This cohort study investigates the association between very high high-density lipoprotein cholesterol (HDL-C) level and outcomes in patients with coronary artery disease.
Key Points
Question
What is the association between very high high-density lipoprotein cholesterol (HDL-C) level and outcomes in patients with coronary artery disease (CAD)?
Findings
In this cohort study of 14 478 participants with CAD enrolled in the UK Biobank (UKB) and 5467 participants in the Emory Cardiovascular Biobank (EmCAB), individuals in the UKB with HDL-C levels greater than 80 mg/dL had a 96% higher risk of all-cause mortality and a 71% higher risk of cardiovascular mortality after adjustment for covariates, compared with those with HDL-C levels in the range of 40 to 60 mg/dL. These findings were replicated in patients with CAD in the EmCAB.
Meaning
Very high HDL-C levels are paradoxically associated with higher mortality risk in individuals with CAD.
Abstract
Importance
Previous studies have shown lower cardiovascular risk with higher high-density lipoprotein cholesterol (HDL-C) levels. However, recent data in the general population have shown increased risk of adverse outcomes at very high HDL-C concentrations.
Objective
To study the association between very high HDL-C levels (>80 mg/dL) and mortality in patients with coronary artery disease (CAD) and to investigate the association of known HDL-C genotypes with high HDL-C level outcomes.
Design, Setting, and Participants
This prospective, multicenter, cohort study, conducted from 2006 to present in the UK and from 2003 to present in Atlanta, Georgia, recruited patients with CAD from the UK Biobank (UKB) and the Emory Cardiovascular Biobank (EmCAB), respectively. Patients without confirmed CAD were excluded from the study. Data analyses were conducted from May 10, 2020, to April 28, 2021.
Exposure
High HDL-C levels (>80 mg/dL).
Main Outcomes and Measures
The primary outcome was all-cause death. The secondary outcome was cardiovascular death.
Results
A total of 14 478 participants (mean [SD] age, 62.1 [5.8] years; 11 034 men [76.2%]) from the UKB and 5467 participants (mean [SD] age, 63.8 [12.3] years; 3632 men [66.4%]) from the EmCAB were included in the study. Over a median follow-up of 8.9 (IQR, 8.0-9.7) years in the UKB and 6.7 (IQR, 4.0-10.8) years in the EmCAB, a U-shaped association with outcomes was observed with higher risk in those with both low and very high HDL-C levels compared with those with midrange values. Very high HDL-C levels (>80 mg/dL) were associated with increased risk of all-cause death (hazard ratio [HR], 1.96; 95% CI, 1.42-2.71; P < .001) and cardiovascular death (HR, 1.71; 95% CI, 1.09-2.68; P = .02) compared with those with HDL-C levels in the range of 40 to 60 mg/dL in the UKB after adjustment for confounding factors. These results were replicated in the EmCAB. These associations persisted after adjustment for the HDL-C genetic risk score within the UKB. Sensitivity analyses demonstrated that the risk of all-cause mortality in the very high HDL-C group was higher among men than women in the UKB (HR, 2.63; 95% CI, 1.75-3.95; P < .001 vs HR, 1.39; 95% CI, 0.82-2.35; P = .23).
Conclusions and Relevance
Results of this cohort study suggest that very high HDL-C levels are paradoxically associated with higher mortality risk in individuals with CAD. This association was independent of the common polymorphisms associated with high HDL-C levels.
Introduction
High-density lipoprotein cholesterol (HDL-C) levels have historically been inversely associated with increased cardiovascular disease (CVD) risk.1 However, recent studies have questioned the efficacy of therapies designed to increase HDL-C levels. For example, cholesteryl ester transfer protein inhibition that increases HDL-C levels does not reduce CVD risk.2,3,4,5,6,7 Additionally, genetic variants associated with HDL-C are not linked to CVD risk.8 Recent epidemiologic studies in populations free of CVD from northern Europe and Canada have linked very high levels of HDL-C with higher mortality risk.9,10 Whether very high HDL-C levels in patients with coronary artery disease (CAD) are associated with mortality risk remains unknown. We investigated the association between very high HDL-C levels and all-cause and cardiovascular mortality risk in 2 independent groups of patients with CAD in the UK Biobank (UKB) and the Emory Cardiovascular Biobank (EmCAB) and the association of known HDL-C genetic variants with this risk. Our hypothesis was that very high HDL-C would be associated with greater mortality risk in individuals with CAD, and there may be a genetic association with this risk.
Methods
This study was approved by the institutional review board of Emory University in Atlanta, Georgia, and the North West Multicentre Research ethics committee in the UK. All participants from both databases provided written informed consent. We analyzed data from a subset of patients in the UKB with CAD at enrollment and patients with CAD in the EmCAB. The UKB and EmCAB participants were followed prospectively for a median of 8.9 (IQR, 8.0-9.7) years and 6.7 (IQR, 4.0-10.8) years, respectively. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
UK Biobank
The design of the UKB has been reported previously.11 Participants aged 40 to 72 years with a history of CAD at enrollment were included from the UK general population between 2006 and 2010. The inclusion criteria have been described previously.12 The UKB data were linked to Hospital Episode Statistics data, which cover all hospital admissions until 2016, dating back to 1997 for England, 1998 for Wales, and 1981 for Scotland. Hospital Episode Statistics data use the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) to record diagnosis information and Office of Population, Censuses, and Surveys: Classification of Interventions and Procedures, version 4 (OPCS-4), to code operative procedures. Death registries used include all deaths until 2018, with both primary and secondary causes of death coded in ICD-10.
We identified individuals with CAD history at enrollment, using in-patient Hospital Episode Statistics data based on the earliest record of ICD-10 codes I20 to I25 and OPCS-4 codes K40-K46, K49, K50, or K75. Patients with a CAD onset date before the enrollment date were included. Follow-up time was defined as the time from enrollment until incident all-cause or cardiovascular death, loss to follow-up, or end of follow-up. Information about date and primary and secondary causes of incident death were obtained from the death registry. We defined incident cardiovascular death using ICD-10 codes of deaths from diseases of the heart (I00-I09, I11, I13, I20-I51), essential hypertension and hypertensive renal disease (I10, I12, I15), cerebrovascular diseases (I60-I69), and peripheral vascular disease (I73).13
HDL-C level, demographics, and relevant risk factors were obtained at enrollment, including age, sex, race and ethnicity, body mass index (BMI), hypertension, smoking, total cholesterol, triglycerides, low-density lipoprotein cholesterol (LDL-C), estimated glomerular filtration rate (eGFR), and self-reported stroke history, heart attack history, diabetes, and frequent alcohol use (consumption ≥3 times per week considered frequent). HDL-C, LDL-C, triglycerides, and total cholesterol were assayed using the AU5800 analytical platform (Beckman Coulter). Genotyping and imputation of the UKB genomic data were described previously.14 Self-reported race and ethnicity were obtained from questionnaires and comprised Asian, Black, White, and mixed background. Inclusion of race and ethnicity information was for the consideration of potential differences or confounding, and it was not a requirement of the funding bodies.
