Excessive Consumption of Alcoholic Beverages and Extremely High Levels of High-Density Lipoprotein Cholesterol (HALP) in the ELSA-Brasil Cohort Baseline

Background: It has already been established that the consumption of alcoholic beverages increases high-density lipoprotein cholesterol (HDL-C) levels in dose–response. Methods and Results:A cross-sectional analysis was carried out with 6132 participants of both sexes aged between 35 and 74 years, who were active and retired workers from six Brazilian states. Heavy drinkers were categorized by sex: men > 210 g/week and women > 140 g/week; moderate drinkers: men ≤ 209 g/week and women ≤ 139 g/week. The HDL-C level was dichotomized into normal (40 mg/dL–82.9 mg/dL) and extremely high (≥83 mg/dL). We used binary logistic regression to assess associations between baseline alcohol intake and HDL-C, which were adjusted for sex, age, income, physical activity, kilocalories and body mass index (BMI), and we found an positive association between extremely high HDL-C and the excessive consumption of alcoholic beverages. These participants were mostly women with a high income, lower waist circumference, kilocalorie consumption and also a higher consumption in all categories of alcoholic beverages. Conclusion: Excessive alcohol consumption was associated with a higher probability of extremely high HDL-C.


Introduction
The consumption of alcoholic beverages increased from 6.4 to 6.6 for the years 2016 to 2020 in per capita consumption in L, with estimates reaching 7.0 L for 2025 [1,2]; other estimates show that by 2030, the world per capita consumption will have reached 7.6 L, and the proportion of drinkers will increase by 0.22% annually, defined as a public health problem [3].
High-density lipoproteins (HDLs) have been considered as the 'good cholesterol' that brings benefits to the body, mainly for cardiovascular health [4], through activity in the cholesterol efflux and anti-inflammatory, anti-oxidative, anti-thrombotic and anti-apoptotic characteristics [5][6][7].
Serum levels of HDL-C have been extensively investigated in relation to cardiovascular health. Low HDL-C levels have been associated with the incidence of cardiovascular diseases (CVDs) [8,9], leading to recommendations to increase this lipoprotein [10]. Currently, for ELSA-Brasilarecurrent or recent (<4 months prior to the first interview) pregnancy, the intention to quit working at the institution in the near future, severe cognitive or communication impairment and, if retired, residence outside of a study center's corresponding metropolitan area. ELSA-Brasil already has data from 3 waves: wave 1 (2008-2010), wave 2 (2012-2014) and wave 3 (2017-2018);ELSA-COVID (2020) and wave 4 (2022-2022) are currently being collected.
On a previously scheduled day, the participants appeared at 7:00 am in each research center for clinical, biochemical and questionnaire exams. In this way, a previously validated, comprehensive set of questionnaires, clinical measurements and laboratory tests wascarried out. As this wasa multicenter study, data collection was standardized.
Data werecollected in 2 phases. The first, lasting approximately1 h, included obtaining informed consent and conducting the initial interview at the participant's job site. The second, comprising additional interviews and examinations, lasted for approximately 6 h and wasconducted at a study clinic.
In this study, wave 1 was analyzed, formed by 6132 participants after we applied the following exclusion criteria: bariatric surgery, cardiovascular disease, implausible kcal consumption, implausible alcohol consumption, abstemious (0 mL/week), low HDL-C and missing data. (Figure 1).
The ELSA-Brasil sample is made up of men and women aged 35-74 years who are eligible for the study; in terms of race/color,52% are white, 28% are brown or mixed color, 16% are black, 3% are Asian (mainly Japanese) and 1% are indigenous.Exclusion criteria for ELSA-Brasilarecurrent or recent (<4 months prior to the first interview) pregnancy, the intention to quit working at the institution in the near future, severe cognitive or communication impairment and, if retired, residence outside of a study center's corresponding metropolitan area. ELSA-Brasil already has data from 3 waves: wave 1 (2008-2010), wave 2 (2012-2014) and wave 3 (2017-2018);ELSA-COVID (2020) and wave 4 (2022-2022) are currently being collected.
On a previously scheduled day, the participants appeared at 7:00 am in each research center for clinical, biochemical and questionnaire exams. In this way, a previously validated, comprehensive set of questionnaires, clinical measurements and laboratory tests wascarried out. As this wasa multicenter study, data collection was standardized.
Data werecollected in 2 phases. The first, lasting approximately1 h, included obtaining informed consent and conducting the initial interview at the participant's job site. The second, comprising additional interviews and examinations, lasted for approximately 6 h and wasconducted at a study clinic.
In this study, wave 1 was analyzed, formed by 6132 participants after we applied the following exclusion criteria: bariatric surgery, cardiovascular disease, implausible kcal consumption, implausible alcohol consumption, abstemious (0 mL/week), low HDL-C and missing data. (Figure 1).

