Diabetic kidney disease (DKD) is one of the most severe microvascular complications of diabetes, with approximately 20–40% of patients with type 2 diabetes (T2D) progressing to DKD (1–3). DKD not only significantly increases the risk of end-stage renal disease but is also closely associated with cardiovascular event-related mortality, posing a major global public health challenge (4, 5). Although strict glycemic and blood pressure control can delay renal function deterioration, some patients still experience DKD progression even after achieving treatment targets (6), suggesting that the potential role of other metabolic factors remains incompletely elucidated.

Abnormal lipid metabolism has recently garnered widespread attention in DKD pathogenesis. Studies indicate that lipids promote DKD progression through mechanisms such as oxidative stress, inflammatory responses, autophagy dysregulation, endoplasmic reticulum stress, and apoptosis, leading to podocyte injury, extracellular matrix deposition, and macrophage infiltration (7–9). Low-density lipoprotein cholesterol (LDL-C), recognized as a core risk factor for atherosclerosis, has long been managed under the traditional paradigm of “the lower, the better” (10, 11). However, the 2024 China HEART project revealed a U-shaped association between LDL-C and all-cause mortality in primary prevention and low-risk populations, suggesting that both elevated and extremely low LDL-C levels may increase mortality risk (12). These findings challenge conventional lipid management strategies. Notably, current clinical studies on LDL-C and DKD risk remain limited and yield conflicting results. Some cohort studies demonstrated a positive correlation between LDL-C levels and albuminuria or estimated glomerular filtration rate (eGFR) decline (13, 14); a multicenter study reported that statins significantly reduced DKD risk by lowering LDL-C, particularly in populations with stringent lipid control (15). Conversely, other clinical studies indicated no statistically significant association between statin-induced LDL-C reduction and DKD cumulative incidence (hazard ratio 0.97, 95% CI 0.85–1.10; p = 0.60) (16). Additionally, a Mendelian randomization study found no causal relationship between LDL-C levels and DKD or urinary albumin excretion (17).

These controversies may arise from multiple factors. First, the dose–response relationship between LDL-C and DKD risk may exhibit non-linear characteristics, whereas prior studies predominantly used linear models, potentially failing to fully elucidate intrinsic associations. Second, the relationship may vary due to heterogeneity in metabolic status, such as differing glycemic control levels. Therefore, our study aimed to systematically explore the complex association patterns between LDL-C and DKD risk through a large-scale retrospective cohort analysis, thereby providing a reference for individualized lipid management in T2D patients.

This retrospective cohort study of 3,040 patients with T2D without baseline DKD demonstrated a complex relationship between baseline LDL-C levels and DKD onset risk. In the overall cohort, LDL-C exhibited a non-monotonic relationship with DKD risk, with the lowest hazard observed in the second quartile. However, stratification by glycemic control uncovered a U-shaped relationship in patients with well-controlled glycemia (HbA1c ≤ 7%), where both low (<2.66 mmol/L) and high (>3.57 mmol/L) LDL-C levels were associated with elevated DKD risk, while intermediate levels (2.66–3.57 mmol/L) conferred the lowest risk. In contrast, no significant association was observed among individuals with poor glycemic control. These findings suggest that the relationship between LDL-C and DKD risk may vary by glycemic control status, emphasizing the need for tailored lipid management strategies in patients with T2D.

The pathogenesis of DKD involves intricate interactions between metabolic disturbances and renal injury (22). LDL-C, traditionally recognized for its atherogenic properties, exerts direct and indirect effects on renal physiology through multiple intersecting pathways (23). Experimental evidence suggests that native LDL particles infiltrate the glomerular basement membrane and renal tubulointerstitium, where local oxidative modification generates oxidized LDL (ox-LDL), a potent activator of innate immunity (24). Ox-LDL binds to scavenger receptors (e.g., LOX-1 and CD36) on podocytes and mesangial cells, triggering NF-κB-driven transcription of pro-inflammatory cytokines (e.g., IL-6 and TNF-α) and NLRP3 inflammasome assembly, thereby promoting macrophage recruitment and sustained renal inflammation (24–26). In the diabetic milieu, hyperglycemia amplifies these effects synergistically by enhancing LDL glycation, forming advanced glycation end product (AGE)-modified LDL, which exhibits stronger affinity for scavenger receptors and exacerbates oxidative stress (27, 28). Despite mechanism insights, clinical studies investigating the impact of LDL-C on DKD risk remain contentious (14–17). Our previous cross-sectional study identified a non-linear association between LDL-C levels and DKD prevalence (29). However, cross-sectional designs inherently limit causal inference, as they cannot establish temporal relationships between LDL-C exposure and DKD onset. To address this gap, we designed the current longitudinal cohort study to systematically explore the complex association patterns between LDL-C and DKD risk, leveraging time-to-event analyses to strengthen causal plausibility.

