Association between plasma trimethylamine N-oxide and cerebral white matter hyperintensity: a cross-sectional study

Cerebral white matter hyperintensity (WMH) is a pivotal imaging feature of cerebral small vessel disease (CSVD), closely correlated with an elevated risk of ischemic stroke (IS). Trimethylamine N-oxide (TMAO), a metabolite of gut microbiota, is increasingly associated with IS and atherosclerosis. However, the intricate relationship between TMAO and WMH remains ambiguous. This study aimed to study the connection between plasma TMAO and WMH. Furthermore, it assessed the potential of TMAO as a risk evaluation instrument for WMH.

In this cross-sectional study, we categorized WMH into periventricular WMH (P-WMH) and deep WMH (D-WMH), based on its locations. The severity of WMH was assessed and grouped according to the Fazekas scale. Plasma TMAO levels were quantitatively determined. We established the correlation between plasma TMAO levels and WMH severity using a Logistic regression model. Additionally, we employed ROC curves to evaluate the diagnostic efficacy of plasma TMAO concentration in distinguishing the severity of WMH.

A higher plasma TMAO tertile was significantly linked to a higher Fazekas score, encompassing the overall score, P-WMH score, and D-WMH score (p < 0.001). A logical regression analysis revealed that plasma TMAO levels were independently associated with overall moderate and severe WMH, compared to overall non-mild WMH, in the unadjusted model (OR = 1.373, 95%CI 1.183-1.594 for moderate; OR = 1.384, 95%CI 1.192-1.607 for severe), the adjusted model a (OR = 1.436, 95%CI 1.214-1.669 for moderate; OR = 1.446, 95%CI 1.222-1.711 for severe) and the adjusted model b (OR = 1.490, 95%CI 1.234-1.800 for moderate; OR = 1.494, 95%CI 1.237-1.805 for severe). The analysis also showed an independent correlation between plasma TMAO levels and WMH severity, irrespective of the unadjusted model, adjusted model a, or adjusted model b, when considering P-WMH and D-WMH severity. The ROC indicated that, in overall WMH and P-WMH, the area under curve (AUC) for non-mild and severe WMH were both>0.5, while the AUC for moderate WMH was<0.5. In contrast, in D-WMH, the AUC for non-mild, moderate, and severe WMH were all>0.5.

Plasma TMAO levels exhibited a significant correlation with both overall and region-specific WMH severity. Furthermore, the plasma TMAO levels displayed robust predictive capability for D-WMH.