EmCAB
The EmCAB is an ongoing prospective cohort of adults 18 years or older undergoing left heart catheterization for suspected or confirmed CAD at 3 Emory Healthcare sites in Atlanta, Georgia.15 Patients without confirmed CAD were excluded. Individuals underwent a baseline evaluation using standardized questionnaires and medical records review. Age, sex (male or female), race and ethnicity, smoking history, and frequent alcohol use frequency (consumption of ≥8 alcohol beverages per week considered frequent) were obtained by self-report. Self-reported race and ethnicity were obtained from questionnaires and comprised Asian, Black, Hispanic, White, and other background (other race and ethnicity included Native American and Pacific Islander). Inclusion of race and ethnicity information was for the consideration of potential differences or confounding, and it was not a requirement of the funding bodies. Medical history and medication use were obtained by self-report and medical records review for hypertension, diabetes, hyperlipidemia, heart failure, and myocardial infarction (MI) history. Blood pressure, weight, and height were measured. BMI for both cohorts was calculated as weight in kilograms divided by height in meters squared. Serum creatinine at enrollment before cardiac catheterization was obtained from routine clinic visits or hospitalizations within the Emory Healthcare system. The eGFR (milliliter per minute per 1.73 m2) was computed using the Chronic Kidney Disease Epidemiology Collaboration equation.16 Medications were obtained by self-report and medical records review.
Fasting total cholesterol, HDL-C, and triglyceride levels were assayed (Beckman Coulter). LDL-C level was calculated using the Friedewald equation. Fasting arterial blood samples for serum were drawn before cardiac catheterization and stored at −80 °C. Serum concentrations of high-sensitivity C-reactive protein (hsCRP) were determined using the Abbott Architect platform (Abbott Laboratories), which has a lower limit of detection of 0.03 mg/L.17
Follow-up data were collected by blinded personnel through telephone interview, medical record review, Social Security Death Index, and state records. Two independent blinded cardiologists adjudicated the cause of death. Medical records were accessed to validate self-reported events. Cardiovascular death was defined as death from myocardial infarction, heart failure, stroke, pulmonary embolism, or as a complication during any cardiovascular-related procedure.
Statistical Analysis
Analysis of variance, Mann-Whitney U test, and χ2 tests were used where appropriate to compare baseline characteristics between 5 categories of HDL-C: 30 mg/dL or less, 30 to 40 mg/dL, 40 to 60 mg/dL, 60 to 80 mg/dL, and greater than 80 mg/dL. Continuous variables were reported as mean (SD) or median (IQR); categorical variables were reported as frequency (proportion).
Kaplan-Meier curves were generated for all-cause and cardiovascular death across HDL-C categories. Cox proportional hazards models were used to compute hazard ratios (HRs) and 95% CIs for all-cause death, and Fine and Gray subdistribution hazard models18 were used for cardiovascular death, treating noncardiovascular death as a competing risk. Individuals with HDL-C levels between 40 and 60 mg/dL formed the reference. Both unadjusted and adjusted HRs were computed. For UKB, models adjusted for age, sex, race and ethnicity, BMI, history of hypertension, diabetes, smoking, triglyceride level, LDL-C level, stroke history, heart attack history, eGFR, frequent alcohol use. For EmCAB, models adjusted for age, sex, race and ethnicity, BMI, history of hypertension, diabetes, current/former smoking, triglyceride level, LDL-C level, heart failure history, myocardial infarction history, eGFR, frequent alcohol use, statin use, aspirin use, β-blocker use, and angiotensin-converting enzyme inhibitor/angiotensin receptor blocker use. Proportional hazards assumption was met, which was assessed by adding a time-dependent interaction term between HDL-C categories and follow-up time.
Using R package smoothHR,19 the associations between HDL-C and both cardiovascular and all-cause death were examined using HR curves that permitted nonlinear associations, treating an HDL-C level of 55 mg/dL as the reference with adjustment of the aforementioned covariates. The 95% CI was computed along a continuous spectrum of HDL-C levels. We performed interaction analyses to determine if the association between very high HDL-C and outcomes differed by sex (male/female), diabetes (yes/no), frequent alcohol use (yes/no), and age (<65 years/≥65 years) adjusting for aforementioned covariates.
For UKB, we created a genetic risk score (GRS) based on the 142 independent single nucleotide variations associated with HDL-C at the genome-wide significance threshold (P < 5 × 10−8) in a large-scale genome-wide association study.20 The GRS was calculated by summing all single nucleotide variations weighted by reported β coefficients, then converted to z scores and included in the fully adjusted models, adjusting population structure by including the top 10 principal components.
Statistical analyses were performed from May 10, 2020, to April 28, 2021, using SAS, version 9.4 (SAS Institute) and R, version 4.0.2 (R Project). Statistical significance was defined as a 2-sided P value < .05.
