Critique 181: Support for a causal role of alcohol consumption in improving the lipid profile — 16 February 2016
Vu KN, Ballantyne CM, Hoogeveen RC, Nambi V, Volcik KA, Boerwinkle E, Morrison AC. Causal Role of Alcohol Consumption in an Improved Lipid Profile: The Atherosclerosis Risk in Communities (ARIC) Study. PLoS One 2016;11:e0148765. doi: 10.1371/journal.pone.0148765. eCollection 2016.
INTRODUCTION: Health benefits of low-to-moderate alcohol consumption may operate through an improved lipid profile. A Mendelian randomization (MR) approach was used to examine whether alcohol consumption causally affects lipid levels.
METHODS: This analysis involved 10,893 European Americans (EA) from the Atherosclerosis Risk in Communities (ARIC) study. Common and rare variants in alcohol dehydrogenase and acetaldehyde dehydrogenase genes were evaluated for MR assumptions. Five variants, residing in the ADH1B, ADH1C, and ADH4 genes, were selected as genetic instruments and were combined into an unweighted genetic score. Triglycerides (TG), total cholesterol, high-density lipoprotein cholesterol (HDL-c) and its subfractions (HDL2-c and HDL3-c), low-density lipoprotein cholesterol (LDL-c), small dense LDL-c (sdLDL-c), apolipoprotein B (apoB), and lipoprotein (a) (Lp(a)) levels were analyzed.
RESULTS: Alcohol consumption significantly increased HDL2-c and reduced TG, total cholesterol, LDL-c, sdLDL-c, and apoB levels. For each of these lipids a non-linear trend was observed. Compared to the first quartile of alcohol consumption, the third quartile had a 12.3% lower level of TG (p < 0.001), a 7.71 mg/dL lower level of total cholesterol (p = 0.007), a 10.3% higher level of HDL2-c (p = 0.007), a 6.87 mg/dL lower level of LDL-c (p = 0.012), a 7.4% lower level of sdLDL-c (p = 0.037), and a 3.5% lower level of apoB (p = 0.058, P overall = 0.022).
CONCLUSIONS: This study supports the causal role of regular low-to-moderate alcohol consumption in increasing HDL2-c, reducing TG, total cholesterol, and LDL-c, and provides evidence for the novel finding that low-to-moderate consumption of alcohol reduces apoB and sdLDL-c levels among EA. However, given the nonlinearity of the effect of alcohol consumption, even within the range of low-to-moderate drinking, increased consumption does not always result in a larger benefit.
This paper is from a large number of subjects in the US-based Atherosclerosis Risk in Communities (ARIC) study. The analyses are based on repeated assessments of alcohol consumption and data on common and rare genetic variants in alcohol dehydrogenase and acetaldehyde dehydrogenase genes, which are known to affect the tendency of subjects to consume alcohol and perhaps modify its biologic effects.
Forum members considered the analyses used in this paper to be an innovative approach to help determine the degree to which the effects on lipids of reported alcohol consumption may be confounded by associated lifestyle factors. This approach should permit a better, unbiased assessment of alcohol as a “causative factor” for the improved lipid profiles generally found among moderate drinkers.
The results strongly support favorable effects of moderate drinking on most lipid values, specifically an increase in HDL-cholesterol and decreases for triglycerides and LDL-cholesterol. The association is non-linear, and further studies will be required to determine the specific amounts of alcohol associated with such lipid changes.
Reviewer Thelle commented: “This is an important paper to all of us, and particularly to me, as I co-authored one of the first papers to ‘rediscover’ alpha-lipoprotein cholesterol, later known as HDL-cholesterol, as a CHD protective factor (Miller et al). This was followed by a cross-sectional study where we showed that alcohol consumption might affect the HDL-cholesterol level (Thelle et al). The caveat is still whether HDL-cholesterol is the causative vehicle for the cardio-protective effect of alcohol (see, e.g., Magnus et al and Holmes et al). However, a few small RCTs of selected subjects (Chiva-Blanch et al B; Droste et al; Chiva-Blanch et al D) have also shown favorable results of alcohol on lipids that are consistent with those from the present study.”
