Serum retinol and risk of subsequent cancer: extension of the Evans County, Georgia, study.

A previous report from the Evans County, Ga., cohort correlated low base-line retinol levels in 1960-62 to an increased risk of cancer developing in the following 12-14 years. So that this inverse association could be retested, retinol determinations were performed on sera in 135 incident cancer cases, identified during a vital status follow-up in 1981, and in 237 controls matched by race, sex, age, and date of venipuncture. The serum was drawn during the second cohort examination between 1967 and 1969 and stored at -20 degrees C until analysis by high-pressure liquid chromatography in January 1982. Eligible for the study were patients whose initial cancer diagnosis was established more than 24 months after venipuncture. Base-line retinol levels of subsequent cancer cases and their matched controls were similar in the 4 race-sex groups. The risk of cancer at all sites associated with the lowest retinol quintile was similar to that associated with the highest quintile. In multivariate analysis, with control for age, cholesterol, body mass, smoking status, and social class, no significant relationship of serum retinol and case status was found. In summary, these data fail to confirm the strong dose-response relationship between baseline retinol levels and subsequent cancer reported in the previous study.