To determine the influence of smoking and alcohol consumption on risk of Al
zheimer disease (AD), 238 AD patients enrolled in the MIRAGE Study at three
centers in the Boston, Massachusetts area were each frequency matched on t
he basis of sex, age, and year of birth to three non-demented control subje
cts from the Framingham Study, a longitudinal study of health in a communit
y 20 miles from Boston. Regular consumers of alcohol, including those drink
ing above and below the USDA recommended limits of 1.0 drinks/day for women
and 2.0 drinks/day for men, had a significantly lower risk of AD compared
to non-drinkers (crude OR's of 0.67 for consumers within the recommended gu
idelines and 0.48 for alcohol consumption above). Ever-smokers were less li
kely to develop AD than non-smokers, but this difference was not significan
t. Analyses evaluating the joint effects of drinking and smoking and adjust
ing for education and APOE epsilon 4 status eliminated the effect of smokin
g but did not lessen the effect of alcohol consumption. However, among epsi
lon 4 carrier women, smokers had a higher risk of AD than non-smokers where
as smoking was somewhat protective among women lacking epsilon 4. Although
these trends were not significant, the interaction between smoking and epsi
lon 4 status was significant (p=0.02). Smoking did not influence AD risk in
men regardless of APOE genotype. The interaction between smoking, gender a
nd APOE may help explain previous contradictory findings about the effect o
f smoking on AD risk. Alcohol consumption within nationally recommended lim
its may protect against AD. However, it would be premature to recommend thi
s as a prophylaxis until the protective mechanism is understood and further
studies are performed to more precisely define the minimal level and durat
ion of exposure necessary to realize a benefit.