Results
UKB
Of the 14 478 UKB participants with CAD (mean [SD] age, 62.1 [5.8] years; 11 034 men [76.2%]; 3444 women [23.8%]; 539 Asian [3.7%], 119 Black [0.8%] 13 586 White [93.8%], 234 mixed [1.6%] race and ethnicity), there were 1795 all-cause deaths (12.4%) and 1137 cardiovascular deaths (7.9%) during follow-up. Individuals with an HDL-C level greater than 80 mg/dL constituted 1.8% (255 of 14 478) of this cohort and were likely to be older (mean [SD] age, 62.8 [5.2]), female (150 [58.8%]), with greater alcohol consumption (frequent alcohol use, 157 [61.6%]) and a lower prevalence of cardiovascular risk factors (hypertension, 139 [54.5%]; diabetes, 28 [11.0%]) and heart attack history (92 [36.1%]) compared with those with lower HDL-C levels. Individuals with an HDL-C level greater than 80 mg/dL also had lower mean (SD) triglyceride levels (93.0 [40.7] mg/dL; to convert to millimoles per liter, multiply by 0.0113) and higher mean (SD) LDL-C levels (110.6 [36.2] mg/dL; to convert to millimoles per liter, multiply by 0.0259) (Table 1) than those with lower HDL-C levels. The distribution of HDL-C across the cohort is shown in eFigure 1 in the Supplement. Within the group with HDL-C levels of 80 mg/dL or more, all-cause death was reported in 43 of 255 participants (16.9%), and cardiovascular deaths were reported 22 of 255 participants (8.6%). The proportions of outcomes by HDL-C are shown in eFigure 2 in the Supplement.
Table 1. Baseline Characteristics: UK Biobank Coronary Artery Disease Cohort by HDL-C Categories.
| Variable | HDL-C level | ||||
|---|---|---|---|---|---|
| ≤30 mg/dL | 30-40 mg/dL | 40-60 mg/dL | 60-80 mg/dL | >80 mg/dL | |
| Total No. (%) | 641 (4.4) | 4103 (28.3) | 7888 (54.5) | 1591 (11.0) | 255 (1.8) |
| Age, mean (SD), y | 61.1 (6.4) | 61.6 (6.1) | 62.2 (5.7) | 62.8 (5.5) | 62.8 (5.2) |
| Women, No. (%) | 39 (6.1) | 441 (10.8) | 2006 (25.4) | 808 (50.8) | 150 (58.8) |
| Men, No. (%) | 602 (93.9) | 3662 (89.3) | 5882 (74.6) | 783 (49.2) | 105 (41.2) |
| Race and ethnicity, No. (%) | |||||
| Asian | 35 (5.5) | 193 (4.7) | 271 (3.4) | 33 (2.1) | 7 (2.8) |
| Black | 3 (0.5) | 28 (0.7) | 70 (0.9) | 12 (0.8) | 6 (2.4) |
| White | 582 (90.8) | 3815 (93.0) | 7425 (94.1) | 1525 (95.9) | 239 (93.7) |
| Mixed | 21 (3.3) | 67 (1.6) | 122 (1.6) | 21 (1.3) | 3 (1.2) |
| Hypertension, No. (%) | 403 (63.4) | 2337 (57.2) | 4254 (54.1) | 825 (52.1) | 139 (54.5) |
| Diabetes, No. (%) | 259 (41.1) | 1117 (27.4) | 1282 (16.4) | 133 (8.4) | 28 (11.0) |
| Heart attack history, No. (%) | 389 (61.2) | 2327 (56.9) | 3956 (50.3) | 675 (42.6) | 92 (36.1) |
| Stroke history, No. (%) | 57 (9.0) | 295 (7.2) | 507 (6.5) | 100 (6.3) | 25 (9.8) |
| eGFR, mean (SD) mL/min/1.73 m2 | 81.1 (19.2) | 82.5 (16.5) | 83.4 (15.3) | 83.7 (14.5) | 85.4 (15.6) |
| Frequent alcohol use, No. (%)a | 120 (18.8) | 1205 (29.4) | 3428 (43.6) | 861 (54.3) | 157 (61.6) |
| Current/former smoker, No. (%) | 459 (72.6) | 2717 (66.9) | 4961 (63.3) | 956 (61.0) | 168 (66.1) |
| Body mass index,b mean (SD) | 31.2 (5.3) | 30.6 (4.9) | 29.1 (4.7) | 27.4 (4.7) | 25.9 (4.7) |
| Total cholesterol, mean (SD), mg/dL | 138.1 (32.5) | 158.4 (33.1) | 176.8 (36.5) | 197.1 (37.5) | 215.4 (43.7) |
| LDL-C, mean (SD), mg/dL | 85.8 (23.8) | 98.3 (24.9) | 106.7 (28.2) | 111.4 (30.5) | 110.6 (36.2) |
| Triglycerides, mean (SD), mg/dL | 243.7 (144.9) | 204.8 (108.9) | 159.0 (84.0) | 115.7 (56.0) | 93.0 (40.7) |
| hsCRP, mg/L, median (IQR) | 2.3 (1.2 to 4.7) | 1.8 (0.9 to 3.6) | 1.4 (0.7 to 3.0) | 1.3 (0.6 to 2.7) | 1.3 (0.6 to 3.1) |
| HDL-C GRS, mean (SD), z score | –0.6 (0.9) | –0.3 (0.9) | 0.0 (1.0) | 0.2 (1.0) | 0.6 (1.0) |
| All-cause death, No. (%) | 123 (19.2) | 552 (13.5) | 888 (11.3) | 189 (11.9) | 43 (16.9) |
| Cardiovascular death, No. (%) | 89 (13.9) | 360 (8.8) | 563 (7.1) | 103 (6.5) | 22 (8.6) |
Abbreviations: eGFR, estimated glomerular filtration rate; GRS, genetic risk score; HDL-C, high-density lipoprotein cholesterol; hsCRP, high-sensitivity C-reactive protein; LDL-C, low-density lipoprotein cholesterol.
SI conversion factors: To convert total cholesterol to millimoles per liter, multiply by 0.0259; LDL-C to millimoles per liter, multiply by 0.0259; triglycerides to millimoles per liter, multiply by 0.0113; CRP to milligrams per deciliter, divide by 10; HDL-C to millimoles per liter, multiply by 0.0259.
Frequent alcohol use defined as alcohol consumption 3 or more times per week.
Calculated as weight in kilograms divided by height in meters squared.
