Würtz P, Cook S, Wang Q, Tiainen M, Tynkkynen T, Kangas AJ, et al. Metabolic profiling of alcohol consumption in 9778 young adults. Int J Epidemiol 2016; pre-publication: doi: 0.1093/ije/dyw175
Background: High alcohol consumption is a major cause of morbidity, yet alcohol is associated with both favourable and adverse effects on cardiometabolic risk markers. We aimed to characterize the associations of usual alcohol consumption with a comprehensive systemic metabolite profile in young adults.
Methods: Cross-sectional associations of alcohol intake with 86 metabolic measures were assessed for 9778 individuals from three population-based cohorts from Finland (age 24–45 years, 52% women). Metabolic changes associated with change in alcohol intake during 6-year follow-up were further examined for 1466 individuals. Alcohol intake was assessed by questionnaires. Circulating lipids, fatty acids and metabolites were quantified by high-throughput nuclear magnetic resonance metabolomics and biochemical assays.
Results: Increased alcohol intake was associated with cardiometabolic risk markers across multiple metabolic pathways, including higher lipid concentrations in HDL subclasses and smaller LDL particle size, increased proportions of monounsaturated fatty acids and decreased proportion of omega-6 fatty acids, lower concentrations of glutamine and citrate (P<0.001 for 56 metabolic measures). Many metabolic biomarkers displayed U-shaped associations with alcohol consumption. Results were coherent for men and women, consistent across the three cohorts and similar if adjusting for body mass index, smoking and physical activity. The metabolic changes accompanying change in alcohol intake during follow-up resembled the cross-sectional association pattern (R2=0.83, slope=0.7260.04).
Conclusions: Alcohol consumption is associated with a complex metabolic signature, including aberrations in multiple biomarkers for elevated cardiometabolic risk. The metabolic signature tracks with long-term changes in alcohol consumption. These results elucidate the double-edged effects of alcohol on cardiovascular risk.
Observational epidemiologic studies relating alcohol consumption to health and disease have been amazingly consistent over many decades: light to moderate alcohol intake is related to improved cardiovascular health and less diabetes, while heavy intake and binge drinking relate primarily to adverse cardiovascular and other disease outcomes. A multitude of animal experiments have identified many mechanisms by which moderate drinking affects cardiovascular disease, including beneficial effects on lipids, coagulation factors, fibrinolysis, glucose metabolism, inflammation, and endothelial function. The mechanisms for a direct effect of heavy drinking on upper airway cancers and liver disease have also been delineated, and suggested mechanisms given for the slight increase in breast cancer risk from even moderate drinking seen in most epidemiologic studies. However, the specific metabolic effects of alcohol have been little studied. The investigators for the present study have collected a large amount of metabolic data from population-based cohorts of relatively young and healthy men and women in Finland. Their analysis provides a wealth of information by relating such data to the reported alcohol intake of subjects.
Contributions of the present study: For the present study, data from a group of almost 10,000 young adults from population-based cohorts in Finland have been evaluated; the investigators have measured associations between self-reported alcohol consumption and a large number of lipid parameters, indices of fat intake, glutamine, citrate, and 56 metabolic measures. The investigators found, as expected, that total HDL-cholesterol increases linearly with reported alcohol intake. In addition, they found that alcohol intake was associated with higher lipid concentrations in HDL sub-classes and smaller LDL particle size, increases in monounsaturated fatty acids and decreases in omega-6 fatty acids, lower concentrations of glutamine and citrate, and a large number of other metabolic effects (many being u-shaped, according to the level of alcohol intake). The findings of this study provide valuable clues to the biologic effects on health, both favorable and adverse, related to alcohol consumption. Further, the paper could also provide an interesting new approach for estimating the level of alcohol intake of individuals in epidemiologic studies, so that alcohol intake is judged not just on what the individual subject self-reports.
Improving our understanding of alcohol’s effects on health: The authors have summarized very well the potential importance of these analyses: “The detailed metabolic phenotyping further clarified the association shapes for numerous established biomarkers related to alcohol and cardiometabolic risk. The metabolic signature of alcohol consumption included molecular perturbations linked with both higher and lower cardiovascular risk. Many metabolic measures displayed an optimum level at modest alcohol intake.” They conclude: “Comprehensive metabolic profiling in these large cohorts elucidated the metabolic influence of alcohol consumption and clarified the double-edged relation between alcohol and cardiometabolic biomarkers.”
Forum member Barrett-Connor commented: “I think this is an excellent paper, based on data from three different population-based cohorts for whom outcome data are available (the great advantage of having personal identifiers in a country). It is a meta-analysis and includes both cross-sectional and, for a subset, longitudinal data that further validates their reported alcohol intake. Associations are reported between alcohol intake and lipids, circulating fatty acid levels, and many other metabolites. The authors note that the dietary determinants of the circulating biomarkers (especially amino acids and other small molecules) remain poorly understood. The most prominent associations were for citrate and glutamine–both strongly inversely associated with alcohol but otherwise weakly correlated with established CVD risk factors.”