Ethical Aspects
This study followed the international ethical standards found in the Declaration of Helsinki (2000). All procedures involving human subjects were approved by the Research and Ethics Committee of each country as follows:669/06 (USP), 343/06 (FIOCRUZ), 041/06 (UFES), 186/06 (UFMG), 194/06 (UFRGS) and 027/06 (UFBA). All participants signed a written informed consent form in both stages, with the anonymity of the information obtained being assured.

Ethical Aspects
This study followed the international ethical standards found in the Declaration of Helsinki (2000). All procedures involving human subjects were approved by the Research and Ethics Committee of each country as follows:669/06 (USP), 343/06 (FIOCRUZ), 041/06 (UFES), 186/06 (UFMG), 194/06 (UFRGS) and 027/06 (UFBA). All participants signed a written informed consent form in both stages, with the anonymity of the information obtained being assured.

Study Variables
All covariates included in the analysis were obtained through face-to-face interviews by clinical or laboratory procedures, and the variables measured in this study were sociodemographic, lifestyle, anthropometric, diet, consumption of alcohol and biochemical (serum lipoproteins). The following socio-demographic variables were evaluated by closed questionnaires withvariables such asage (years), sex (male or female), income (in tertile), education level (incomplete primary, primary, high school or university), marital status (married, separated/divorced, single, widower or other-with previous union), ethnicity (not white or white). Lifestyle variables were evaluated using closed questionnaires or specific measures of lipid-lowering drugs (yes no no);body mass index (BMI) was used to classify nutritional status (thin <18.5 kg/m 2 ; normal ≥18.5 kg/m 2 and <24.9 kg/m 2 , overweight, obese ≥25 kg/m 2 ) [25], and physical activity in leisure (low, moderate or high), smoking (never smoked, ex-smoker or smoker), consumption of alcoholic beverages (moderate or excessive), anthropometric measurements (weight (kg), height (mts) and waist circumference (cm)) were collected in a standardized way [30].Diet variables, (kilocalories, lipids, carbohydrates and proteins) were collected using the FFQ validated for the Brazilian population [31,32], as well as alcohol consumption (mL/week), beer (mL/week), spirits (ml/week), total alcohol (ml/week) and total alcohol (g/week)using the closed question questionnaire, and serum lipoproteins TC (mg/dL), triglycerides(mg/dL), HDL-C(mg/dL) and LDL-C (mg/dL)).

Alcohol Consumption
Alcohol consumption was reported through structured questionnaires with closed questions, which were used in each ELSA-Brasil research center to determine the types of alcoholic beverages (beer, wine and spirits), and the frequency and amount of consumption (daily, weekly or monthly) [33].
For this study, we worked with the total consumption of alcoholic beverages derived from the sum of millimeters/week of each participant and transformed it to g/week, classifying alcoholic beverage consumers as moderate and excessive drinkers according to the following equation: Volume milliliter week alcohol × alcohol content/100 = Volume Volume × 0.8 = grams/week Heavy drinkers were categorized by sex: men >210 g/week and women >140 g/week; and moderate drinkers: men ≤209 g/week and women ≤139 g/week [34].