Reviewer Ursini made a biochemical contribution: “From a biochemical point of view this set of results is reminiscent of a hormetic (or better para-hormetic) effect. Alcohol (better wine) in moderation activates the Keap1/Nrf2 signaling pathway. This increases the expression of ATP-binding cassette and thus the reverse cholesterol transport and finally the plasma level of HDL. Also the increased nucleophicic tone contributes–seemingly through a redox activation of APMK- to shift metabolism from anabolic to catabolic pathways. The expected outcome is a decrease of plasma value of cholesterol and triglycerides. Obviously, enzymes with different kinetic properties can give an account for individual differences.”
Forum member Sluik also contributed to the critique: “I also think this is a very well conducted study using a high quality study design. Mendelian Randomization reduces confounding and reverse causation bias, which are of course important factors to consider when studying alcohol consumption in relation to health. Furthermore, the authors consider the non-linearity of the association. That being said – when the authors additionally exclude never, former, and heavy drinkers to assess in a sensitivity analysis, it would also have been nice to exclude those with prevalent diseases that might have (recently) changed their alcohol habits. “Next, it would have been interesting to see a cross-tabulation of the observed alcohol consumption versus the predicted alcohol consumption. The authors have regressed the SNPs on alcohol consumption, but since the results are shown in categories of alcohol intake, I am curious about the overlap across the categories. Moreover, the authors might have reflected more on the differences and similarities in the associations between predicted and observed alcohol intake and lipid levels.” Forum member Ellison agreed that such data would have been useful for comparing these results with those from previous observational studies.
Forum member Keil stated: “The ARIC Study investigators have done a good job in their study. The meta-analysis by Brien et al reported favourable effects on the levels of HDL-cholesterol and on apolipoprotein A1 from alcohol consumption. The new study with ARIC data now contributes to our knowledge with the following findings: increased HDL2-c; reduced triglycerides, total cholesterol, LDL-c, sdLDL-c and apoB levels. This is a major step forward in our understanding of the mechanisms of the protective effect of light to moderate alcohol consumption on coronary heart disease.” Keil added: “Some lipidologists have been firmly convinced that alcohol consumption would increase triglyceride levels, which was not seen in the present study. Further, I am still not sure what the mechanisms are that are associated with increased HDL-cholesterol occurring both from alcohol consumption and from exercise.”
Reviewer Svilaas commented: “People with combined hyperlipidaemia (light to moderate elevated cholesterol, low HDL-C, elevated TG), or hypertriglyceridaemia get higher TG with alcohol consumption and have, to a certain degree, to be careful about drinking, partly because of the risk of pancreatitis.” Also, Forum member Van Velden pointed out that in their own studies they found an increase in triglycerides after the administration of red or white wine.
Using Mendelian Randomization to decrease confounding: Using Mendelian Randomization (MR), the investigators created a genetically modified alcohol value to test for an association with a variety of lipids. Supposedly, this adjusts for genetically determined factors and somehow still allows an unbiased assessment of the association between alcohol and lipids.
The authors point out that none of the SNPs (affecting ADH and ALDH) were correlated with potential confounders and that they chose a total of 5 SNPs that met the instrumental variable (IV) assumptions. The investigators state that the final genetic instruments were coded to ensure consistent effect direction of increasing alcohol consumption, and then were combined into an unweighted genetic risk score, which was used to fit the IV regression models.
Forum member Zhang provided some background information: “The basic idea of Mendelian Randomization is to use genetic variables as instrumental variables, i.e., genes associated with an exposure (here alcohol consumption or alcohol metabolism) but not directly related to outcome (here, HDL-cholesterol, etc.). Since the genetic pattern is determined before birth, it should not (at least, in theory) be confounded by later lifestyle exposures or outcome variables. Thus, it allows the investigators to make causal inference.
“The statistical method for analyzing an IV in relation to outcome is standard, using a 2-stage model. The IV approach is not equal to a randomized clinical trial (RCT), as it has its own problems. The main issue is the association between the IV variable and the exposure variable, akin to treatment assignment and actually having received the treatment (exposure) in a RCT. If the association is weak (or many subjects did not comply their treatment assigned in a RCT), then such an approach is highly model-dependent; if they are highly correlated, then it is great. Second, if an IV directly affects the outcome, then this method should not be used. If an IV is associated with an outcome through unmeasured confounding factors, then its value is doubtful. If it is associated with an outcome through measured confounders, then one can adjust for these through various approaches (multivariable analysis, matching, stratification, etc.).