Kaplan-Meier curves by HDL-C categories are shown in eFigure 3 in the Supplement. In unadjusted models, compared with the reference category with HDL-C levels of 40 to 60 mg/dL, those with low HDL-C levels (≤30 mg/dL) had an expected higher risk of all-cause and cardiovascular mortality, even after adjustment for the aforementioned covariates (all-cause mortality: HR, 1.33; 95% CI, 1.07-1.64; P = .009; cardiovascular mortality: HR, 1.42; 95% CI, 1.09-1.85; P = .009). Importantly, compared with the reference category, individuals with very high HDL-C levels (>80 mg/dL) also had a higher risk of all-cause death (HR, 1.58; 95% CI, 1.16-2.14; P = .004), but cardiovascular death rates were not significantly greater in unadjusted analyses (HR, 1.23; 95% CI, 0.80-1.87; P = .34). However, after adjustment, the highest HDL-C group also had an increased risk of both all-cause death (HR, 1.96; 95% CI, 1.42-2.71; P < .001) and cardiovascular death (HR, 1.71; 95% CI, 1.09-2.68; P = .02). Further, individuals with HDL-C levels in the range of 60 to 80 mg/dL also had an adjusted increased risk for all-cause death (HR, 1.27; 95% CI, 1.08-1.50; P = .005). Using adjusted nonlinear HR curves, a U-shaped association between HDL-C and outcomes was evident (Figure 1).
Figure 1. Nonlinear Association Between High-Density Lipoprotein (HDL) Cholesterol Levels and Adverse Outcomes .
UK Biobank (UKB) coronary artery disease cohort model of all-cause death (A) and cardiovascular death (C) adjusted for age, sex, race and ethnicity, body mass index, hypertension, diabetes, smoking, triglycerides, low-density lipoprotein (LDL) cholesterol, stroke history, heart attack history, estimated glomerular filtration rate (eGFR), and frequent alcohol use (defined as alcohol consumption ≥3 times per week). Emory Cardiovascular Biobank (EmCAB) model of all-cause death (B) and cardiovascular death (D) adjusted for age, sex, race and ethnicity, body mass index, hypertension, diabetes, current/former smoking, triglycerides, LDL cholesterol, heart failure history, myocardial infarction history, eGFR, frequent alcohol use (defined as ≥8 alcohol beverages per week), statin use, aspirin use, β-blocker use, and angiotensin-converting enzyme inhibitor/angiotensin receptor blocker use. HR indicates hazard ratio.
Significant or near-significant interactions with sex were observed in the group with HDL-C level greater than 80 mg/dL for all-cause death (β = 0.68; SE = 0.33; P = .04) and cardiovascular death (β = 0.94; SE = 0.49; P = .06). Compared with women, men with HDL-C levels greater than 80 mg/dL had higher all-cause death risk (HR, 2.63; 95% CI, 1.75-3.95; P < .001 vs HR, 1.39; 95% CI, 0.82-2.35; P = .23) and cardiovascular death risk (HR, 2.50; 95% CI, 1.47-4.27; P < .001 vs HR, 0.89; 95% CI, 0.39-2.07; P = .79) (Figure 2 and eTable 1 in the Supplement).
Figure 2. Forest Plot of High-Density Lipoprotein (HDL) Cholesterol (>80 mg/dL) Compared With Reference HDL Cholesterol (40-60 mg/dL).
UK Biobank (UKB) coronary artery disease cohort model of all-cause death (A) and cardiovascular death (C) adjusted for age, sex, race and ethnicity, body mass index, hypertension, diabetes, smoking, triglycerides, low-density lipoprotein (LDL) cholesterol, stroke history, heart attack history, estimated glomerular filtration rate (eGFR), and frequent alcohol use (defined as alcohol consumption ≥3 times per week), excluding the variable of stratification. Emory Cardiovascular Biobank (EmCAB) model of all-cause death (B) and cardiovascular death (D) adjusted for age, sex, race and ethnicity, body mass index, hypertension, diabetes, current/former smoking, triglycerides, LDL cholesterol, heart failure history, myocardial infarction history, eGFR, frequent alcohol use (defined as ≥8 alcohol beverages per week), statin use, aspirin use, β-blocker use, and angiotensin-converting enzyme inhibitor/angiotensin receptor blocker use, excluding the variable of stratification. P values indicate test for interaction between HDL cholesterol greater than 80 mg/dL group and variable of stratification. HR indicates hazard ratio.
aP value < .05.
EmCAB
The baseline characteristics by HDL-C categories for the EmCAB patients are shown in Table 2. The distribution of HDL-C levels is shown in eFigure 1 in the Supplement. Of the 5467 patients (mean [SD] age, 63.8 [12.3] years; 3632 men [66.4%]; 1835 women [33.6%]; 95 Asian [1.7%], 1217 Black [22.3%], 51 Hispanic [0.9%], 4004 White [73.2%]; 100 other [1.8%] race and ethnicity) included from the EmCAB database, 90 (1.6%) had HDL-C levels greater than 80 mg/dL, and this group was more likely to be Black (34 [37.8%], female (60 [66.7%]), with greater alcohol intake (8 [9.8%]), lower mean (SD) triglyceride levels (86.4 [54.9] mg/dL), a lower prevalence of myocardial infarction history (11 [12.4%]), and statin (63 [70.0%]), aspirin (61 [67.8%]), β-blocker (53 [58.9%]), and angiotensin-converting enzyme inhibitor/angiotensin receptor blocker use (37 [41.1%]) compared with those with lower HDL-C levels. During follow-up, there were a total of 1471 all-cause deaths (26.9%) and 757 cardiovascular deaths (13.8%). Within the group with HDL-C levels greater than 80 mg/dL, all-cause and cardiovascular deaths were observed in 27 of 90 participants (30.0%) and 15 of 90 participants (16.7%), respectively, during follow-up. The proportions of outcomes by HDL-C are shown in eFigure 2 in the Supplement.
Table 2. Baseline Characteristics: Emory Cardiovascular Biobank by HDL-C Categories.
