Association between alcohol and cardiovascular disease: Mendelian randomisation analysis based on individual participant data
BMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g4164 (Published 10 July 2014) Cite this as: BMJ 2014;349:g4164
All rapid responses
According to the authors, “This approach, known as Mendelian randomisation, avoids some of the key limitations of observational studies, since allocation of genetic variants is random with regard to potential confounders.”
Yes, in order for an interpretation and conclusions of a Mendelian randomization study to be valid, the allocation of the genetic variants involved must be random with regard to potential confounders. However, in this case, they distinctively are not.
In the study’s Supplementary Figures, see Figure S1: ADH1B rs1229984 A-allele frequency in the 56 collaborating studies, arranged by geographic region. Note that the A-allele frequency is consistently about 20% in the studies of Southern European populations, about 10% in the studies from North America, and about 5% in the other studies from Europe. The A-allele proportions range from 37% of the population in the Cyprus study, to 20%, 19%, 17% and 18% of the Southern European populations in the IMPROVE Perugia, EPIC Turin, IMPROVE Milan, and Portuguese Stroke studies, respectively, to 1% in the two Finnish Kuopio studies and 3%, 4%, and 3% in the Swedish ULSAM, Improve Stockholm, and Nordic NORDIL studies. many of the 1% of Finns or 3% of Swedes who have the A-allele are likely of Southern European heritage--and have all the cultural and genetic confounders that go along with that. Which are, of course, totally unaccounted for in the study.
Clearly, the allocation of the A-allele variants is far from random. And so this analysis and its conclusion are simply not valid. Case closed. The authors and reviewers should be embarrassed.
Competing interests: No competing interests
I ought to be happy with this paper and its conclusions because I have contributed to the emergence of Mendelian Randomization [1] and because I oppose the promotion of alcohol for preventing CHD. [2] However, the lack of an association of HDL cholesterol levels with ADH1B genotype is worrying. HDL levels are consistently and universally associated with alcohol intake (cf. Figure S3 in the Supplementary Appendix) and are even used as a biomarker for intake. [3] So why do people with higher alcohol intakes not have higher HDL levels in this study?
In the study of Hines et al. [4] – which the authors did not discuss --, people with a genetically less active form of another alcohol dehydrogenase, ADH1C (a.k.a. ADH3), had a lower risk of CHD than those with the gene for the more active form. They also had higher HDL levels. This may explain why there was no association of HDL – and CHD? -- with ADH1B genotype in the present study: gram for gram, alcohol may have raised HDL more in those with an impaired ability to metabolize alcohol than in subjects with the more active form of the enzyme, compensating for the difference in intake.
Thus the lack of an effect of alcohol on CHD in this study could mean that the smaller amount of alcohol consumed by carriers of the minor allele has the same physiological effect as the larger amount consumed by the others.
There is also a possibility that something went wrong with data management and analyses. This is not a nice thing to say, and I have no concrete reason to believe that there are errors in this study, but those who have been involved in complicated data analyses know how easily things can go wrong. An independent re-analysis could lay this issue to rest.
1. Keavney B. Commentary: Katan’s remarkable foresight: genes and causality 18 years on. Int J Epidemiol 2004;33(1):11 –14.
2. Katan MB. [Alcohol should not be recommended for the prevention of heart disease]. Ned Tijdschr Geneeskd 2007;151(49):2717.
3. Berger D, Williams EC, Bryson CL, Rubinsky AD, Bradley KA. Alcohol questionnaires and HDL: screening scores as scaled markers of alcohol consumption. Alcohol Fayettev N 2013;47(6):439–45.
4. Hines LM, Stampfer MJ, Ma J, et al. Genetic Variation in Alcohol Dehydrogenase and the Beneficial Effect of Moderate Alcohol Consumption on Myocardial Infarction. N Engl J Med 2001;344(8):549–55.
Competing interests: No competing interests
The response submitted by Dr. Yedy Israel is a more accurate statement of my objections to this paper than I provided in my prior responses, which I retract in favor of his comments. However, I maintain that the adverse effects of light ethanol consumption on the health of persons with ADH mutations should not be extrapolated to the general public. Mendelian analysis does not overturn the substantial observational evidence that light ethanol ingestion is associated with enhanced cardiovascular health and longevity for the average person. I regret the factual errors I made regarding the underlying biochemistry, which Dr. Israel has set straight. The bottom line is that people who harbor the ADH mutation have an unpleasant reaction to even light ethanol consumption, but others can remain assured that, for them, observational evidence associates light ethanol consumption with enhanced cardiovascular health.