Blood Analysis
Blood samples were obtained by venipuncture using scalp and vacuum collection tubes. Fasting samples were collected, and at the time of collection, the participant was informed about the procedure, and we verified the fulfillment of the given guidelines through a questionnaire. The temperature of the collection room was maintained between 20 and 24 • C, and the samples were properly stored and transported to the project's Central Laboratory, located at the University Hospital of São Paulo [35].
In this study, the biochemical parameters were obtained in two stages: after fasting for 8 to 12 h and 2 h after ingesting an overload of glucose. The samples were quickly processed to obtain serum, which was stored locally at −80 • C until it was sent to the ELSA-Brasil Central Laboratory (São Paulo) for the monthly determination of the analytes. The blood collection, processing, and biological transport protocol were standardized as described by Fedeli et al. [35]; the variables analyzed in this study were total cholesterol(enzymatic, colorimetric cholesterol oxidase method-ADVIA 1200 Siemens ® ), LDL-C (homogeneous colorimetric enzymatic method without precipitation), HDL-C (homogeneous colorimetric method without precipitation) and TGs (glycerolphosphate peroxidase method according to Trinder (enzymatic and colorimetric)). The tubes used for analysis, storage and transport were BD Vacutainer tubes with 9 mL volumes and BD Vacutainer scalps measuring 21 G and 23 G. For storage, a 2 mL Greiner Cryogenic Tube was used. The extremely high HDL-C cut-off point was definedby means of the 90th percentile, being, for this population, defined asHDL-C(≥83 mg/dL).Then, it was dichotomized into normal (40 mg/dL-82.9 mg/dL) and extremely high (≥83 mg/dL).

Statistical Analysis
The categorical (sociodemographic, lifestyle, anthropometric, consumption of alcohol and biochemical) variables dependent on HDL-C were analyzed using the chi-square test. The continuous (anthropometric, diet and biochemical) variables were expressed as mean and standard deviation and evaluated by the t-test. Likewise, they were dichotomized by HDL-C levels (normal and extremely high-HALP) and beverage consumption (moderate and excessive), performing the same statistical tests, depending on the nature of the variables.
Crude and adjusted binary logistic regression models were createdto estimate odds ratios (OR) and 95% confidence intervals (95% CI) to assess the association between HDL-C levels (dependent variable) and alcohol consumption (independent variable). The models used the dichotomized variables of HDL-C (normal and extremely high) and alcohol consumption (moderate and excessive). Model 1 was adjusted for sex, age and income, and model 2 was adjusted for model 1+ physical activity, kilocalories and BMI. All analyses were performed using Stata Statistical Software (release 13, StataCorp LP, College Station, TX, USA), and the level of significance was 5%.

Results
The study sample had 6132 participants, who were predominantly male (55.6%) and hada 6.8% prevalence of HALP. We analyzed socio-demographic, lifestyle, anthropometric, consumption and biochemical variables using the HDL-C levels. It was evident that most participants with HALP were female (p < 0.001), in the third income tertile (p < 0.001), completed higher education (p < 0.001), were married (p < 0.001), did not consume lipidlowering drugs (p < 0.001), had a normal nutritional status (p < 0.001), had low leisure-time physical activity (p < 0.001) and had a moderate consumption of alcoholic beverages (p < 0.001) Table 1. The highest means for HALP were for age (p < 0.001), total cholesterol (p < 0.001) and HDL-C (p < 0.001) Table 2.  Table 3 shows the socio-demographic, lifestyle, anthropometric, consumption and biochemical variables dependent on HDL-C levels and alcohol consumption. When analyzing the consumption of alcoholic beverages, we found participants categorized with HALP in the two categories of consumption (moderate and excessive) were mostlywomen (p < 0.001), had a university degree (p = 0.005) and had a normal body mass index (p = 0.050). Higher proportions were observed in moderate consumption for those who had never smoked and in excessive consumption for ex-smokers (p < 0.001). When analyzing the anthropometric, consumption, and biochemical variables depending on HDL-C levels and alcohol consumption, we found higher means in excessive consumption for weight (p = 0.023), height (p = 0.006) waist circumference (p = 0.011), kilocalorie consumption (p = 0.040) and all the categories of alcoholic beverages (p < 0.001) and for the consumption of wine (p = 0.007) ( Table 4).  Table 5 presents the binary logistic regression with progressive adjustments (sex, age, income, physical activity, kilocalories and BMI). We found a positive association with extremely high levels of HDL-C (HALP) and excessive alcohol consumption (OR = 1.92; 95% CI (1.4-2.5)). Table 5. Binary logistic regression (OR (95% CI)) between alcoholic beverage consumption and extremely high levels of HDL-C (HALP).

