Construction Of Prognostic Model Based On GLIM Criteria In Cancer Patients

Purpose:Malnutrition is a highly prevalent and neglected syndrome in cancer patients,and inflammatory load is an important factor contributing to malnutrition.Malnutrition in cancer patients increases the incidence of adverse effects of antitumor therapy,affects patients’ quality of life and long-term prognosis.Accurate identification and proper assessment of malnutrition is essential to improve clinical outcomes in patients with cancer.The Global Leadership Initiative in Malnutrition(GLIM)standard is currently the most recognized method of malnutrition detection,but single-dimensional assessment is often difficult to accurately predict patient prognosis.In this study,the state of malnutrition was identified based on GLIM criteria,and a model combined with serological indexes was constructed to predict the prognosis of Chinese cancer patients.Methods:We collected 2146 patients with pathologically confirmed cancer with basic information,anthropometric indicators,anthropometric data and serological indicators.Patients were randomly divided into a training cohort(n=1502)and a validation cohort(n=644)according to 7:3.In the training cohort,the optimal cut-off values of relevant indicators were defined by time-dependent receiver operating characteristic(ROC)curves,and the predictive effect of nutritional markers on overall survival(OS)was analyzed by univariate and multivariate Cox regression.Prognostic prediction models were constructed based on nutritional markers,and model validity was verified in a validation cohort.Kaplan-Meier survival analysis was used to analyze the relationship between each model and patients’ survival.Subanalysis of specific tumor types was conducted by Cox regression.Results:1.In the training cohort,univariate and multivariate Cox regression analyses showed that neutrophil-to-lymphocyte ratio(NLR)(HR=1.768,95% CI: 1.378～2.267,P<0.001),albumin(HR=1.578,95% CI: 1.190～2.092,P=0.002),and malnutrition diagnosed by GLIM criteria(HR=1.601,95% CI: 1.223 ～ 2.095,P=0.001)were independent risk factors for OS in patients with cancer.2.In the training cohort,based on tumor type,tumor stage,NLR,albumin,and malnutrition diagnosed by GLIM criteria,these indicators were used to construct the prognostic model and a nomogram was established.The C index of the model was 0.786,and AUC for predicting 1-year,3-year,and 5-year survival rates were 0.83,0.86,and 0.90,respectively.The calibration curves in the validation cohort showed strong agreement between the survival rates predicted by the model and the actual survival rates.3.Eight nutritional models were formed by the combination of the three indicators of NLR,albumin,and malnutrition diagnosed by GLIM criteria,and the eight models represented a statistically significant difference in OS.Among them,model 8(NLR+,albumin+,GLIM+)represented the shortest OS(HR=4.288,95% CI: 3.117～5.899,P<0.001).Conclusions:1.NLR,albumin,and malnutrition diagnosed by GLIM criteria are prognostic nutritional markers for cancer patients.2.A prognostic model constructed based on tumor type,tumor stage,NLR,albumin,and malnutrition diagnosed by GLIM criteria can accurately predict the prognosis of tumor patients.3.A simple inflammationnutrition model based on nutritional status defined by GLIM criteria,Alb,and NLR can rapidly determine the prognosis.4.Patients who meet low albumin level,high inflammatory level,and malnutrition have the worst prognosis,short overall survival,and high risk of death.