Competing interests: No competing interests
This article in The BMJ (Holmes MV et al. doi July 10, 2014) applies a recent technique, the Mendelian randomization analysis, to determine whether low alcohol consumption reduces cardiovascular disease. Cardiovascular disease was compared in two sets of subjects according to their alcohol intake. One of the groups carried a polymorphic gene coding for an alcohol dehydrogenase that metabolizes ethanol at a very fast rate, which deters subjects from alcohol intake. The second group carried the gene for the more common slow alcohol dehydrogenase. Subjects carrying the fast variant drank 20% less alcohol and showed less cardiovascular disease, at all levels of alcohol intake. This is expected when reducing a high intake of ethanol, a drug known to be cardiotoxic, but it was unexpected when the intake of very low levels was reduced, as it has been proposed that at low intakes, ethanol has a protective effect on cardiovascular disease. Thus, these studies could be taken as a demostration that the putative protective effect of low ethanol intake on cardiovascular disease actually does not exist, and that ethanol is cardiotoxic at any level. This may indeed be the case, as there are very few controlled prospective studies where low alcohol levels are administered for prolonged times, versus total abstinence.
However, a word of caution is needed. The gene examined codes for an alcohol dehydrogenase that is 40-times more active than the control enzyme; a high activity leads to increases in acetaldehyde (as indeed implied by the authors) . One can ask if the higher levels of acetaldehyde in these subjects counter any putative protective cardiovascular effect of low ethanol intakes. It is noted that two variables change in this study (acetaldehyde and ethanol intake) while only one outcome is measured (cardiovascular disease). Changes in cardiovascular disease are purported to be due to alcohol intake. It is also assumed (vide infra) that these data applies to 95% of Caucasians who carry the gene that codes for the slow, rather than the fast alcohol dehydrogenase. Acetaldehyde has been proposed by many authors to have tissue deleterious effects and to bind to DNA, as well as binding to proteins, generating neoantigens that correlate with organ damage. This damage might start at lower levels of ethanol consumption in a population that carries the more active alcohol dehydrogenase, which generates acetaldehyde at a higher rate. Without taking issue as to whether low ethanol intake protects or does not protect from cardiovascular disease, a word of caution is needed vis-a-vis the extrapolation of these findings. In their highlighted chart (p 10) the study indicates: “Under the principles of mendelian randomisation, the(se) findings suggest that reduction of alcohol consumption, even for light to moderate drinkers, is beneficial for cardiovascular health”. This recommendation is certainly correct for a small subset of Caucasian (European) subjects who carry the fast alcohol dehydrogenase-coding gene; but generalizing this recommendation to other populations requires caution
Competing interests: No competing interests
In the study by Holmes et al. it is concluded that “the reduction of alcohol consumption, even for light to moderate drinkers, is beneficial for cardiovascular health” [1].
However in Fig. 2 there are no statistically significant effects for the groups of "light drinkers", "moderate drinkers" and "heavy drinkers" when these groups are taken separately (the 95% confidence intervals cross the "no effect" line).
A statistically significant effect is reported for the broader group of "drinkers only". What confuses me is that the average protective effect for the "drinkers only" group is larger than the average effect of any of the three following groups: "light drinkers", "moderate drinkers" and "heavy drinkers". This should not be so.
From Fig. 2 we see the "light drinkers", "moderate drinkers" and "heavy drinkers" groups can be combined to a total of 9827 cases and 97243 individuals. In the "drinkers only" group there are 10130 cases and 107478 individuals. This leaves us with some unknown group (let’s call it X).
Group X consists of very strange people who are classified as "drinkers only" but not as "light drinkers", "moderate drinkers" or "heavy drinkers" and in this group there are 303 cases and 10235 individuals.
The proportion of cases to the number of individuals is as follows:
Non-drinkers: 5883/43029 ~ 0,137
Light: 4686/47246 ~ 0,099
Moderate: 3222/33772 ~ 0,095
Heavy: 1919/16225 ~ 0,118
Group X: 303/10235 ~ 0,03
This means that this group X has 3-4 times less cases per individual than any other mentioned group. When this strange group is mixed with the “light”, “moderate” and “heavy” drinkers the “protective effect” appears to increase so dramatically for the group of “drinkers only” that the effect becomes statistically significant and stronger (by average effect size) than in any of the three groups (of "light", "moderate" and "heavy" drinkers).
This suggests the presence of a discrepancy in the data that could have led to an erroneous conclusion.
