Before discussing the implications of these observations, it is necessary to consider their reliability. The data in Table 1 are shown because of the possibility that the cases having the background information about alcohol use might not be representative of those for whom this information was not recorded. While the table shows a trend for values to be larger in the group with known history, only one pair of values (prevalence of emphysema) approached statistical significance (p = 0.Q5) using the x2 test.
While we have found no plausible explanation for the above trend, it does not appear sufficient to interfere with the validity of the much larger differences observed between drinkers and nondrinkers. Tables 2 and 3, relating to prevalence of emphysema, were also analyzed by the x2 test and, as noted, gave values at or just above the usually accepted borderline of statistical significance (p = 0.Q5). However, the logistic analysis used with respect to the measured variables (Table 6) is a more reliable statistical tool, since it uses all available information simultaneously. It showed the effect of alcohol on extent of emphysema to have a low likelihood (p = 0.008) of having occurred by chance. Since the prevalence data deal with the same population, the trends shown in lables 2 and 3 are considered valid even though the x2 tests on individual tables show only borderline values.
The accuracy of background historic data in clinical records is often questioned. In our experience, however, such data have been very useful and have permitted the demonstration of a number of highly significant differences between smokers and nonsmokers and even between cigarette and pipe-cigar smokers. It is logical to believe that the history of alcohol consumption is as reliable as that for smoking. The major criticism regarding the background information in this study is that it was possible only to identify broad groups of cases but not to specify the type or amount of alcohol used or the duration. It is possible that a few drinkers may have reported themselves as nondrinkers, and it is likely that some cases who were in feet heavy drinkers placed themselves in the “slight-to-moderate” class. The effect of errors such as these would be to reduce the size of the differences between groups. Therefore, it seems that the true differences have been reduced, rather than exaggerated, by the retrospective nature of the smoking and drinking histories.
We did not foresee the results reported. We expected that alcohol use would probably have little or no effect on CLE, but, if any, the effect would be in the direction of exacerbation of the known smoking effect. However, after seeing the data and reflecting on current concepts of the pathogenesis of emphysema and the cellular responses to alcohol, the result seems considerably less surprising. Indeed, perhaps the results could have been predicted.
The rationale is summarized as follows: an emphysema-like lesion can be produced in animals by administration of proteolytic enzymes, especially elastase. Such enzymes are probably liberated in the human lung by inflammatory cells, both granulocytes and macrophages. Normal individuals have antiproteases in the circulating blood and probably in interstitial tissue and persons who lack these enzymes clearly have an increased risk of emphysema. Thus, it appears that the local balance between proteolytic and antiproteolytic activity may determine whether emphysema occurs. The association between smoking and prevalence of emphysema appears to result from the excess numbers of both granulocytes and macrophages in the lungs of smokers, which may liberate a greater amount of proteolytic enzyme than some individuals are able to inactivate adequately. The data in Table 3 suggest that more than half (56.1 percent) of nondrinking smokers develop CLE of trace or greater extent. In an earlier report it was shown that the lungs from nonsmokers in the entire autopsy series had a CLE prevalence of 18 percent. Thus, smoking upset the proteolytic-antiproteolytic balance in about 38 percent (56 —18 percent) of this population. The 44 percent of smokers without CLE presumably have sufficient antiproteolytic activity to prevent alveolar septal destruction even in the presence of smoking.
Alcohol has been shown to interfere with many activities of both granulocytes and macrophages. These include mobilization, migration, phagocytosis, bactericidal activity, and lysosomal enzyme release. One report is of special interest in this regard. It involved an examination of phagocytic activity of Kupffer cells in the liver, tested in vivo. The study was done in patients admitted to hospital for acute alcoholism. Phagocyte activity was shown to be reduced and did not return to normal until four to seven days after admission. This suggests that alcohol effects on inflammatory cells may persist for several days, even in individuals who do not drink daily. Reduction in protease production is also likely, but has not been specifically demonstrated in any reports known to the authors. Because of the known effects, however, it is logical that alcohol would tend to shift the proteolytic-antiproteolytic balance in the direction favoring prevention of CLE. Again, referring to Table 3, the prevalence was reduced from 56.1 percent in nondrinkers to 35.1 percent in heavy drinkers, or by 37 percent (21 percent/56.1 percent). Reductions in measured extent of CLE (Tables 4 and 5) were of comparable degree.
Data concerning nonsmokers are not presented in this report because the population includes only 31 nonsmokers. It is, however, pertinent to note that none of them had as much as 20 percent involvement by emphysema and the mean extent was only about 2 percent for all cases and 7 percent in those who did have CLE. These values are comparable to those already reported for the entire nonsmoker population: 1.7 percent, all cases and 6.9 percent, cases with CLE.
The data presented here cannot, of course, be extrapolated directly to a nonhospitalized or even a nonautopsied population. A phenomenon known as “Berksons bias” may have exerted some effect at both levels of selection. This bias could result if the hospital admission rate (or autopsy rate) were different for patients with smoking-related lesions who used alcohol from the rate for those who did not drink. However, we are not aware of any such systematic difference. Additionally, the feet that a careful clinical study of a nonhospitalized population has shown a tendency for some aspects of pulmonary function to be closer to predicted values in smokers who drink than in nondrinking smokers seems to support the validity of the present observation. Finally, the feet that some of the smoking effects observed here appeared to be enhanced by drinking (Tables 6 to 9), while others including CLE were diminished, suggests that the effects cannot be entirely attributed to that bias. A description of the bias is beyond the scope of this article, since it cannot be summarized concisely. Readers who are not familiar with it should examine the above references.
Further studies are planned to attempt to clarify the role of these sources of bias. For this purpose, major causes of death are being coded for all 871 cases constituting the entire population involved in the project. Patterns of variation of CLE among various diagnostic groupings of cases may permit evaluation of the importance of these potential effects. For example, one could describe some groupings of causes of death which would exaggerate the bias and others which would minimize it. The degree to which these groupings alter the prevalence or mean extent of CLE should provide evidence as to the importance of these phenomena in this study.
The possibility that emphysema may be a preventable disease was considered in detail in a recent series of papers. It was concluded that both individuals and society would benefit from such prophylaxis. The method contemplated involved administration of anti-proteolytic agents with the intent to inactivate elastases liberated in the lung by inflammatory cells. The present study supports the concept that emphysema prophylaxis is possible and permits proposal of an alternative mechanism to accomplish it; namely, the inhibition of elastase liberation in the lung.
Although one can scarcely recommend regular alcohol consumption as a general prophylactic mechanism, it seems possible that there may be one or more drugs with inhibitory effects on inflammatory cells similar to those of alcohol but without its habituating effect and toxic effect on other organs. Before embarking on a search for such a drug, further studies should be carried out to confirm these observations and to show that inflammatory cells obtained from lungs of alcohol users are indeed different from those of nonusers. It has been postulated that the various inhibitory effects of alcohol on inflammatory cells may result from alteration of cell metabolism so that an excess of cyclic AMP accumulates in the cytoplasm. This effect has been shown to result from stimulation of adenylate-cyclase by levels of circulating alcohol commonly achieved in vivo. If such effects are verified in pulmonary cells, it should be possible to propose one or more potential prophylactic drugs. It may be hoped that an effective prophylactic drug might reduce risk of emphysema even below that accomplished by alcohol consumption.