Pyridoxal 5'-phosphate and risk of stroke: triangulation of evidence from a nationally representative cohort and bidirectional Mendelian randomization analysis.

Background: Stroke is a leading cause of mortality and disability worldwide. Identifying predictive biomarkers and modifiable risk factors is crucial for stroke prevention in the context of predictive, preventive, and personalized medicine (PPPM). We aimed to investigate the association of serum pyridoxal 5'-phosphate (PLP) levels with stroke prevalence in a nationally representative cohort and to assess the causal relationship using bidirectional Mendelian randomization (MR) analysis, with a focus on the implications for PPPM strategies in stroke management.

Methods: We included 6839 participants aged ≥ 18 years from the National Health and Nutrition Examination Survey (NHANES) 2005-2013. Serum PLP levels were measured by high-performance liquid chromatography. Stroke prevalence was ascertained by self-report. We used generalized linear models, Kaplan-Meier curves, restricted cubic splines, stratified analysis, receiver operating characteristic (ROC) curves, and bidirectional two-sample MR to examine the association of PLP levels with stroke prevalence and assess the causal relationship.

Results: In the fully adjusted model, participants with low serum PLP levels had significantly higher odds of stroke compared to those with high levels (odds ratio (OR) = 6.51e-01, 95% confidence interval (CI) 4.46e-01-9.50e-01, P = 2.74E-02). Kaplan-Meier curves showed significantly lower survival probability in the low PLP group (P < 0.05). The restricted cubic spline analysis revealed a non-linear association, with the highest stroke risk at lower PLP levels. The stratified analysis showed significant associations in several subgroups. The ROC analysis indicated good predictive performance of the fully adjusted model (area under the curve (AUC) > 0.7). The MR analysis supported a protective causal effect of PLP on stroke risk (OR = 0.7723581, 95% CI 0.6388086-0.9336201, P = 0.00345), while the reverse MR analysis did not suggest a causal effect of stroke on PLP levels.

Conclusions: Low serum PLP levels are significantly associated with higher stroke prevalence in a nationally representative the United States (US) sample. Integration of observational and genetic evidence supports a protective causal role of PLP in stroke risk. Serum PLP may serve as a promising predictive biomarker for stroke risk assessment and a potential target for personalized nutritional interventions in stroke prevention, in line with PPPM strategies. Our findings highlight the importance of maintaining optimal vitamin B6 status for effective PPPM-guided stroke prevention and management.

Supplementary information: The online version contains supplementary material available at 10.1007/s13167-024-00392-2.