1. Holmes MV, Dale CE, Zuccolo L, et al. Association between alcohol and cardiovascular disease: Mendelian randomisation analysis based on individual participant data. Bmj 2014;349:g4164 doi: 10.1136/bmj.g4164 (Published 10 July 2014)
Competing interests: No competing interests
People who have loss-of-function mutations in their gene for alcohol dehydrogenase (ADH), whom I shall refer to as mutants, do not metabolize ethanol as efficiently as people with 2 fully-functional wild-type ADH alleles, whom I shall refer to as normals. Because of their impaired metabolism of ethanol, mutants experience a given dose of ethanol as equivalent, in effect, to a larger dose of ethanol ingested by normals. For normal Caucasian men, the consistently-observed minimum risk for cardiovascular disease occurs with a chronic daily ingestion of about 1 to 2 ounces of ethanol per day. If a mutant metabolizes ethanol with (for example) 10% of the normal ADH enzymatic efficiency, then he would require only about 10% of the normal's optimal ethanol dose to obtain the same cardiovascular benefit. Therefore, it is a fallacy to try to estimate the effects of ethanol ingestion on normals by studying its effect on persons whose alcohol metabolism is significantly reduced compared with normals.
In summary, persons with loss-of-function ADH mutations are not an accurate surrogate for persons with wild-type ADH who consume low levels of ethanol, because the ethanol consumption levels of mutants must be corrected for by the degree to which their ADH mutation impairs ethanol metabolism efficiency. The results of this study only apply to ADH mutants and cannot be projected to normals. All other factors being equal, ADH mutants will probably be found to have an optimal ethanol dose for cardiac protection equal to 1 to 2 ounces multiplied by a factor which corrects for the impaired rate constant of their mutated ADH enzyme. In the example I gave of a mutant ADH with only 10% of normal function, the optimal dose of ethanol would therefore be about 0.1 to 0.2 ounces of ethanol per day.
Competing interests: No competing interests
People who have the rs1229984 mutation in their alcohol dehydrogenase (ADH) gene ("mutants") do not metabolize ethanol as efficiently as people with 2 copies of the wild-type ethanol dehydrogenase allele ("normals"). Because of their impaired metabolism of ethanol, mutants experience a given dose of ethanol as equivalent, in effect, to a larger dose of ethanol ingested by normals. For normals, the observed minimum risk for cardiovascular disease occurs with a chronic daily ingestion of 1 to 2 ounces of ethanol per day. If a mutant metabolizes ethanol with (for example) 10% of the normal ADH enzymatic rate constant, then he would require only 10% of the normal's optimal ethanol dose to obtain the same cardiovascular benefit. Because mutants experience flushing and other adverse effects of ethanol at low doses, they are less likely than normals to drink enough to cause direct cardiotoxic effects on the myocardium. They may even generate enough endogenous ethanol (via other metabolic pathways) to achieve their optimum dose for maximal cardioprotective effect without ingesting any exogenous ethanol at all. In summary, these mutants are not an accurate surrogate for consumers of low levels of ethanol, after their consumption levels are corrected for their impaired ethanol metabolism efficiency.
Competing interests: No competing interests
155 authors for only one scientific paper are ridiculous. Only 34 references mean that less than every fifth of the authors had to read only one reference article. Every author had to elaborate only 30 words to gain a publication volume of 4650 words? And that the Mendelian randomization meta-analysis was founded on 56 epidemiological studies is absolutely contradictory with 34 references only!
The scientific basis is very weak. It is an 'ex post' Mendelian randomization meta-analysis without any preexisting scientific hypothesis and illogical conclusions. People with a genetic disorder of alcohol metabolism were to be taken as testimony for lower cardiovascular morbidity in comparison with inadequately differentiated alcohol- consuming people with regular alcohol metabolism.
This is like testing albino rats against normally pigmented rats in a randomized controlled test design (RCT) with UV-light and wondering why they won't get sun-tanned.
Sincerely Yours
Dr. med. Thomas G. Schaetzler
Family Medicine Unit
24 Kleppingstr.
D 44135 Dortmund/Germany
th.g.schaetzler@gmx.de
Competing interests: None declared
Competing interests: No competing interests
ETHANOL: A CARCINOGENIC SUBSTANCE FOR THE PREVENTION OF CARDIOVASCULAR DISEASE? AN ETHICAL QUESTION.
Gianni Testino (hepatologist), Silvia Leone (under training toxicologist) and Paolo Borro (hepatologist)
Centro Alcologico Regionale – Regione Liguria, UO Alcologia, Dept. of General Internal and Specialistic Medicine, IRCCS AOU San Martino- National Institute for Cancer Research-IST, Genova, Italy
University of Genova, Genova, Italy
We read with great interest the excellent study by Holmes et al. (1). The authors concluded that, ‘the reduction of alcohol consumption, even for light to moderate drinkers, is beneficial for cardiovascular health’.
