Blood lead is a predictor of homocysteine levels in a population-based study of older adults

Abstract

Lead and homocysteine are both associated with cardiovascular disease and cognitive dysfunction. We evaluated the relations among blood lead, tibia lead, and homocysteine levels by cross-sectional analysis of data among subjects in the Baltimore Memory Study, a longitudinal study of 1,140 randomly selected residents in Baltimore, Maryland, who were 50-70 years of age. Tibia lead was measured by (superscript)109(/superscript)Cd K-shell X-ray fluorescence. The subject population had a mean +/- SD age of 59.3 +/- 5.9 years and was 66.0% female, 53.9% white, and 41.4% black or African American. Mean +/- SD blood lead, tibia lead, and homocysteine levels were 3.5 +/- 2.4 microg/dL, 18.9 +/- 12.5 microg/g, and 10.0 +/- 4.1 micromol/L, respectively. In unadjusted analysis, blood lead and homocysteine were moderately correlated (Pearson's r = 0.27, p < 0.01). After adjustment for age, sex, race/ethnicity, educational level, tobacco and alcohol consumption, and body mass index using multiple linear regression, results revealed that homocysteine levels increased 0.35 micromol/L per 1.0 microg/dL increase in blood lead (p < 0.01). The relations of blood lead with homocysteine levels did not differ in subgroups distinguished by age, sex, or race/ethnicity. Tibia lead was modestly correlated with blood lead (Pearson's r = 0.12, p < 0.01) but was not associated with homocysteine levels. To our knowledge, these are the first data to reveal an association between blood lead and homocysteine. These results suggest that homocysteine could be a mechanism that underlies the effects of lead on the cardiovascular and central nervous systems, possibly offering new targets for intervention to prevent the long-term consequences of lead exposure.

Figure 1. Crude relation of log₁₀-transformed blood lead levels and log₁₀-transformed tibia lead levels for 955 participants in the Baltimore Memory Study. Dashed line represents a locally weighted smoothing fit (lowess bandwidth, 0.10) (Cleveland 1979).

Figure 2. Smoothed plot of residuals of blood lead and homocysteine levels, controlling for covariates. Values were adjusted for age, sex, race/ethnicity, BMI, educational level, and tobacco and alcohol use. The three data points with blood lead concentrations > 15 μg/dL have been excluded from the plot (but not from the regression model) so that the portion of the plot with the most data could be more clearly visualized. The solid line is predicted linear fit, and the dashed line is from a locally weighted smoothing fit (lowess bandwidth, 0.05) (Cleveland 1979).