Our study found that, in the overall population, although the trend test for the association between LDL-C quartiles and DKD risk (p = 0.006) and RCS analysis (P_non-linear = 0.075) suggested no significant non-linear relationship between baseline LDL-C and DKD risk, the absence of a significant linear association when LDL-C was analyzed as a continuous variable (p = 0.061), the non-monotonic changes in hazard ratios across quartiles (lowest risk in Q2 → partial rebound in Q3 → decline in Q4), and the “descending-then-flattening” trend of the RCS curve collectively indicated potential non-linear association patterns not captured by traditional models. This complexity may stem from the bidirectional pathological effects of LDL-C on the kidney. Mechanistic studies demonstrate that excessively high LDL-C induces cholesterol overload in renal cells via CD36-mediated endocytosis, activating endoplasmic reticulum stress and the unfolded protein response, which triggers mitochondrial dysfunction and Bax/Bak-dependent apoptosis, ultimately promoting proteinuria and glomerulosclerosis (30–32). Conversely, extremely low LDL-C levels may disrupt the structural integrity of cholesterol-rich lipid rafts in renal cell membranes. These rafts serve as critical platforms not only for insulin receptor clustering and signaling (33, 34) but also for organizing podocyte survival signals and nephrin localization (35). Disruption of lipid rafts exacerbates insulin resistance in proximal tubules, impairing glucose reabsorption and creating a vicious cycle of metabolic dysregulation (33, 36). More importantly, in glomerular podocytes, raft disintegration may destabilize slit diaphragm complexes via diminished nephrin phosphorylation, directly promoting albuminuria and podocyte detachment (37). Additionally, LDL-C provides essential precursors for steroid hormone synthesis. Depletion of cholesterol substrate may reduce the production of vitamin D metabolites and anti-inflammatory glucocorticoids (30, 38). Vitamin D deficiency is linked to upregulated renal renin–angiotensin–aldosterone system activity and TGF-β1-driven fibrosis, while impaired glucocorticoid signaling attenuates the suppression of pro-inflammatory cytokines in renal tissues (39). Such dual impairment of steroidogenic pathways likely compromises the kidney’s adaptive response to oxidative and inflammatory insults, thereby accelerating DKD progression in the context of extreme hypocholesterolemia.

Based on the aforementioned hypotheses, this study revealed through interaction analysis that baseline LDL-C levels and glycemic control status (HbA1c ≤ 7% vs. >7%) had a significant interactive effect on DKD risk. Further subgroup stratification analysis demonstrated that, in T2D patients with well-controlled glycemia (HbA1c ≤ 7%), LDL-C exhibited a clear U-shaped association with DKD risk: the risk significantly increased when LDL-C was <2.66 mmol/L or >3.57 mmol/L, while maintaining LDL-C within 2.66–3.57 mmol/L conferred the lowest risk. While current guidelines advocate aggressive LDL-C reduction (<1.8 mmol/L) for cardiovascular protection (10, 11), our data suggest that, for T2D patients with optimal glycemic control, maintaining LDL-C at 2.66–3.57 mmol/L may optimize nephroprotection. This divergence underscores the necessity of disease-specific LDL-C targets. However, no significant association was observed in the subgroup with poor glycemic control (HbA1c > 7%). This phenomenon might be related to differences in the pathophysiological effects of LDL-C in different glycemic states. Previous studies have suggested that a hyperglycemic environment enhanced the renal toxicity of LDL by promoting LDL glycosylation modification and oxidative stress: AGE-modified LDL could be more readily taken up by glomerular mesangial cells, activating the NF-κB pathway and inducing the expression of pro-fibrotic factors, thereby accelerating glomerulosclerosis (40). Under well-controlled glycemia, the degree of LDL glycosylation was reduced, and its pro-inflammatory properties might be partially suppressed (40, 41); then, the bidirectional effects of LDL-C on the kidneys (low levels disrupting lipid raft signaling and high levels inducing endoplasmic reticulum stress) might become the dominant mechanisms. Additionally, the integrity of insulin signaling pathways might regulate LDL-C metabolism: insulin resistance accompanying hyperglycemia might impair the LDL receptor-mediated cholesterol clearance capacity in the liver, leading to a decoupling between elevated circulating LDL-C levels and renal LDL deposition (42, 43), thereby masking its association with DKD risk. These mechanisms collectively supported the findings of this study, indicating that glycemic control status was a critical modulator of the relationship between LDL-C and DKD risk.

This study has several limitations. First, as a single-center retrospective cohort study, the findings might lack generalizability to populations with different ethnic backgrounds or healthcare settings, necessitating validation through multicenter prospective studies. Second, reliance on baseline LDL-C measurements precluded assessment of cumulative LDL-C exposure and temporal variability, which could better reflect long-term lipid metabolism dynamics. Third, data on glycemic variability and hypoglycemia episodes were not available in our cohort, which may confound the interpretation of HbA1c-based stratification. Fourth, DKD was defined by ACR/eGFR thresholds, which may misclassify non-diabetic kidney disease or overlook early tubular injury, potentially attenuating the observed associations. Fifth, unmeasured confounders such as dietary lipid intake, genetic polymorphisms (e.g., APOE), and residual effects of prior lipid-lowering therapy could not be completely accounted for, which may bias the observed U-shaped association. Future studies should incorporate longitudinal lipid profiling, standardized multi-omics renal assessments, and comprehensive lifestyle data. More importantly, prospective interventional trials are warranted to validate whether individualized LDL-C targets can be translated into clinical practice.

In conclusion, our study demonstrated a significant interaction between LDL-C levels and glycemic control in influencing DKD risk among patients with type 2 diabetes. Specifically, a U-shaped relationship between LDL-C and DKD progression was observed exclusively in individuals with optimal glycemic control, suggesting potential limitations of the traditional “lower is better” approach to LDL-C management in this population. These findings indicate that maintaining LDL-C within an intermediate range may reduce DKD risk in patients with well-controlled glycemia, while no such association was evident in those with suboptimal glycemic control. This highlights the importance of considering glycemic status when evaluating LDL-C targets.