To investigate the usefulness of serum carotenoids as biomarkers of fruit and vegetable consumption.

Reproducibility study on three repeat measurements of serum carotenoids. Correlation analysis of carotenoids and dietary food intake, and regression analysis of potential predictive parameters for serum carotenoid levels.

New York, USA.

Women participating in the New York Women's Health Study, a prospective study of sex hormones, diet and breast cancer. Forty-eight women with three repeat blood samples and 302 women having a blood sample and a dietary history questionnaire.

Serum carotenoid concentrations were highly reproducible between one- and two-year repeat samples. Estimated fruit and vegetable consumption was positively correlated with serum carotenoid concentrations but correlation coefficients were low. Consumption of fruit was predictive for serum levels of beta-carotene, alpha-carotene and beta-cryptoxanthin, while vegetable consumption was predictive for serum beta-carotene, lutein, zeaxanthin and lycopene. Serum concentrations of cholesterol and triglycerides were predictive for serum carotenoids but adjustment for their levels had little or no influence on the correlation between fruit and vegetable consumption and serum carotenoid concentrations.

One single serum measurement of alpha-carotene, beta-carotene and lutein can accurately rank subjects according to their usual serum level. Serum concentrations, however, correlate only moderately with estimated dietary intake of fruits or vegetables and should therefore be used with caution as biomarkers of fruit and vegetable intake.