‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease?*

Lung cancer and cardiovascular disease are major causes of death in the United States. It has been proposed that carotenoids and retinoids are agents that may prevent these disorders. We conducted a multicenter, randomized, double-blind, placebo-controlled primary prevention trial -- the Beta Carotene and Retinol Efficacy Trial -- involving a total of 18,314 smokers, former smokers, and workers exposed to asbestos. The effects of a combination of 30 mg of beta carotene per day and 25,000 IU of retinol (vitamin A) in the form of retinyl palmitate per day on the primary end point, the incidence of lung cancer, were compared with those of placebo. A total of 388 new cases of lung cancer were diagnosed during the 73,135 person-years of follow-up (mean length of follow-up, 4.0 years). The active-treatment group had a relative risk of lung cancer of 1.28 (95 percent confidence interval, 1.04 to 1.57; P=0.02), as compared with the placebo group. There were no statistically significant differences in the risks of other types of cancer. In the active-treatment group, the relative risk of death from any cause was 1.17 (95 percent confidence interval, 1.03 to 1.33); of death from lung cancer, 1.46 (95 percent confidence interval, 1.07 to 2.00); and of death from cardiovascular disease, 1.26 (95 percent confidence interval, 0.99 to 1.61). On the basis of these findings, the randomized trial was stopped 21 months earlier than planned; follow-up will continue for another 5 years. After an average of four years of supplementation, the combination of beta carotene and vitamin A had no benefit and may have had an adverse effect on the incidence of lung cancer and on the risk of death from lung cancer, cardiovascular disease, and any cause in smokers and workers exposed to asbestos.

Observational studies suggest that people who consume more fruits and vegetables containing beta carotene have somewhat lower risks of cancer and cardiovascular disease, and earlier basic research suggested plausible mechanisms. Because large randomized trials of long duration were necessary to test this hypothesis directly, we conducted a trial of beta carotene supplementation. In a randomized, double-blind, placebo-controlled trial of beta carotene (50 mg on alternate days), we enrolled 22,071 male physicians, 40 to 84 years of age, in the United States; 11 percent were current smokers and 39 percent were former smokers at the beginning of the study in 1982. By December 31, 1995, the scheduled end of the study, fewer than 1 percent had been lost to follow-up, and compliance was 78 percent in the group that received beta carotene. Among 11,036 physicians randomly assigned to receive beta carotene and 11,035 assigned to receive placebo, there were virtually no early or late differences in the overall incidence of malignant neoplasms or cardiovascular disease, or in overall mortality. In the beta carotene group, 1273 men had any malignant neoplasm (except nonmelanoma skin cancer), as compared with 1293 in the placebo group (relative risk, 0.98; 95 percent confidence interval, 0.91 to 1.06). There were also no significant differences in the number of cases of lung cancer (82 in the beta carotene group vs. 88 in the placebo group); the number of deaths from cancer (386 vs. 380), deaths from any cause (979 vs. 968), or deaths from cardiovascular disease (338 vs. 313); the number of men with myocardial infarction (468 vs. 489); the number with stroke (367 vs. 382); or the number with any one of the previous three end points (967 vs. 972). Among current and former smokers, there were also no significant early or late differences in any of these end points. In this trial among healthy men, 12 years of supplementation with beta carotene produced neither benefit nor harm in terms of the incidence of malignant neoplasms, cardiovascular disease, or death from all causes.

Great efforts and expense have been expended in attempts to detect genetic polymorphisms contributing to susceptibility to complex human disease. Concomitantly, technology for detection and scoring of single nucleotide polymorphisms (SNPs) has undergone rapid development, extensive catalogues of SNPs across the genome have been constructed, and SNPs have been increasingly used as a means for investigation of the genetic causes of complex human diseases. For many diseases, population-based studies of unrelated individuals--in which case-control and cohort studies serve as standard designs for genetic association analysis--can be the most practical and powerful approach. However, extensive debate has arisen about optimum study design, and considerable concern has been expressed that these approaches are prone to population stratification, which can lead to biased or spurious results. Over the past decade, a great shift has been noted, away from case-control and cohort studies, towards family-based association designs. These designs have fewer problems with population stratification but have greater genotyping and sampling requirements, and data can be difficult or impossible to gather. We discuss past evidence for population stratification on genotype-phenotype association studies, review methods to detect and account for it, and present suggestions for future study design and analysis.