BackgroundLipid management in clinic is critical to the prevention and treatment of Chronic kidney disease (CKD), while the manifestations of lipid indicators vary in types and have flexible association with CKD prognosis.PurposeExplore the associations between the widely used indicators of lipid metabolism and their distribution in clinic and CKD prognosis; provide a reference for lipid management and inform treatment decisions for patients with non-dialysis CKD stage 3–5.MethodsThis is a retrospective cohort study utilizing the Self-Management Program for Patients with Chronic Kidney Disease Cohort (SMP-CKD) database of 794 individuals with CKD stages 3–5. It covers demographic data, clinical diagnosis and medical history collection, laboratory results, circulating lipid profiles and lipid distribution assessments. Primary endpoint was defined as a composite outcome(the initiation of chronic dialysis or renal transplantation, sustained decline of 40% or more in estimated glomerular filtration rate (eGFR), doubled of serum creatinine (SCr) from the baseline, eGFR less than 5 mL/min/1.73m2, or all-cause mortality). Exposure variables were circulating lipid profiles and lipid distribution measurements. Association were assessed using Relative risks (RRs) (95% confidence intervals (CIs)) computed by multivariate Poisson models combined with least absolute shrinkage and selection operator (LASSO) regression according to categories of lipid manifestations. The best model was selected via akaike information criterion (AIC), area under curve (AUC), receiver operating characteristic curve (ROC) and net reclassification index (NRI). Subgroup analysis and sensitivity analysis were performed to assess the interaction effects and robustness..Results255 individuals reached the composite outcome. Median follow-up duration was 2.03 [1.06, 3.19] years. Median age was 58.8 [48.7, 67.2] years with a median eGFR of 33.7 [17.6, 47.8] ml/min/1.73 m2. Five dataset were built after multiple imputation and five category-based Possion models were constructed for each dataset. Model 5 across five datasets had the best fitness with smallest AIC and largest AUC. The pooled results of Model 5 showed that total cholesterol (TC) (RR (95%CI) (per mmol/L) :1.143[1.023,1.278], P = 0.018) and percentage of body fat (PBF) (RR (95%CI) (per percentage):0.976[0.961,0.992], P = 0.003) were significant factors of composite outcome. The results indicated that comprehensive consideration of lipid metabolism and fat distribution is more critical in the prediction of CKD prognosis..ConclusionComprehensive consideration of lipid manifestations is optimal in predicting the prognosis of individuals with non-dialysis CKD stages 3–5.