Abstract
BACKGROUND: Elderly individuals are susceptible to the accrual of White Matter Lesions (WMLs), a subcategory of cerebral small-vessel disease. WMLs are strongly linked to an increased risk of strokes, intracerebral hemorrhages, and dementia. While the relationship between blood glucose levels and the development of WMLs has been investigated in previous studies, the findings remain inconsistent. Some evidence suggests that glucose dysregulation, including both hypo- and hyperglycemia, may contribute to WML formation through mechanisms such as endothelial dysfunction and chronic inflammation. However, other studies report no significant correlation. This inconsistency underscores the need for further investigation. METHODS: In this investigation, the primary data were derived from a predictive mathematical model designed to estimate WMLs based on parameters obtained from routine medical examinations, with head MRI scans serving as the reference standard for WML diagnosis and quantification. We leveraged multivariable logistic regression analysis to scrutinize the relationship between blood glucose concentrations and WMLs. Additionally, we employed a restricted cubic spline regression model to investigate a potential non-linear relationship between these variables. RESULTS: There were 1904 participants who underwent medical check-ups which included a head MRI. Generally, the relationship between blood glucose levels and white matter lesions followed an asymmetric U-shaped curve (P for non-linearity = 0.004). A consistent finding was that compared to the individuals in the 2nd and 3rd quartiles (95 to 107 mg/dl), the 1st quartile (OR, 1.71; 95% CI: 1.26-2.30) and 4th quartile (OR, 1.57; 95%CI: 1.12-2.20) had white matter lesions were significantly higher. CONCLUSION: An asymmetric U-shaped relationship exists between blood glucose and WMLs, with the lowest risk occurring at 95-107 mg/dl. Management of blood glucose can help prevent the occurrence and development of WMLs. However, the study's cross-sectional design limits causal inference, and the reliance on pre-existing data constrained the availability of variables.