“There is a two-step process for determining a valid IV using this technique. The first stage of an IV estimator involves regressing the phenotype of interest (X, e.g., alcohol) on the genetic IVs; for example, using three independent polymorphisms as IVs (Z1-Z3): E(X) = B0 + B1*Z1 + B2*Z2 + B3*Z3. The predicted value of X is used as the independent variable in a second stage regression with the outcome (Y, lipid values) as the dependent variable: E (Y) = A0 + A1* E (X). When external information about the association between the proposed genetic instruments and the phenotype is available, the predicted value of X in Equation 2 can be estimated based on this information.”
Problems with Mendelian Randomization when the instrumental variable is weak: While MR can be an important approach for seeking unbiased estimates of the effects of alcohol consumption, if the effects of the specific genetic factors on alcohol consumption are weak, little may be added to usual analyses. Forum member Ellison discussed this paper with several genetic epidemiologists in the Framingham Heart Study, who pointed out that the genetic score used in this paper was said to explain only approximately 0.1% variance of alcohol consumption; thus, it appears that the inclusion of MR in the analyses may have added little to their results. It was also noted that the authors used an unweighted genetic score in their analyses, whereas the data they present indicate that, of the 5 instrumental SNPs used, only one (rs2066702 influencing ADH1B) had much of an effect, so combining the others into an unweighted genetic score would not be expected to improve their results.
Previous extensive epidemiologic observational data from well-done cohort studies have clearly shown that moderate alcohol consumption is associated with a lower risk of cardiovascular disease (summarized well by Ronksley et al). Further, as noted below by Estruch, there have indeed been a number of important clinical trials of the effects of alcohol and wine on cardiovascular disease mechanisms (e.g., Chiva-Blanch et al A,B,C,D; Droste et al; Holmes et al). Brien et al have summarized the considerable previous data from clinical trials on the mechanisms of alcohol’s effects, and an huge number of animal experiments have produced results very similar to the effects seen in the present study
In describing their own studies, Forum member Estruch stated: “I think that results from randomized clinical trials have had stronger evidences that those obtained by Mendelian randomization approaches. In addition, cross-over intervention design studies help to better understand the effects of an intervention (alcohol, for instance), avoiding the possible effects of different genetic profiles, since the same individual receives two or more different interventions.
“We have performed several randomized cross-over intervention trials comparing the effects of alcoholized wine (regular wine), dealcoholized wine, and gin on different clinical and biochemical outcomes. In our experience, moderate consumption of red wine containing alcohol increased HDL cholesterol and Apo A1 to the same extent as gin, whereas dealcoholized red wine did not change the lipid profile. In these trials, plasma triglycerides did not change (Estruch et al)(Chiva-Blanch et al, A)(Chiva-Blanch et al, C) (Chiva-Blanch et al, D). On the other hand, we analyzed the effects of wine consumption on the prevalence of metabolic syndrome in 5801 participants in the PREDIMED trial and we found that red wine consumption, at doses of 0.1 – 1 drink a day, significantly decreased triglyceride concentration compared to that of subjects not consuming alcohol, OR 0.76 (95% CI 0.68 – 0.86)(P < 0.001), and at a dose of > 1 drink a day, triglycerides did not change: OR 0.87 (CI 0.71 – 1.06; P = 0.18)(Tresserra-Rimbau et al).”
Hence, despite potential weaknesses of the MR aspects of these analyses, the present study does provide important data indicating that light-to-moderate alcohol consumption has favorable effects on most lipid values. The effects shown for HDL-cholesterol support much earlier data, and the effects on small dense LDL-cholesterol and ApoB provide key information. The effects on lipids are surely important factors in the protective effects of moderate drinking against cardiovascular disease that have been seen in almost all well-done cohort studies.
References from Forum review
Brien SE, Ronksley PE, Turner BJ, Mukamal KJ, Ghali WA. Effect of alcohol consumption on biological markers associated with risk of coronary heart disease: systematic review and meta-analysis of interventional studies. BMJ 2011;342:d636; doi:10.1136/bmj.d636.
