Weiqiang Qin, Bo Liu, Xiaoling Yi, Juan Lei, Lan Mai
NHHR was independently associated with incident T2DM, with the majority of its effect (77.3%) operating independently of BMI. UA's association with T2DM was largely mediated through BMI, indicating potentially distinct pathway profiles between the two markers. Joint exposure analysis revealed that the high-UA/low-NHHR group (G3) had an incidence rate comparable to the reference group, while elevated NHHR was the more prominent risk marker across combined groups. These findings suggest that NHHR may complement uric acid assessment in risk awareness among metabolically abnormal individuals with normal BMI; however, the incremental predictive value of NHHR was limited, and clinical implementation requires validation in independent cohorts.
BACKGROUND: Serum uric acid (UA) and the non-high-density lipoprotein cholesterol to high-density lipoprotein cholesterol ratio (NHHR) are both established risk factors for type 2 diabetes mellitus (T2DM), yet their metabolic pathways to diabetes may differ substantially. This study compared the independent associations of UA and NHHR with incident T2DM, quantified the mediating effects of BMI along each pathway, and evaluated the incremental predictive value of combined assessment.
METHODS: Using the CHARLS 2011-2018 prospective cohort (8,150 diabetes-free adults ≥ 45 years), we examined associations of UA and NHHR with incident T2DM via Cox regression across four models, characterized dose-response relationships with RCS, and conducted exploratory mediation analysis (VanderWeele's counterfactual framework) to quantify BMI-mediated proportions. Incremental predictive value was assessed by C-index, NRI, IDI, and DCA.
RESULTS: In the fully adjusted model (n = 6,762), each 1-unit increase in NHHR was independently associated with T2DM (HR = 1.104, 95% CI: 1.050-1.161, P < 0.001), whereas UA showed no independent association (HR = 1.048, 95% CI: 0.978-1.122, P = 0.184). Mediation analysis revealed that 77.3% of NHHR's total association was not mediated through BMI (BMI-mediated proportion: 22.7%). In contrast, 51.4% of UA's association with T2DM was mediated through BMI, with the non-BMI-mediated direct effect remaining non-significant. Adding NHHR to the base model showed marginal improvement in C-index (Δ=+0.005, P = 0.091) and limited net benefit in decision curve analysis (ΔNB = 0.0049 at 11.6% prevalence threshold). Dyslipidemia significantly modified the NHHR-T2DM association (interaction P = 0.015).
CONCLUSION: NHHR was independently associated with incident T2DM, with the majority of its effect (77.3%) operating independently of BMI. UA's association with T2DM was largely mediated through BMI, indicating potentially distinct pathway profiles between the two markers. Joint exposure analysis revealed that the high-UA/low-NHHR group (G3) had an incidence rate comparable to the reference group, while elevated NHHR was the more prominent risk marker across combined groups. These findings suggest that NHHR may complement uric acid assessment in risk awareness among metabolically abnormal individuals with normal BMI; however, the incremental predictive value of NHHR was limited, and clinical implementation requires validation in independent cohorts.
CLINICAL TRIAL NUMBER: Not applicable.