| Variable | HDL-C level | ||||
|---|---|---|---|---|---|
| ≤30 mg/dL | 30-40 mg/dL | 40-60 mg/dL | 60-80 mg/dL | >80 mg/dL | |
| Total No. (%) | 852 (15.6) | 1907 (34.9) | 2139 (39.1) | 479 (8.8) | 90 (1.6) |
| Age, mean (SD), y | 61.5 (12.4) | 63.4 (12.0) | 64.7 (12.3) | 64.8 (13.0) | 63.3 (11.9) |
| Women, No. (%) | 128 (15.0) | 466 (24.4) | 892 (41.7) | 289 (60.3) | 60 (66.7) |
| Men, No. (%) | 724 (85.0) | 1441 (75.6) | 1247 (58.3) | 190 (39.7) | 30 (33.3) |
| Race and ethnicity, No. (%) | |||||
| Asian | 9 (1.1) | 36 (1.9) | 43 (2.0) | 7 (1.5) | 0 (0) |
| Black | 163 (19.1) | 397 (20.8) | 474 (22.2) | 149 (31.1) | 34 (37.8) |
| Hispanic | 14 (1.6) | 13 (0.7) | 20 (0.9) | 4 (0.8) | 0 (0) |
| White | 648 (76.1) | 1421 (74.5) | 1569 (73.4) | 311 (64.9) | 55 (61.1) |
| Othera | 18 (2.1) | 40 (2.1) | 33 (1.5) | 8 (1.7) | 1 (1.1) |
| Hypertension, No. (%) | 687 (81.2) | 1504 (79.1) | 1704 (80.1) | 385 (80.9) | 74 (82.2) |
| Diabetes, No. (%) | 389 (46.0) | 766 (40.3) | 730 (34.4) | 135 (28.4) | 25 (28.1) |
| Heart failure history, No. (%) | 343 (40.3) | 690 (36.2) | 764 (35.7) | 175 (36.5) | 33 (36.7) |
| MI history, No. (%) | 272 (32.4) | 547 (29.1) | 492 (23.5) | 104 (22.0) | 11 (12.4) |
| Obstructive CAD, No. (%)b | 607 (84.3) | 1356 (83.9) | 1394 (79.8) | 292 (75.8) | 52 (71.2) |
| eGFR, mean (SD), mL/min/1.73 m2 | 70.5 (25.5) | 72.1 (23.1) | 70.7 (24.1) | 71.8 (25.3) | 72.9 (27.1) |
| Frequent alcohol use, No. (%)c | 30 (4.0) | 106 (6.3) | 141 (7.5) | 30 (7.2) | 8 (9.8) |
| Current/former smoker, No. (%) | 582 (68.3) | 1256 (65.9) | 1342 (62.7) | 281 (58.7) | 60 (66.7) |
| BMI, mean (SD)d | 31.0 (5.9) | 30.3 (6.0) | 29.1 (6.0) | 27.5 (5.8) | 25.8 (6.0) |
| Total cholesterol, mean (SD) mg/dL | 151.3 (48.4) | 159.6 (41.4) | 170.4 (43.9) | 183.3 (41.1) | 197.8 (39.5) |
| LDL-C, mean (SD), mg/dL | 87.1 (39.7) | 92.1 (35.2) | 97.3 (38.9) | 96.5 (38.0) | 91.5 (32.5) |
| Triglycerides, mean (SD), mg/dL | 211.1 (202.8) | 163.0 (114.0) | 127.1 (82.9) | 100.3 (72.9) | 86.4 (54.9) |
| hsCRP, mg/L, median (IQR) | 4.1 (1.5 to 11.0) | 3.1 (1.3 to 7.8) | 2.5 (1.0 to 6.4) | 2.9 (0.9 to 7.7) | 1.9 (0.8 to 4.8) |
| Statin use, No. (%) | 656 (77.0) | 1488 (78.0) | 1592 (74.4) | 323 (67.4) | 63 (70.0) |
| Aspirin use, No. (%) | 704 (82.6) | 1583 (83.0) | 1706 (79.8) | 351 (73.3) | 61 (67.8) |
| β-Blocker use, No. (%) | 636 (74.7) | 1376 (72.2) | 1482 (69.3) | 279 (58.3) | 53 (58.9) |
| ACEI/ARB use, No. (%) | 525 (61.6) | 1161 (60.9) | 1191 (55.7) | 256 (53.4) | 37 (41.1) |
| HDL-C GRS, mean (SD), z score | –0.3 (1.0) | –0.1 (0.9) | 0.1 (1.0) | 0.6 (1.2) | 0.7 (0.7) |
| All-cause death, No. (%) | 284 (33.3) | 477 (25.0) | 548 (25.6) | 135 (28.2) | 27 (30.0) |
| Cardiovascular death, No. (%) | 152 (17.8) | 258 (13.5) | 268 (12.5) | 64 (13.4) | 15 (16.7) |
Abbreviations: ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; BMI, body mass index; CAD, coronary artery disease; eGFR, estimated glomerular filtration rate; GRS, genetic risk score; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; hsCRP, high-sensitivity C-reactive protein; MI, myocardial infarction.
SI conversion factors: To convert total cholesterol to millimoles per liter, multiply by 0.0259; LDL-C to millimoles per liter, multiply by 0.0259; triglycerides to millimoles per liter, multiply by 0.0113; CRP to milligrams per deciliter, divide by 10; HDL-C to millimoles per liter, multiply by 0.0259.
Other race and ethnicity included Native American and Pacific Islander.
Obstructive CAD defined as 50% or more stenosis in 1 or more epicardial arteries.
Frequent alcohol use defined as 8 or more alcohol beverages per week.
Calculated as weight in kilograms divided by height in meters squared.
Kaplan-Meier survival curves by HDL-C categories are shown in eFigure 3 in the Supplement. As compared with individuals with HDL-C levels in the range of 40 to 60 mg/dL, those in the lowest (≤30 mg/dL) and the highest (>80 mg/dL) groups had a significant or near-significant greater risk of all-cause death in unadjusted models (HR, 1.26; 95% CI, 1.09-1.46; P = .002; HR, 1.43; 95% CI, 0.97-2.10; P = .07, respectively). Similar associations were observed with cardiovascular death (HR, 1.37; 95% CI, 1.13-1.68; P = .002; HR, 1.46; 95% CI, 0.88-2.44; P = .14, respectively). In fully adjusted models, the group with the lowest HDL-C levels had a higher risk of all-cause death (HR, 1.22; 95% CI, 1.03-1.45; P = .02) and cardiovascular death (HR, 1.35; 95% CI, 1.06-1.72; P = .01). Importantly, the highest group also had a significant or near-significant higher risk of all-cause death (HR, 1.63; 95% CI, 1.09-2.43; P = .02) and cardiovascular death (HR, 1.57; 95% CI, 0.95-2.61; P = .08) Table 3. Using adjusted HR curves, a U-shaped association between HDL-C and adverse events was evident with higher mortality at both very high and low HDL-C levels (Figure 1).