Chiva-Blanch G (A), Urpi-Sarda M, Llorach R, Rotches-Ribalta M, Guillén M, Casas R, Arranz S, Valderas-Martinez P, Portoles O, Corella D, Tinahones F, Lamuela-Raventos RM, Andres-Lacueva C, Estruch R. Differential effects of polyphenols and alcohol of red wine on the expression of adhesion molecules and inflammatory cytokines related to atherosclerosis: a randomized clinical trial. Am J Clin Nutr 2012;95:326-334. doi: 10.3945/ajcn.111.022889. Epub 2011 Dec 28.
Chiva-Blanch G (B), Arranz S, Lamuela-Raventos RM, Estruch R. Effects of wine, alcohol and polyphenols on cardiovascular disease risk factors: evidences from human studies. Alcohol Alcohol 2013;48:270-277. doi: 10.1093/alcalc/agt007. Epub 2013 Feb 13
Chiva-Blanch G (C), Urpi-Sarda M, Ros E, Valderas-Martinez P, Casas R, Arranz S . . . Estruch R. Effects of red wine polyphenols and alcohol on glucose metabolism and the lipid profile: a randomized clinical trial. Clin Nutr 2013;32:200-206. doi:10.1016/j.clnu.2012.08.022
Chiva-Blanch G (D), Magraner E, Condines X, Valderas-Martínez P, Roth I, Arranz S . . . Estruch R. Effects of alcohol and polyphenols from beer on atherosclerotic biomarkers in high cardiovascular risk men: a randomized feeding trial. Nutr Metab Cardiovasc Dis 2015;25:36-45. doi:10.1016/j.numecd.2014.07.008
Droste DW, Iliescu C, Vaillant M, Gantenbein M, De Bremaeker N, Lieunard C . . . Chioti A. A daily glass of red wine associated with lifestyle changes independently improves blood lipids in patients with carotid arteriosclerosis: results from a randomized controlled trial. Nutr J 2013;12:147. doi:10.1186/1475-2891-12-147.
Estruch R, Sacanella E, Mota F, Chiva-Blanch G, Antúnez E, Casals E, Deulofeu R, Rotilio D, Andres-Lacueva C, Lamuela-Raventos RM, de Gaetano G, Urbano-Marquez A. Moderate consumption of red wine, but not gin, decreases erythrocyte superoxide dismutase activity: a randomised cross-over trial. Nutr Metab Cardiovasc Dis 2011;21:46-53. doi:10.1016/j.numecd.2009.07.006. Epub 2009 Oct 12.
Holmes MV, Dale CE, Zuccolo L, Silverwood RJ, Guo Y, Ye Z . . . Consortium I. Association between alcohol and cardiovascular disease: Mendelian randomisation analysis based on individual participant data. BMJ 2014;349:g4164.
Magnus P, Bakke E, Hoff DA, Høiseth G, Graff-Iversen S, Knudsen GP . . . Mørland J. Controlling for high-density lipoprotein cholesterol does not affect the magnitude of the relationship between alcohol and coronary heart disease. Circulation 2011;124:2296-2302. doi:10.1161/CIRCULATIONAHA.111.036491
Miller NE, Thelle DS, Forde OH, Mjos OD. The Tromsø heart-study. High-density lipoprotein and coronary heart-disease: a prospective case-control study. Lancet 1977;1(8019):965-968.
Ronksley PE, Brien SE, Turner BJ, Mukamal KJ, Ghali WA. Association of alcohol consumption with selected cardiovascular disease outcomes: a systematic review and meta-analysis. BMJ 2011;342:d671; doi:10.1136/bmj.d671
Thelle DS, Shaper AG, Whitehead TP, Bullock DG, Ashby D, Patel I. Blood lipids in middle-aged British men. Br Heart J 1983;49:205-213.
Tresserra-Rimbau A, Medina-Remón A, Lamuela-Raventós RM . . . Estruch R, for the PREDIMED Study Investigators. Moderate red wine consumption is associated with a lower prevalence of the metabolic syndrome in the PREDIMED population. Br J Nutr 2015;113 Suppl 2:S121-130. doi: 10.1017/S0007114514003262.