Reviewer Finkel was also pleased by this paper. “I particularly note the youth of the subjects, which brings to mind one of the old-time tenets of drinking’s oft-expressed health benefits: drinking moderately was usually said to be associated with cardiovascular and survival benefits only for middle-aged and older people, whereas this paper shows strong relations between alcohol and biomarkers among the young. Atherosclerosis is a chronic, slowly building disease: its clinical appearance in middle or old age surely means it started long before.”
Forum member Ursini was less impressed by these analyses. “As a basic scientist, interested on mechanisms and metabolism, I must admit I did not learn too much. Are we going to discover new biomarkers? Citrate and glutamate will become the new cholesterol? This type of well-organized fishing expedition without an educated working hypothesis will never solve relevant medical issues unless validated by a severe experimental approach.”
Reviewer Thelle noted: “I also am somewhat skeptical of the cross-sectional aspects of this study, while admitting that some such studies have provided the basis for new hypotheses. And that is the major contribution from this paper; it ignites the interest for experimental studies, and of course the inclusion of new variables in prospective studies.”
Reviewer Skovenborg tended to agree with the views of Ursini and Thelle, stating “Results from this cross-sectional study may best be used to generate new hypotheses regarding cardio-vascular risk factors and maybe candidates for laboratory confirmation of data on alcohol intake. I question whether the results from these Finnish populations – from a country with binge drinking traditions – are generalizable to populations with a Mediterranean drinking pattern. Also, I am not clear as to why participants with alcohol consumption above 500 g/week were excluded due to the likelihood that these intake volumes were spurious. Most literature regarding people with heavy alcohol consumption would not consider 6-7 drinks/day as improbable.”
Forum member Lanzmann-Petithory had some cogent comments on the lipid changes noted in this study. “Reading this paper I was struck by the fatty acid profiles in plasma in the cross-sectional association with alcohol. The pattern with alcohol shown in the present study is very similar to that from the intervention group given canola oil in the Lyon Diet Heart study (de Lorgeril et al; Renaud et al), i.e., increased MUFA and decreased linoleic acid. The Lyon study showed a large degree of cardiovascular protection from more mono-unsaturated, more omega 3 fatty acids and less linoleic acid. It is unclear why the essential fatty acid profile should be the same in Finnish drinkers: could there be a metabolic interaction of alcohol with diet or a confounding with diet?
“We know that the diet in Finland has turned towards a Mediterranean-type diet from the beginning of the North Karelia Project in 1972; rapeseed oil has become widely used, including in the margarine industry (Pietinen et al). During recent decades, there has also been a rapid decline in coronary heart disease in Finland that remains unexplained by classical risk factor correction (Laatikainen et al). The group led by Federico Leighton in Chile carried out clinical trials investigating the effect of wine on plasma fatty acids; they found that wine improved fatty acid profile in subjects following a Mediterranean diet but not in subjects with more of an Occidental one (Urquiaga et al). Furthermore, we know from statistics of IREB that Finland turned more and more towards wine consumption: between 1961 and 2001, the consumption of pure alcohol soared from 2 liters to 7.4 liters /capita /year, and wine consumption between 1990 and 2001 from 6.4 liters to 20.1 liters/capita/year. Still, we do not know why the fatty acid profile associated with alcohol intake in Finland was so similar to that shown in France from increased canola oil. It deserves more investigation.”
Can the data in this study be used to help provide an accurate assessment of alcohol consumption of individuals? In addition to providing information on potential mechanisms by which alcoholic beverages affect health, reviewer Ellison has brought up another potential use of these data: can they be used to help scientists get a better estimate of the alcohol intake of individuals? “In almost all observational studies, estimates of the alcohol intake of individuals come from self-report, from food diaries or specific questions regarding beverages consumed within the preceding days, weeks, or years. Given that it is generally assumed that some people under-estimate their intake, the ability to obtain a more objective measure could be very helpful in seeking to relate alcohol intake to health and disease.
