Lipid phenotypes and diabetic kidney disease: a cluster-based study integrating traditional and nontraditional lipid parameters

BACKGROUND: This study aimed to identify lipid phenotypes integrating traditional and nontraditional lipid parameters and examine their associations with diabetic kidney disease (DKD). METHODS: The study analyzed 4,936 Chinese adults with type 2 diabetes mellitus recruited from 24 primary medical institutions. K-means clustering was performed using triglycerides (TG), high-density lipoprotein cholesterol (HDL), low-density lipoprotein cholesterol (LDL), atherogenic index of plasma, and remnant cholesterol (RC). Multivariable logistic regression was used to evaluate associations between lipid phenotypes and DKD and interactions with C-reactive protein were assessed on multiplicative and additive scales. The higher TG/RC phenotype exhibited more severe metabolic-inflammatory status and a higher prevalence of DKD than the other phenotype (52.44% vs 44.00%). After full adjustment, the higher TG/RC phenotype was associated with higher odds of DKD (OR 1.370, 95% CI 1.172-1.602), early DKD (OR 1.238, 95% CI 1.059-1.446), and reduced eGFR (OR 2.062, 95% CI 1.420-2.994). Participants with both the higher TG/RC phenotype and elevated C-reactive protein levels showed the highest odds of DKD (OR 1.881, 95% CI 1.540-2.297), with significant multiplicative and additive interactions. The clustered phenotype also provided incremental discriminative information beyond individual lipid markers, and the findings remained consistent across sensitivity analyses. CONCLUSIONS: Integrating traditional and nontraditional lipid measures may help identify clinically relevant lipid phenotypes associated with DKD and support more targeted renal assessment in T2DM patients.