Table 3. Association Between HDL-C Levels and Adverse Outcomes.
| Cohort | Model | HDL-C level | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| ≤30 mg/dL | 30-40 mg/dL | 40-60 mg/dL (Reference) | 60-80 mg/dL | >80 mg/dL | ||||||
| HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value | |||
| All-cause death | ||||||||||
| UKB | Unadjusted | 1.78 (1.48-2.15) | <.001 | 1.20 (1.08-1.34) | <.001 | 1 | 1.07 (0.91-1.25) | .42 | 1.58 (1.16-2.14) | .004 |
| Adjusteda | 1.33 (1.07-1.64) | .009 | 1.00 (0.89-1.13) | .96 | 1 | 1.27 (1.08-1.50) | .005 | 1.96 (1.42-2.71) | <.001 | |
| GRS adjustedb | 1.31 (1.06-1.63) | .01 | 1.01 (0.89-1.13) | .91 | 1 | 1.28 (1.08-1.51) | .005 | 1.96 (1.42-2.72) | <.001 | |
| EmCAB | Unadjusted | 1.26 (1.09-1.46) | .002 | 0.93 (0.82-1.05) | .21 | 1 | 1.19 (0.99-1.44) | .07 | 1.43 (0.97-2.10) | .07 |
| Adjusteda | 1.22 (1.03-1.45) | .02 | 0.97 (0.85-1.11) | .65 | 1 | 1.20 (0.97-1.47) | .09 | 1.63 (1.09-2.43) | .02 | |
| Cardiovascular death | ||||||||||
| UKB | Unadjusted | 2.01 (1.61-2.52) | <.001 | 1.24 (1.08-1.41) | .002 | 1 | 0.91 (0.74-1.12) | .39 | 1.23 (0.80-1.87) | .34 |
| Adjusteda | 1.42 (1.09-1.85) | .009 | 0.99 (0.85-1.14) | .86 | 1 | 1.17 (0.94-1.47) | .17 | 1.71 (1.09-2.68) | .02 | |
| GRS adjustedb | 1.44 (1.10-1.89) | .007 | 1.01 (0.87-1.17) | .92 | 1 | 1.18 (0.94-1.48) | .16 | 1.68 (1.07-2.63) | .02 | |
| EmCAB | Unadjusted | 1.37 (1.13-1.68) | .002 | 1.05 (0.88-1.24) | .59 | 1 | 1.10 (0.83-1.44) | .52 | 1.46 (0.88-2.44) | .14 |
| Adjusteda | 1.35 (1.06-1.72) | .01 | 1.13 (0.93-1.36) | .22 | 1 | 1.04 (0.76-1.41) | .82 | 1.57 (0.95-2.61) | .08 | |
Abbreviations: CAD, coronary artery disease; eGFR, estimated glomerular filtration rate; EmCAB, Emory Cardiovascular Biobank; GRS, genetic risk score; HDL-C, high-density lipoprotein cholesterol; HR, hazard ratio; LDL-C, low-density lipoprotein cholesterol; UKB, UK Biobank.
SI conversion factor: To convert HDL-C to millimoles per liter, multiply by 0.0259.
UKB CAD cohort model adjusted for age, sex, race and ethnicity, body mass index, hypertension, diabetes, smoking, triglycerides, LDL-C, stroke history, heart attack history, eGFR, frequent alcohol use (defined as alcohol consumption ≥3 times per week); EmCAB model adjusted for age, sex, race and ethnicity, body mass index, hypertension, diabetes, current/former smoking, triglycerides, LDL-C, heart failure history, myocardial infarction history, eGFR, frequent alcohol use (defined as ≥8 alcohol beverages per week), statin use, aspirin use, β-blocker use, and angiotensin-converting enzyme inhibitor/angiotensin receptor blocker use.
UKB CAD cohort model adjusted for covariates listed above plus HDL-C GRS plus top 10 principal components.
Significant interaction with diabetes for all-cause death (β = 0.89; SE = 0.41; P = .03), cardiovascular death (β = 1.02; SE = 0.49; P = .04), and interaction with age (β = −1.22; SE = 0.49; P = .01) for cardiovascular death were noted in the group with HDL-C levels greater than 80 mg/dL (Figure 2). Compared with patients without diabetes, patients with diabetes and an HDL-C level greater than 80 mg/dL had a higher risk of all-cause death (HR, 3.03; 95% CI, 1.62-5.65; P < .001 vs HR, 1.15; 95% CI, 0.68-1.94; P = .61) and cardiovascular death (HR, 3.14; 95% CI, 1.55-6.38; P = .002 vs HR, 0.97; 95% CI, 0.48-1.94, P = .93). Patients younger than 65 years had a higher risk of cardiovascular death than patients 65 years or older (HR, 3.05; 95% CI, 1.52-6.11; P = .002 vs HR, 1.02; 95% CI, 0.50-2.11; P = .95) (Figure 2 and eTable 1 in the Supplement).
Association of the GRS
A positive linear association between the HDL-C GRS and HDL-C levels was observed, with a 1-SD higher HDL-C GRS being associated with a 3.03-mg/dL higher HDL-C level (95% CI, 2.83-3.22; P < .001; R2 = 0.06) (eFigure 4 in the Supplement). In unadjusted models, the HDL-C GRS was not associated with risk of all-cause death (HR, 1.01; 95% CI, 0.97-1.06; P = .65) or cardiovascular death (HR, 1.05; 95% CI, 0.99-1.11; P = .13). After adding the HDL-C GRS to the fully adjusted models, the association with HDL-C level greater than 80 mg/dL was not attenuated (Table 3), indicating that HDL-C genetic variations in the GRS do not contribute substantially to the risk.
Discussion
Results of this cohort study suggest that in individuals with CAD, compared with those with normal HDL-C levels (40-60 mg/dL), those with very high HDL-C levels (>80 mg/dL) have a higher risk of all-cause and cardiovascular death, independent of traditional cardiovascular risk factors and alcohol use. Even though prevalence of some traditional risk factors, such as diabetes, myocardial infarction history, heart attack history, and BMI, decreased with increasing HDL-C levels, we observed a higher risk of mortality in this group after adjusting for these risk factors. Our study addresses an important knowledge gap as current risk calculators used in the general population input high HDL-C level as a protective factor, whereas at very high levels, this protective effect does not appear to hold true and, in fact, may confer increased risk.
