Vitamin K Concentration and Cognitive Status in Elderly Patients on Anticoagulant Therapy: A Pilot Study

Abstract

Objectives: Recent studies have suggested that vitamin K may exert significant effects on the central nervous system. The present study investigates the relationship between vitamin K plasmatic levels and cognitive functions in elderly patients on oral anticoagulant therapy (OAT).

Design: At the Thrombosis Centre of Haematology, "Sapienza" University of Rome, 85 patients on OAT, aged between 75 and 92, were randomly enrolled in the study. Patients were on OAT with vitamin K antagonists (VKAs). Vitamin K1 concentrations were determined using standardized High-Performance Liquid Chromatography (HPLC). Cognitive functions were assessed using the Milan Overall Dementia Assessment (MODA).

Results: MODA scores are positively correlated to vitamin K1 concentration. Patients with vitamin K1 below 0.100 μg/L and between 0.100 and 0.400 μg/L and between 0.100 and 0.400 μg/L and between 0.100 and 0.400 p < 0.001). Even long-term OAT (>10 years) does not affect MODA scores. Education seems to exert a greater role on the cognitive status in comparison with aging.

Conclusions: The study shows a positive association between vitamin K1 concentration and cognitive status in elderly patients (≥75 years) on OAT. The relationship between vitamin K1 concentration and MODA scores is described by a linear model. Cognitive status is not influenced by the duration of OAT but by the years of education.

Figure 1

MODA scores vs. vitamin K1 plasmatic levels. The dispersion of the values tends to decrease as the vitamin K1 concentration increases (see also in the Supplementary Material). This leads to larger mean MODA scores for higher concentrations of the vitamer, in particular for concentrations above 0.400 μg/L (blue dashed line).

Figure 2

Analysis of MODA scores vs. vitamin K1 levels after binning the data. The linear relationship between MODA scores and vitamin K1 plasmatic levels appears clear after binning the data according to vitamin K1 concentration bicentiles. The value of χ²_ν of 0.3 is not <<1 thus suggesting that we should not reject the null hypothesis of a linear relationship between vitamin K1 levels and MODA scores. This is supported by the normalised residuals being randomly distributed around the zero and not showing any clear structures. The value for the Pearson correlation coefficient of 0.94 shows a very high positive correlation between MODA scores and vitamin K levels, thus further confirming that the linear regression is a good model for the data.