The present paper is based on data from more than 10,000 Americans of European descent who were participants in the Atherosclerosis Risk in Communities (ARIC) study. Common and rare variants in alcohol dehydrogenase and acetaldehyde dehydrogenase genes were evaluated using Mendelian randomization (MR), and then a genetic score based on 5 SNPs was used in regression analyses for the association of alcohol with lipids. Results reported by the authors state: “Alcohol consumption significantly increased HDL2-c and reduced TG, total cholesterol, LDL-c, sdLDL-c, and apoB levels. For each of these lipids a non-linear trend was observed. Compared to the first quartile of alcohol consumption, the third quartile had a 12.3% lower level of TG (p < 0.001), a 7.71 mg/dL lower level of total cholesterol (p = 0.007), a 10.3% higher level of HDL2-c (p = 0.007), a 6.87 mg/dL lower level of LDL-c (p = 0.012), a 7.4% lower level of sdLDL-c (p = 0.037), and a 3.5% lower level of apoB (p = 0.058, P overall = 0.022).”
The authors thus conclude that “This study supports the causal role of regular low-to-moderate alcohol consumption in increasing HDL2-c, reducing TG, total cholesterol, and LDL-c, and provides evidence for the novel finding that low-to-moderate consumption of alcohol reduces apoB and sdLDL-c levels among European-Americans.”
Forum members considered this to be a well-done and important study, the results of which indicate favorable effects of moderate alcohol consumption on lipid factors. However, they doubt that the inclusion of the genetic score from MR played a large part in the results; the authors state that the genetic score used in this paper explained only approximately 0.1% variance of alcohol consumption, suggesting that it was a weak instrumental variable for the MR. Several Forum members stated that randomized control trials, especially cross-over trials, testing the effects of alcohol intake on risk factors, may provide even better evaluation of alcohol’s effects on lipids. Until large-scale RCTs of alcohol and disease occurrence can be are carried out, such RCTs may provide the most reliable data on mechanisms of alcohol’s effects on the risk of cardiovascular disease.
Overall, the present study provides important data indicating that light-to-moderate alcohol consumption has favorable effects on most lipid values. The effects shown for HDL-cholesterol support much earlier data, and the effects on small dense LDL-cholesterol and ApoB provide key new information. The effects on lipids are surely important factors in the protective effects of moderate drinking against cardiovascular disease that have been seen in almost all well-done cohort studies.
* * * * * * * * * *
Comments on this critique by the International Scientific Forum on Alcohol Research were provided by the following members:
Diewertje Sluik, DrPH, Division of Human Nutrition, Wageningen University, NL.
Yuqing Zhang, MD, DSc, Clinical Epidemiology, Boston University School of Medicine, Boston, MA, USA
Fulvio Ursini, MD, Dept. of Biological Chemistry, Universityof Padova, Padova, Italy
R. Curtis Ellison, MD, Professor of Medicine & Public Health, Boston University School of Medicine, Boston, MA, USA
Luc Djoussé, MD, DSc, Dept. of Medicine, Division of Aging, Brigham & Women’s Hospital and Harvard Medical School, Boston, MA, USA
Ulrich Keil, MD, PhD, Professor Emeritus, Institute of Epidemiology & Social Medicine, University of Muenster, Germany
Dag S. Thelle, MD, PhD, Senior Professor of Cardiovascular Epidemiology and Prevention, University of Gothenburg, Sweden; Senior Professor of Quantitative Medicine at the University of Oslo, Norway
David Van Velden, MD, Dept. of Pathology, Stellenbosch University, Stellenbosch, South Africa
Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway
Ramon Estruch, MD, PhD. Associate Professor of Medicine, University of Barcelona, Spain
Fulvio Mattivi, MSc, Head of the Department of Food Quality and Nutrition, Research and Innovation Centre, Fondazione Edmund Mach, in San Michele all’Adige, Italy
Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy
Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark
Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA
Creina Stockley, PhD, MSc Clinical Pharmacology, MBA; Health and Regulatory Information Manager, Australian Wine Research Institute, Glen Osmond, South Australia, Australia