This detailed study seriously calls into question the moderate consumption of alcohol in the prevention of cardiovascular diseases. The last few studies in the literature have criticised the methodological correctness of some research.
Furthermore, we must point out that the moderate doses commonly indicated as protectors (10-12 g / day for women, 20-25 g / day for men) lead to other diseases (e.g. hypertension), and cancer in particular.
It is widely recognised and proven that acetaldehyde and ethanol are carcinogenic for humans. Ethanol, acetaldehyde and alcoholic beverages are included in the IARC (International Agency for Cancer Research) Group 1 (WHO) (2).
Moreover, an evaluation by the Model of Exposure (MOE) identified a substance in ethanol with a high carcinogenic potential (3).
Alcohol consumption raises the risk of the following cancers: oral cavity, pharynx, larynx, esophagus, liver, colon, rectum and breast. It also increases the risk of pancreatic cancer at high doses.
Some tumors can occur even with low amounts of alcohol (approximately 10 g /day): oral cavity, pharynx, larynx, esophagus and breast. The risk of breast cancer in particular increases with these amounts by 5-7%.
It is clear, therefore, that alcohol intake of 10 g/day for women and 20 g/day for men should no longer be considered ‘recommended dosages’ or ‘safe dosages’, but at the very least dosages with a ‘low risk’ of carcinogenic effects.
A comprehensive review of the literature shows that several authors and the WHO have repeatedly stated that the possible prevention of one disease while favouring another is not to be considered as medical progress.
Currently, scientific institutions all over the world (Institute National du Cancer in France, the World Cancer Research Fund, the American Institute for Cancer Research, the Association of European Cancer Leagues, the Cancer Council of Australia, the Istituto Superiore di Sanità in Italy; the WHO, etc.) argue strongly that the risk of cancer is dose-dependent and that there is, therefore, a threshold level of safety (4).
The data from Holmes et al. (1) help us to further strengthen our idea: it is not helpful and even unethical to use a toxic and a potentially carcinogenic substance to prevent coronary arterial diseases. Sooner or later those who speak the truth are discovered!
Physicians and cardiologists should inform their patients that even if moderate doses of alcohol reduce the risk of cardiovascular disease, at the same time they can increase the risk of cancer. They will just have to follow common sense advice and safety tips such as smoking cessation, weight reduction, reduction of salt intake, a balanced diet and proper physical exercise (5).
Competing Interest: none declared
References
1) Holmes MV, Dale CE, Zuccolo L, Silverwood RJ, Guo Y, Ye Z et al. Association between alcohol and cardiovascular disease: Mendelian randomisation analysis based on individual participant data. BMJ 2014; 349: g4164; doi: 10.1136/bmj.g4164
2) IARC. A review of human carcinogens. IARC Monogr Eval Carcinog Risks Hum 2012; 100 part E: 377-478
3) Lachenmeier DW, Przybylski MC, Rehm J. Comparative risk assessment of carcinogens in alcoholic beverages using the margin of exposure approach. In J Cancer 2012; 131: E995-1003; doi: 10.1002/ijc.27553
4) Nelson DE, Jarman DW, Rehm J, Greenfield TK, Rey G, Kerr WC et al. Alcohol-attributable cancer deaths and years of potential life lost in the United States. Am J Public Health 2013; doi: 10.2105/AJPH.2012.301199
5) World Health Organization. Regional Office for Europe. Alcohol in the European Union. Consumptiom, harm and policy approaches, 2012
Competing interests: No competing interests
Re: Association between alcohol and cardiovascular disease:
May I first offer two comments -
1. Almost a year has gone by since the paper was published.
There is still no response from the authors to the responses (over a dozen), many very critical.
2. It is pleasing (to me) that the authors have avoided the term " Caucasian" which is bandied about in epidemiology these days, without definition. They have used the phrase " European descent" .
Unfortunately the definition of Europe is also a little awkward.
Are the Sephardics classified as European, in this study? The Ashkenazis are, I assume.
Further, the Italians and the Maltese must have, over the centuries, taken in a fair amount of chromosomes which originated from the Maghreb (which is not quite Europe).
May I now ask two questions:
1.Would the authors of the paper care to reply to the comments so far?
2. Would the BMJ care to publish the Reviewers' views on the points made by the Rapid Responders so far?
Thank you
Competing interests: No competing interests