Prior studies have indicated that low HDL-C levels are associated with higher risk of adverse outcomes compared with those with normal HDL-C levels (ie, 40-60 mg/dL).21,22 More recently, studies have indicated that very high HDL-C level is also associated with adverse outcomes in the general population without known CAD.9,10,23 To our knowledge, this is the first study investigating the association of very high HDL-C levels with cardiovascular mortality among patients with CAD. Even a moderately high HDL-C level of 60 to 80 mg/dL was associated with higher all-cause death risk in the UKB. Within both cohorts with CAD, only less than 2% of the participants had HDL-C levels greater than 80 mg/dL, a much lower prevalence than the 7% participants reported in the study of general population in Canada,10 as well as approximately 8% to 9% reported in the general population of northern China.24 The other Danish study reported 31% of women and 11% of men had HDL-C levels of 77 mg/dL or greater.9 These data may imply reduced survival of this subgroup of patients with CAD and very high HDL-C levels.
A sex interaction with men having greater mortality risk was noted in the UKB, a finding that was not replicated in the EmCAB. Within the EmCAB, the association between very high HDL-C levels and mortality risk was stronger among patients with diabetes. The interaction with diabetes was not identified in the UKB, likely owing to a lower proportion of patients with diabetes in the high HDL-C category. Future studies of larger sample size are needed to replicate these observed interactions. Further, higher cardiovascular mortality risk within the groups with HDL-C levels greater than 80 mg/dL was more prominent among participants younger than 65 years compared with participants 65 years or older within the EmCAB, with a similar pattern observed in the UKB. This result of stratification by age group is consistent with a previous study of all-cause mortality among the general population of northern China.24
Although the HDL-C GRS was closely associated with HDL-C levels in a linear fashion, the GRS only accounts for 6% of the variation of HDL-C levels. This explains the persistence of a U-shaped association between HDL-C level and outcomes after adjustment for the HDL-C GRS. Both genetic profile and environmental exposures can affect HDL-C levels, but the associations with all-cause and cardiovascular mortality appear to be largely driven by nongenetic causes. Further, mendelian randomization studies have not supported the causal role for HDL-C in cardiovascular events.8,25 Moreover, raising HDL-C levels with pharmacologic interventions has also not influenced cardiovascular risk.2,3,4,5,6,7
Prior studies have indicated that excessive alcohol intake may be a confounder that raises HDL-C level; however, adjusting for alcohol frequency use did not attenuate the association with very high HDL-C levels in either of the populations studied. Even adjustment for alcohol intake using a more continuous scale or only including participants without frequent alcohol use did not significantly affect the association between very high HDL-C levels and outcomes (eTables 1 and 2 in the Supplement).
The association between very high HDL-C level and risk of mortality may be driven by heterogenous groups of HDL-C particles, which may play critical roles beyond the measured levels of HDL-C. Further, the function of HDL-C particles could be converted from anti-inflammatory to proinflammatory under certain circumstances, such as increased oxidative stress.26 Future studies are warranted to identify specific mechanisms by investigating the functionality of various major HDL-C apolipoprotein components.
Strengths and Limitations
There were several strengths to our study. This was a large study conducted in 2 independent high-risk populations with CAD in 2 countries on separate continents who had long-term follow-up for both incident all-cause and cardiovascular death events. Our findings may not, however, be generalizable to nonatherosclerotic cardiovascular diseases. Further studies will need to investigate whether the risk of very high HDL-C levels only pertains to men as noted in the UKB cohort, but not validated in the EmCAB cohort. The measure of frequent alcohol consumption differed among the 2 cohorts, and it may not capture the full pattern of alcohol use in both cohorts. The definition of CAD in the UKB was based on ICD-10 codes, which have been developed for billing purposes. Although such definitions have been widely used in clinical and epidemiologic research, potential misclassification is possible. Future investigations will also have to address the value of apolipoprotein A levels and family history in patients with very high HDL-C levels. Finally, very high HDL-C levels showed similar increased risk in all-cause and cardiovascular death. Although the observed association has been adjusted extensively for cardiovascular risk factors, other unmeasured confounders may contribute to the association. Further studies are warranted to investigate the association of very high HDL-C level with other cause-specific outcomes.
Conclusions
In this cohort study, results suggest that an HDL-C level greater than 80 mg/dL was associated with higher risk of all-cause and cardiovascular death in populations with CAD, compared with those with normal HDL-C levels, a finding that until now has only been reported in individuals without known CAD. Our findings have critical and meaningful clinical impact in risk prediction.
eTable 1. Association Between HDL-C Levels and Adverse Outcomes, Sensitivity Analysis
eTable 2. Association Between HDL-C Levels and Adverse Outcomes, Using the Continuous Measure of Alcohol Use as a Covariate
eFigure 1. HDL-C Distributions
eFigure 2. Proportion of Events by HDL-C Levels
eFigure 3. Kaplan-Meier Curves for HDL-C Categories and Adverse Outcomes
eFigure 4. Association Between the HDL-C GRS and HDL-C Levels in UKB
References
- 1.Wilson PW, Abbott RD, Castelli WP. High-density lipoprotein cholesterol and mortality: the Framingham Heart Study. Arteriosclerosis. 1988;8(6):737-741. doi: 10.1161/01.ATV.8.6.737 [DOI] [PubMed] [Google Scholar]
- 2.Boden WE, Probstfield JL, Anderson T, et al. ; AIM-HIGH Investigators . Niacin in patients with low HDL cholesterol levels receiving intensive statin therapy. N Engl J Med. 2011;365(24):2255-2267. doi: 10.1056/NEJMoa1107579 [DOI] [PubMed] [Google Scholar]