“Klatsky and colleagues (2006, 2014) have described a method for using multiple medical records from individuals to help identify those who are, or are not, under-reporting their alcohol intake (those deemed likely to be under-reporters showing more adverse effects of moderate drinking). Further, other investigators (Boniface et al, Livingston et al) have described common characteristics of subjects who tend to under-report their intake. Such approaches should improve epidemiologic studies by providing better assessments of actual alcohol intake. Another approach for estimating intake involves the relation of genetic factors that affect alcohol consumption to disease outcomes, using what is known as Mendelian randomization technique to find a genetic surrogate measure of intake (Smith & Ebrahim). Unfortunately, there are currently not available simple genetic factors that are specific enough to serve as an accurate estimate of intake, and the limited Mendelian studies thus far reported provide incomplete information on alcohol effects on health and disease (e.g., Holmes et al, which was reviewed by our Forum; available at www.bu.edu/alcohol-forum/critique-143). Some Forum members wondered if a set of lipids, metabolites, etc., shown in this study to relate strongly to self-reported intake, could be used to help determine the accuracy of self-reported drinking.”
Forum member Klatsky reviewed the present paper and commented on this potential use of the results of this study: “Although my knowledge of it is limited, my impression from the considerable literature about the use of biomarkers to estimate alcohol intake in individuals is that estimation of recent intake is pretty good, but chronic intake is another matter. I’m also sure that the fact of under-estimation (“yes” or “no”) in an individual that claimed abstinence can quite accurately be determined by just a few tests, or even one. In terms of screening for epidemiological surveys, one would be limited to the available data. I know that clinicians in Kaiser Permanente that evaluated examinees undergoing multiphasic screening felt that, in the absence of other explanations, the triad of high HDL, elevated AST, and high MCV was a highly specific marker for alcoholism. Sensitivity was presumably much lower. But even these 3 tests were not consistently available over the years when screening was widely performed.”
Klatsky continued: “We used available AST and ALT values, available in a subgroup of examinees, to validate our suspected under-reporter concept in our study of that phenomenon, but didn’t use labs as part of the definition. So I think your concept is sound, but wonder whether there are available cohorts that have the relevant data. Also, how does one validate the fact of under-reporting?”
Reviewer Zhang commented that “While many such endeavors are for dichotomous outcomes (i.e., yes or no for risk of a specific disease), others have attempted to determine the predictive ability for continuous outcome variables (e.g., Ellison et al). Thus, it is certainly doable in theory to use these metabolites to predict alcohol consumption; however I am not sure what the purpose is of doing it. If the current ‘gold standard’ of alcohol consumption is self-reported intake (and this paper is using that to relate to metabolic outcomes), why not just obtain our estimates through questionnaire; why should we bother to get these biomarkers and use them to predict alcohol consumption? Among non-drinkers, it is likely that distribution of these metabolites may vary. Sometimes, X may be a good predictor for Y, but that does not mean that Y will be a good predictor for X. In other words, while self-reported alcohol consumption (which is susceptible to misclassification bias) is associated with levels of these metabolites, it does not mean that these metabolites would predict self-reported alcohol consumption. Other factors, such as dietary pattern, may affect metabolites as well; in fact, there is a strong association between smoking and alcohol intake.”
Reviewer Ellison argued: “Any additional data on potential biases for self-reported data, whether based on genetic, behavioral, laboratory, or other information (preferably on all), might allow epidemiologists to have more precise and accurate information on alcohol intake when relating such to disease outcomes. As noted, Klatsky et al previously showed that among ‘moderate drinkers,’ data from those deemed unlikely to be under-reporters showed no significant effect of alcohol on blood pressure or risk of cancer; those deemed to be under-reporters showed adverse effects of their supposedly ‘moderate drinking’ on these outcomes. Indeed, a more accurate assessment of exposure could benefit our analyses.” Added reviewer Thelle: “What comes to my mind is an extensive alcohol experiment taking into consideration genetic variation, and assessing the effects on all the variables mentioned in the paper (plus some others as well). Feasibility and ethics may be questionable, but still, the idea deserves to be mentioned.”
References in Forum critique
Boniface S, Kneale J, Shelton N. Drinking pattern is more strongly associated with under-reporting of alcohol consumption than socio-demographic factors: evidence from a mixed-methods study. BMC Public Health 2014;14:1297. DOI: 10.1186/1471-2458-14-1297
de Lorgeril M, Renaud S, Mamelle N, Salen P, Martin J-L, Monjaud I, Guidollet J, Touboul P, Delaye J. Mediterranean alpha-linolenic acid-rich diet in secondary prevention of coronary heart disease. Lancet 1994;143:1454–1459.
Ellison RC, Zhang YQ, Qureshi MM, Knox S, Arnett DK, Province MA. Lifestyle determinants of high-density lipoprotein cholesterol: the National Heart, Lung, and Blood Institute Family Heart Study. Am Heart J 2004;147:529-535.
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.
Klatsky AL, Gunderson E, D G, Kipp H, Udaltsova N, Friedman GD. Higher prevalence of systemic HTN among moderate alcohol drinkers: exploring the role of under-reporting. J Stud Alcohol 2006;67:421–428.
