (CI: 1.64 to 5.46) vs. 7.40 months (CI: 6.08 to 10.16), p=0.004. When the CRP threshold was increased to >
10mg/l, malnutrition type 1 was observed in fewer patients (30.4%), median survival was shorter (1.91 [CI: 1.25 to 2.99] vs. 9.86 months [CI: 7.27 to 14.7], p<
0.001) and in both univariable (HR: 2.91 [CI: 1.94 to 4.63], p<
0.001) and multivariable (HR: 2.32 [CI: 1.50 to 3.60], p <
0.001) analyses, the risk of death increased. CONCLUSION: The results suggest that the inflammatory component of GLIM appears superior compared to reduced intake in predicting OS and notably, a higher CRP threshold correlates with shorter OS. Therefore, whilst GLIM has multiple potential combinations, all treated with equal regard, these data suggest that the inflammatory aetiological component, should be hierarchical to others.