- 3.Landray MJ, Haynes R, Hopewell JC, et al. ; HPS2-THRIVE Collaborative Group . Effects of extended-release niacin with laropiprant in high-risk patients. N Engl J Med. 2014;371(3):203-212. doi: 10.1056/NEJMoa1300955 [DOI] [PubMed] [Google Scholar]
- 4.Lincoff AM, Nicholls SJ, Riesmeyer JS, et al. ; ACCELERATE Investigators . Evacetrapib and cardiovascular outcomes in high-risk vascular disease. N Engl J Med. 2017;376(20):1933-1942. doi: 10.1056/NEJMoa1609581 [DOI] [PubMed] [Google Scholar]
- 5.Schwartz GG, Olsson AG, Abt M, et al. ; dal-OUTCOMES Investigators . Effects of dalcetrapib in patients with a recent acute coronary syndrome. N Engl J Med. 2012;367(22):2089-2099. doi: 10.1056/NEJMoa1206797 [DOI] [PubMed] [Google Scholar]
- 6.Bowman L, Hopewell JC, Chen F, et al. ; HPS3/TIMI55–REVEAL Collaborative Group . Effects of anacetrapib in patients with atherosclerotic vascular disease. N Engl J Med. 2017;377(13):1217-1227. doi: 10.1056/NEJMoa1706444 [DOI] [PubMed] [Google Scholar]
- 7.Barter PJ, Caulfield M, Eriksson M, et al. ; ILLUMINATE Investigators . Effects of torcetrapib in patients at high risk for coronary events. N Engl J Med. 2007;357(21):2109-2122. doi: 10.1056/NEJMoa0706628 [DOI] [PubMed] [Google Scholar]
- 8.Voight BF, Peloso GM, Orho-Melander M, et al. Plasma HDL cholesterol and risk of myocardial infarction: a mendelian randomisation study. Lancet. 2012;380(9841):572-580. doi: 10.1016/S0140-6736(12)60312-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Madsen CM, Varbo A, Nordestgaard BG. Extreme high high-density lipoprotein cholesterol is paradoxically associated with high mortality in men and women: 2 prospective cohort studies. Eur Heart J. 2017;38(32):2478-2486. doi: 10.1093/eurheartj/ehx163 [DOI] [PubMed] [Google Scholar]
- 10.Ko DT, Alter DA, Guo H, et al. High-density lipoprotein cholesterol and cause-specific mortality in individuals without previous cardiovascular conditions: the CANHEART Study. J Am Coll Cardiol. 2016;68(19):2073-2083. doi: 10.1016/j.jacc.2016.08.038 [DOI] [PubMed] [Google Scholar]
- 11.Sudlow C, Gallacher J, Allen N, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12(3):e1001779. doi: 10.1371/journal.pmed.1001779 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Huang Y, Hui Q, Gwinn M, et al. Sexual differences in genetic predisposition of coronary artery disease. Circ Genom Precis Med. 2021;14(1):e003147. doi: 10.1161/CIRCGEN.120.003147 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.World Health Organization . International Statistical Classification of Diseases and Related Health Problems. 10th Revision, 2nd ed. World Health Organization; 2004. [Google Scholar]
- 14.Bycroft C, Freeman C, Petkova D, et al. The UK Biobank resource with deep phenotyping and genomic data. Nature. 2018;562(7726):203-209. doi: 10.1038/s41586-018-0579-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ko YA, Hayek S, Sandesara P, Samman Tahhan A, Quyyumi A. Cohort profile: the Emory Cardiovascular Biobank (EmCAB). BMJ Open. 2017;7(12):e018753. doi: 10.1136/bmjopen-2017-018753 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Levey AS, Stevens LA, Schmid CH, et al. ; CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) . A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9):604-612. doi: 10.7326/0003-4819-150-9-200905050-00006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Jeppesen J, Hansen TW, Olsen MH, et al. C-reactive protein, insulin resistance, and risk of cardiovascular disease: a population-based study. Eur J Cardiovasc Prev Rehabil. 2008;15(5):594-598. doi: 10.1097/HJR.0b013e328308bb8b [DOI] [PubMed] [Google Scholar]
- 18.Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446):496-509. doi: 10.1080/01621459.1999.10474144 [DOI] [Google Scholar]
- 19.Meira-Machado L, Cadarso-Suárez C, Gude F, Araújo A. smoothHR: an R package for pointwise nonparametric estimation of hazard ratio curves of continuous predictors. Comput Math Methods Med. 2013;2013:745742. doi: 10.1155/2013/745742 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Klarin D, Damrauer SM, Cho K, et al. ; Global Lipids Genetics Consortium; Myocardial Infarction Genetics (MIGen) Consortium; Geisinger-Regeneron DiscovEHR Collaboration; VA Million Veteran Program . Genetics of blood lipids among ~300,000 multiethnic participants of the Million Veteran Program. Nat Genet. 2018;50(11):1514-1523. doi: 10.1038/s41588-018-0222-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Goldbourt U, Yaari S, Medalie JH. Isolated low HDL cholesterol as a risk factor for coronary heart disease mortality: a 21-year follow-up of 8000 men. Arterioscler Thromb Vasc Biol. 1997;17(1):107-113. doi: 10.1161/01.ATV.17.1.107 [DOI] [PubMed] [Google Scholar]
- 22.Gordon T, Castelli WP, Hjortland MC, Kannel WB, Dawber TR. High-density lipoprotein as a protective factor against coronary heart disease: the Framingham Study. Am J Med. 1977;62(5):707-714. doi: 10.1016/0002-9343(77)90874-9 [DOI] [PubMed] [Google Scholar]
- 23.Liu C, Dhindsa D, Almuwaqqat Z, Sun YV, Quyyumi AA. Very high high-density lipoprotein cholesterol levels and cardiovascular mortality. Am J Cardiol. 2022;167:43-53. doi: 10.1016/j.amjcard.2021.11.041 [DOI] [PubMed] [Google Scholar]
- 24.Li X, Guan B, Wang Y, et al. Association between high-density lipoprotein cholesterol and all-cause mortality in the general population of northern China. Sci Rep. 2019;9(1):14426. doi: 10.1038/s41598-019-50924-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Holmes MV, Asselbergs FW, Palmer TM, et al. ; UCLEB Consortium . Mendelian randomization of blood lipids for coronary heart disease. Eur Heart J. 2015;36(9):539-550. doi: 10.1093/eurheartj/eht571 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Navab M, Reddy ST, Van Lenten BJ, Fogelman AM. HDL and cardiovascular disease: atherogenic and atheroprotective mechanisms. Nat Rev Cardiol. 2011;8(4):222-232. doi: 10.1038/nrcardio.2010.222 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Association Between HDL-C Levels and Adverse Outcomes, Sensitivity Analysis
eTable 2. Association Between HDL-C Levels and Adverse Outcomes, Using the Continuous Measure of Alcohol Use as a Covariate
eFigure 1. HDL-C Distributions
eFigure 2. Proportion of Events by HDL-C Levels
eFigure 3. Kaplan-Meier Curves for HDL-C Categories and Adverse Outcomes
eFigure 4. Association Between the HDL-C GRS and HDL-C Levels in UKB