Abstract
BACKGROUND: The relationship between low-density lipoprotein cholesterol (LDL-C) levels and diabetic kidney disease (DKD) risk remains controversial, with limited evidence on its interaction with modifiable risk factors. This study aimed to investigate the dose-response relationship between LDL-C and DKD risk in patients with type 2 diabetes (T2D). METHODS: A retrospective cohort of 3,040 patients with T2D without baseline DKD was followed. Association between LDL-C and DKD risk was analyzed using Cox regression analysis, interaction analysis, and restricted cubic splines (RCS). Sensitivity analyses excluded lipid-lowering medication users, and threshold effects were validated using piecewise regression and survival analysis. RESULTS: A total of 665 (21.9%) patients developed DKD during the follow-up (median: 3.13 years). In the fully adjusted model, LDL-C as a continuous variable showed no significant association with DKD risk (p = 0.061). When analyzed by quartiles, the hazard ratios (HRs) displayed a non-monotonic pattern: Compared to Q1, Q2 had the lowest risk (HR = 0.69, p = 0.001), followed by a partial rebound in Q3 (HR = 0.80, p = 0.046), and a subsequent decline in Q4 (HR = 0.72, p = 0.005), suggesting potential non-linearity. A significant LDL-C-by-glycemia control interaction was observed (P(interaction) = 0.013). In the HbA1c ≤ 7% subgroup, RCS analysis demonstrated a U-shaped relationship between LDL-C and DKD risk (P(non-linear) < 0.001), with nadir risk observed at 2.66-3.57 mmol/L. The risk increased below 2.66 mmol/L (HR = 1.55, p = 0.015) and trended upward above 3.57 mmol/L (HR = 1.47, p = 0.121). In this subgroup, sensitivity analyses excluding lipid-lowering drug users confirmed robustness, and survival curves showed lower DKD incidence in the intermediate LDL-C group (2.66-3.57 mmol/L) vs. low/high groups (p = 0.004). No associations were found in the HbA1c > 7% subgroup. CONCLUSION: Glycemic control modulates the LDL-C-DKD risk association in patients with T2D, with a U-shaped relationship observed in those with good glycemic control, thereby emphasizing the necessity of integrating glycemic status into LDL-C target evaluations.