Discussion
The objective was to evaluate the excessive consumption of alcoholic beverages and the association with extremely high levels of HDL-C (HALP) in a Brazilian population; a positive relationship was found.
Participants with excessive drinking are more likely to have extremely high levels of HDL-C (HALP). These participants are mostly women with a high income, lower waist circumference, lower kilocalorie consumption and higher consumption in all categories of alcoholic beverages.
We found a prevalence of 11.8% of excessive alcohol consumption, which agrees with a population-based longitudinal study [29] and with the Brazilian population [36]. In our study, a prevalence of 6.8% of HALP was also estimated, which is lower in North Americans (3.8%) [37] and Asians (1.9%) [11], because the cut-off points for defining HALP in these populations were higher than in our study.
HDL-C levels differ when compared by sex, which is evident in our study, as the largest proportion of participants with HALP were women. The behavior of this lipoprotein and the corresponding changes depend on the physiological and hormonal statuses of women, showing an inverse association between the serum levels of HDL-C and estrogens [38]. Menopause is, therefore, a key factor in this relationship, as postmenopausal women have higher HDL-C values when compared to premenopausal women [39]. The physiological stage present in our population, determined by the average age of our sample, is consistent with the onset of these hormonal and physiological changes.
There is a direct relationship between education and cognition, relating to behaviors in favor of health care [40]. Consistent evidence shows this association in women [41][42][43], and the highest proportions of HALP were found in participants with high incomes and education levels, which is a result also evident in Chilean women [44]. It is important to recognize the nature of our sample, as they are employed or retired from federal universities, characteristics that impact sociodemographic variables in relation to the general population, especially those with higher education levels and incomes.
Our findings show a lower value of triglycerides in participants with HALP, agreeing with a study that evaluated sixNorth American cohorts, explaining this physiological behavior by abnormalities in lipid metabolic pathways, especially in the exchange of HDL-C to VLDL particles [45].
Another important factor is nutritional status: an imbalance in this variable is consideredas a risk factor for the global burden of disease [46]. In turn, obesity increases the risk of cardiovascular diseases, generating atheroma and changes in lipoproteins (dyslipidemias) [47]. The inverse association between nutritional status and serum HDL-C levels is well known. Increased BMI and WC present greater chances of low HLD-C, explaining this relationship in 57.1% and 36%of cases, respectively [48].
The guidelines which aim to reduce weight as a strategy to increase HDL-C [49,50] were found to be relevant in our study, as participants with HALP have lower levels of anthropometric variables. It is well known that the effect of weight loss by dietary modification, physical exercise or bariatric surgery influence changes in lipoproteins [51].
Different health outcomes are associated with the consumption of alcoholic beverages, depending on the quantity and type of these beverages, mainly related to cardiovascular effects [52], diabetes [53] and cancer [54], leading to mortality from all causes [55]. The findings relating the consumption of alcoholic beverages and HDL-C are emphatic in the relationship between these two variables, in dose response [27] or also in the form of "u"or "J"-shaped responses, attributing the protective role to HDL-C. These responses may result from activity on the efflux of cholesterol, and anti-inflammatory, antioxidant, antithrombotic and antiapoptotic characteristics [5][6][7], leading to a false hypothesis of cardiovascular protection with the association of alcoholic beverages and HDL-C levels [12]. However, this increase in HDL-C levels, dependent on the consumption of alcoholic beverages, must be carefully evaluated. Currently, there are some counterpoints to be analyzed when establishing these relationships. In a large population-based study in Japan, a genetic mutation was found in the cholesterol ester transport protein enzyme (CETP), responsible for the metabolism and levels of HDL-C. These findings report an association ofthe incidence of ischemic changes, being statistically significant and U-shaped, with very low and extremely high levels (HALP) of HDL-C [11].