Klatsky AL, Udaltsova N, Li Y, Baer D, Nicole Tran H, Friedman GD. Moderate alcohol intake and cancer: the role of underreporting. Cancer Causes Control 2014;25:693-699. doi: 10.1007/s10552-014-0372-8.
Laatikainen T, Critchley J, Vartiainen E, Salomaa V, Ketonen M, Capewell S . Explaining the decline in coronary heart disease mortality in Finland between 1982 and 1997. Am J Epidemiol 2005; 162:764-773.
Livingston M, Callinan S. Underreporting in alcohol surveys: whose drinking is underestimated? J Stud Alcohol Drugs 2015;76:158-164
Pietinen P, Vartiainen E, Seppänen R, Aro A, Puska P. Changes in diet in Finland from 1972 to 1992: impact on coronary heart disease risk. Prev Med 1996;25:243-250.
Renaud S, M de Lorgeril M, Delaye J, et al. Cretan Mediterranean diet for prevention of coronary heart disease. Am J Clin Nutr 1995;61(Suppl):1360S-1367S.
Smith GD, Ebrahim S. ‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease? International Journal of Epidemiology 2003;32:1–22. DOI: 10.1093/ije/dyg070.
Urquiaga I, Guasch V, Marshall G, San Martín A, Castillo O, Rozowski J, Leighton F. Effect of Mediterranean and Occidental diets, and red wine, on plasma fatty acids in humans. An intervention study. Biol Res 2004;37:253-261.
While observational epidemiologic studies for many decades have consistently shown that moderate drinkers have a lower risk of cardiovascular diseases than do non-drinkers or heavy drinkers, the specific metabolic effects of alcohol have been little studied. In the present analysis, among almost 10,000 young adults from three population-based cohorts in Finland, associations of alcohol intake with 86 metabolic measures were assessed. Circulating lipids, fatty acids and metabolites were quantified by high-throughput nuclear magnetic resonance metabolomics and biochemical assays. The investigators found that alcohol consumption was associated with a complex metabolic signature, including aberrations in multiple biomarkers for reduced as well as elevated cardiometabolic risk; many factors showed different associations according to the estimated amount of alcohol consumed.
Among key associations found for greater alcohol intake were increases in HDL-cholesterol and its subclasses, decreases in LDL size, an increase in monounsaturated fatty acids and a decrease in omega-6 fatty acids, and lower concentrations of glutamine and citrate. For unexplained reasons, the changes in fatty acids from alcohol in this study were similar to those occurring following the administration of canola oil in other research. Some Forum reviewers pointed out that the analyses were not theory driven, and should only be used to generate hypotheses that would need to be tested in future experiments. As stated by one reviewer, the major contribution from this paper could be that it ignites the interest for experimental studies and provides new variables to be evaluated in prospective studies. In any case, the findings of this study provide valuable clues to the biologic effects on health, both favorable and adverse, related to alcohol consumption.
While not discussed by the authors, Forum members considered that the results of this paper might also be useful in providing a new approach for judging the level of alcohol intake of individuals in epidemiologic studies. At present, alcohol intake is judged almost exclusively from self-reports by subjects. Previous attempts designed to identify subjects more likely to be under-reporting their intake have shown that they may sharpen relations between estimated intake and health outcomes. Forum members believe that approaches that identify sources of bias for self-reported data, whether based on genetic, behavioral, physiological, or other information (preferably on all), might allow epidemiologists to have more precise and accurate information on alcohol intake when relating such to disease outcomes. A new approach for providing more accurate and unbiased estimates of alcohol intake is suggested by this excellent analysis.
* * * * * * *
Comments on this critique by the International Scientific Forum on Alcohol Research were contributed by the following members:
Elizabeth Barrett-Connor, MD, Distinguished Professor, Division of Epidemiology, Department of Family Medicine and Public Health and Department of Medicine, University of California, San Diego, La Jolla, CA USA
R. Curtis Ellison, MD, Professor of Medicine & Public Health, Boston University School of Medicine, Boston, MA, USA
Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA
Arthur Klatsky, MD, Dept. of Cardiology, Kaiser Permanente Medical Center, Oakland, CA, USA
Dominique Lanzmann-Petithory, MD, PhD, Nutrition Geriatrics, Hôpital Emile Roux, APHP Paris, Limeil-Brévannes, France
Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark
Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway
Dag S. Thelle, MD, PhD, Department of Biostatistics, Institute of Basic Medical Sciences,
University of Oslo, Norway; Section for Epidemiology and Social Medicine, Sahlgrenska Academy, University of Gothenburg, Sweden
Fulvio Ursini, MD, Dept. of Biological Chemistry, University of Padova, Padova, Italy
Yuqing Zhang, MD, DSc, Clinical Epidemiology, Boston University School of Medicine, Boston, MA, USA