Abstract
A large number of prospective studies have observed an inverse relationship between a moderate intake of alcohol and coronary heart disease morbidity and mortality. Concerning death from all-causes, results are not unanimous. Alcohol intake was associated with a protection of all-cause mortality in England and USA physicians and the large study of the American Cancer Society. None of these studies separated the effects of different alcoholic beverages.
In our prospective studies in France on 35 000 middle-aged men, we observed that only wine at moderate intake, was associated with a protective effect on all-cause mortality. The reason was that in addition to the known effect on cardiovascular diseases, a very moderate intake of wine, protected also from cancer and other causes as confirmed by Gronbaek in Denmark. Our recent results also indicate that the protective effect of a moderate intake of wine on all-cause mortality is observed at all levels of blood pressure and serum cholesterol.
Key words: Alcohol, mortality, wine.
Abbreviations : BMI : body mass index; CHD: coronary heart disease.

Introduction
A large number of studies have reported that amoderate intake of alcohol is associated with a lower risk of morbidity and mortality (1, 2) from coronary heart disease (CHD), more efficient on morbidity. These results however, do not imply that longevity is increased in alcohol drinkers. In Great Britain, Shaper et al. reported (3) in a prospective study on 7735 middle aged men, that in moderate drinkers there was a 34 % lower mortality rate from CHD but no benefit in total cardiovascular and all-cause mortality. Another study concluded: “Our results are inconsistent with the belief that daily consumption of a few glasses of wine has salutary effects” (4). In that study, they did not evaluate wine drinkers but rather beer and spirit drinkers. Thus, they anticipated that the effect of wine could not be different from that of other alcoholic beverages.

Alcohol and all-cause mortality.
Do we have further indications that moderate alcohol drinking does not lower death from all causes? In 1994, the same year of the Shaper study (3) Sir Richard Doll et al. (5), also in Great Britain, in their study on British physicians, reported that a moderate intake of alcohol (21 units/week) reduced by more than 20% mortality from all causes, in addition to the usual protective effect on CHD mortality.
In USA, a prospective study on 22 701 men, also physicians, reported that the consumption of 2 to 6 drinks of alcohol per week lowered mortality from all causes by 21 to 28 % (6).
The question raised by these two studies involves the kind of alcoholic beverage drank by these physicians. Could it be that they normally drank more wine than the general population? However, in USA, in the largest study ever performed on the effect of alcohol (238 206 subjects) (7) on mortality, but in motivated subjects, very moderate drinkers, a 20% reduction in the mortality from all causes was registered, for 1 to 3 drinks per day, a result similar to that observed in the British (5) and American physicians (6).
Thus, at least three large prospective studies have observed a similar reduction in mortality from all causes for a moderate intake of alcohol. The question raised is why large studies have observed a similar protective effect of alcohol on all-cause mortality and not the previous smaller
studies? Such a discrepancy could be due to the size of the cohort or/and possibly the type of alcoholic beverage consumed.
Unfortunately, none of these prospective studies have evaluated the type of the alcoholic beverage consumed.

Type ok alcohol and all-cause mortality
The first prospective study suggesting that the type of alcoholic beverage in addition to its quantity, could be important for the protective effect on mortality, was that of Gronbaek (8).
In that study, for 3 to 5 drinks per day cardiovascular mortality was reduced by 47% in the wine drinkers, but the other causes were also 50% lower. An important issue was that the other causes were not reduced in the drinkers of beer and spirits.
Since French males are considered the largest consumers of wine in the world, we conducted a study in Eastern France that included 34 014 middle-aged men (40 - 60 years at baseline) in which we evaluated mortality in relation to their alcohol intake, with a follow-up of 10 - 15 years (9), In that study, every social class consumed wine, and the largest consumers were workers and tradesmen (48.9 and 47.6 g/ day respectively). Higher classes such as high-level managers consumed only 36.2 g/day of wine. In that cohort, wine was the main alcohol consumed (82% of the total alcoholic beverages). A moderate intake of wine (2 - 3 glasses per day) was associated with a 24 - 31% reduction in all-cause mortality, in smokers, ex-smokers and nonsmokers.

In a subsequent study (10) with 36 250 men, we evaluated separately the effect of wine, beer, and of wine + beer. Significant lower all-cause mortality was observed only
with wine at a moderate intake (1 to 5 glasses/day). With beer, either representing more than 40 % of alcohol or less, there was no reduction of the all-cause mortality, at any level of alcohol consumption. The reduction of all-cause mortality in moderate wine drinkers could be attributed to a lower
death from cancer (by 20%) and from other causes (by 28 to 42 %) shown in Figure 1.
The specificity of wine for lowering the risk of all-cause mortality and cancer has been confirmed recently by Gronbaek et al. (11). It was only in wine drinkers that the all-cause mortality was lowered by more than 20% for an intake of up to 3 drinks per day. They also found that it was only wine but not spirits or beer that reduced cancer mortality by up to 20% (3 glasses of wine/day). As for cancer, the subsequent question was to evaluate on what types of cancer wine may have protective effects.
Gronbaek et al. have already shown that wine drinking does not increase the risk of upper digestive tract cancer (12) as compared to beer or spirits.
In our cohort from Eastern France mentioned above, in 946 cases of cancer we evaluated the effect of wine drinking compared to abstainers (Fig. 2). The risk of both lung cancer and of other cancers was lowered by about the same extent for 1 - 3 glasses of wine (other cancers), and for 3 - 5 glasses (lung cancers). Above 5 glasses /day the risk of cancer was increased by 50 %.

Since it is only wine that offers some protection on the risk of cancer, that protection does not seem to be due to alcohol per se. Among the phenolic substances contained in wine, resveratrol is one substance that has been extensivelys tudied in relation to cancer and other problems (13). It has been shown that resveratrol modulates lipoprotein metabolism, presents antioxidant and antiinflammatory activities, inhibits platelet aggregation and cellular events associated with cancer i.e. initiation, promotion and progression (13).