Increased HDL-C levelshavebeen taken as evidence that therapeutic recommendations need to be made for the benefit of cardiovascular health;it is known that CETPis a mediator in the exchange of cholesterol ester and reverse cholesterol. For this purpose, specialized drugs have been developedfor the inhibition of CETP, demonstrating that when this protein is inhibited, HDL-C levels increase considerably, reaching extremely high levels (HALP) with a significant increase in mortality and cardiovascular events [56,57]; it has been failedto hypothesize that just increasing HDL-C at very high levels could be beneficial to cardiovascular health. This level of HDL-C may favor atherogenesis, because the molecules that predominate are HDL-2, rich in cholesterol esters and less able to carry out reverse cholesterol transport [58].
The antioxidant potential attributed to HDL-C in plasma has also been analyzed. There are other components with greater antioxidant potential than isolated HDL-C, such as fibrinogen, immunoglobulin G, uric acid albumin and ascorbate, with albumin being the dominant antioxidant and HDL-C with lower power 1-2% total antioxidant contribution in plasma [59], leading researchersto reflect on the antioxidant role of this lipoprotein.
The findings suggest that the relationship between HDL-C levels and cardiovascular health is still complex, going beyond just plasma concentrations [60]. HDL-C is a heterogeneous lipoprotein with different components, so the association with CVD should be further analyzed. HDL-C has been found to contain alpha-2 macroglobulin, CoC3 (complement C3), HP (haptoglobin) or PLMG (plasminogen), leading to the idea that not only will the increase in HDL-C contribute to cardiovascular protection, but also, the balance between the intrinsic components of these lipoproteins [61] and molecules that interact in HDL-C metabolism [23] should bothbeevaluated.
The effects of alcohol on health are still controversial, mainly due to itscharacteristics, referring to type, consumption pattern and quantity. It has already been described that excessive consumption has a negative impact on the health of the population and has implications for public health [1,62].
The possible mechanism by which the consumption of alcoholic beverages brings benefits to cardiovascular health is attributed to the increase in HDL-c levels, with a false belief that extremely high levels could bring even greater benefits [29].
In a recent meta-analysis, it wasshown that interest in studying the effects of alcohol consumption has increased considerably in recent years, in addition to reports of an increase in HDL-C levels (3.66 mg/dL 2.22-5.13) per 30 g of alcohol/day [63], associated with an increase in the rate of lipoprotein transport [64]. However, the results are still controversial, as the mechanisms by which alcohol influences HDL-C are not fully understood.
Our results should be considered in terms of clinical implications, as the increase in the dose-response of alcohol consumption and HDL-C levels should be analyzed carefully, as it would lead to the idea that excessive consumption would further increase HDL-C levels. C. It is known that extremely high levels of HDL-C are associated with dysfunctions in lipid metabolism or genetic load, reflected in non-functional HDL-C and correlated with greater risks in cardiovascular health.
Therefore, when an extremely high level of HDL-C is observed, it must be classified as a risk indicator, and health treatment must be given.As we currently think that HDL-C has no risk point, the recommendations only suggest increasing it without a cutoff point; this has clinical implications because the determination of HDL-C levels is a routine and easy test that brings important information that is sometimes undervalued.
The strengths and weaknesses of this study werethat ELSA-Brasil is not a representative study of the Brazilian population, but it represents an important portion of the population, providing important data for South American countries. The alcoholic beverage instrument has memory bias, but it was applied by trained personnel and had data quality control. Participants with implausible consumption were excluded.Thisis one of the first findings in a South American population, presentingimportant information about this population that has specific characteristics.
Finally, the consumption of alcoholic beverages in the world is increasing, and the mechanisms of association with cardiovascular health should be better elucidated. Few studies have tried to understand the specific effect of HDL-C, and this research is a pioneer in this association.

Conclusions
Excessive alcohol consumption was positively associated with extremely high HDL-C (HALP). The increase in serum levels caused by the consumption of alcoholic beverages must be carefully analyzed, because HDL-C is a heterogeneous lipoprotein, and high levels do not provide cardiovascular protection.