Urinary excretion of dithiocarbamates and self-reported Cruciferous vegetable intake: application of the 'method of triads' to a food-specific biomarker

Objective: Greater intake of Cruciferous vegetables (e.g. broccoli) may prevent cancer at several sites. Urinary excretion of isothiocyanate conjugates (dithiocarbamates, DTC) provides a specific biomarker of Cruciferous vegetable consumption suitable for epidemiological investigations. However, no gold-standard referent is available for evaluating urinary DTC levels as an estimator of Cruciferous vegetable consumption. We compared urinary DTC levels to intake as measured by two self-reported dietary assessment techniques. Design: Cruciferous vegetable consumption was measured before and after a behavioural dietary intervention using multiple 24-hour recalls (24HR), a food-counting questionnaire (VFQ) and urinary DTC excretion levels. Analysis included a structural equation approach (Method of Triads) combining these three assessment techniques to estimate the relationship between DTC level and the study population's 'true' Cruciferous vegetable intake. Setting: The intervention curriculum assisted participants in consuming about 2 servings per day for a 6-week period. Participants attended four classes emphasising problem-solving skills, dietary counselling and vegetable preparation skills. There were no dietary restrictions. Subjects: Thirty-three healthy, free-living, post-menopausal women. Results: Although few participants reported Cruciferae consumption prior to the intervention, 30 participants reported Cruciferae consumption after the intervention (Post-intervention). Urinary DTC levels were correlated with estimated intake derived from either the 24HR (r = 0.57; 95% confidence interval (95% CI) 0.28, 0.76) or VFQ (r = 0.49; 95% CI 0.17, 0.71). The validity coefficient (Method of Triads) between urinary DTC excretion and an index of true Cruciferous intake was stronger than the Pearson correlation (r(v) = 0.65; 95% CI 0.35, 0.90), and comparable to estimates derived from the 24HR (r(v) = 0.82; 95% CI 0.65, 1.00) or VFQ (r(v) = 0.76; 95% CI 0.47, 0.92) method. These associations were not affected by adjustment for body mass index, energy intake, or social approval or desirability response sets. Conclusions: Food-frequency questionnaires (FFQ) suitable for large epidemiological studies may not be designed to measure all Cruciferae, and cannot capture exposure to phytochemicals derived from those vegetables. Urinary DTC measurement was significantly correlated with Cruciferae intake derived from two dietary assessment approaches, and urinary DTC levels could supplement traditional FFQ data by providing an index of recent Cruciferous vegetable intake not susceptible to reporting biases.