Abstract
BACKGROUND:
Lipoprotein(a) (Lp(a)) has not been well-studied in a nationally representative US cohort.
OBJECTIVE:
The objective of this study was to investigate the distribution of Lp(a) and its associations with nonfatal cardiovascular events in a nationally representative cohort.
METHODS:
Cross-sectional analysis using the National Health and Nutrition Examination Survey III cohort (1991–1994). We compared Lp(a) levels across demographics and tested the associations between Lp(a) and patient-reported nonfatal myocardial infarction (MI) and/or stroke using multivariate logistic regression.
RESULTS:
Median Lp(a) was 14 mg/dL (interquartile range [IQR]: 3, 32) (n = 8214). 14.7% (95% CI: 13.6%-15.9%) had Lp(a) ≥50 mg/dL. Women had slightly higher median Lp(a) than men (14 mg/dL [IQR: 4, 33] vs 13 [(IQR: 3, 30], P = .001). Non-Hispanics blacks had the highest median Lp(a) (35 mg/dL [IQR: 21, 64]), followed by non-Hispanic whites (12 mg/dL [IQR: 3, 29]) and Mexican Americans (8 mg/dL [IQR:1, 21]). In multivariate analysis, Lp(a) was associated (odds ratio per SD increase [95% CI], P-value) with MI (1.41 [1.14–1.75], P = .001), but not stroke (1.14 [0.91–1.44], P = .26). Lp(a) associated with MI in men (1.52 [1.13–2.04], P = .006), non-Hispanic whites (1.60 [1.27–2.03], P < .001), and Mexican Americans (2.14 [1.29–3.55], P = .003), but not women or non-Hispanic blacks. Lp(a) was not associated with stroke among any subgroups.
CONCLUSION:
In a nationally representative US cohort, 1 in 7 had Lp(a) ≥50 mg/dL, the guidelines-recommended threshold to consider Lp(a) a risk enhancing factor. Lp(a) was associated with nonfatal MI but not stroke, although there were differential associations by sex and race/ethnicity. Future nationally representative cohorts should test Lp(a) to get an updated estimation.
Keywords: Lipoprotein (a), Epidemiology, Myocardial infarction, Stroke, Cardiovascular disease, Risk
Introduction
Lipoprotein(a) (Lp(a)) is a plasma lipoprotein composed of cholesterol, cholesteryl esters, apolipoprotein B100, apolipoprotein(a), and a small amount of triglycerides and carbohydrates.1,2 Mendelian randomization and genome-wide association studies have shown that Lp(a) confers additional risk for cardiovascular disease (CVD) that increases linearly and is independent from LDL.1,3–6 Lp(a) is a recognized risk factor for diverse cardiovascular disorders, including coronary heart disease (CHD), stroke, aortic stenosis, peripheral vascular disease, and heart failure.1,7–9 In meta-analysis, each 1 standard deviation (SD) increase in Lp(a) is associated with an increased risk of CHD (risk ratio 1.13 [95% CI: 1.09 to 1.18]) and stroke (1.10 [95% CI: 1.02 to 1.18]).8 It appears that the risk for CHD and stroke emerges at Lp(a) levels above 30 mg/dL and 50 mg/dL, respectively.8 In cholesterol guidelines, the currently recognized threshold above which Lp(a) is considered a risk-enhancing factor is 50 mg/dL (125 nmol/L).10,11
Given that Lp(a) increases the risk for multiple cardiac conditions, accurate estimations for the frequency of elevated Lp(a) in the US population have important implications for screening, prevention, and therapeutic purposes. Among 531,144 patients from a US national referral laboratory, Varvel et al. found a mean (SD) Lp(a) of 34.0 mg/dL (40.0 mg/dL) with a median of 17 mg/dL (interquartile range [IQR]: 7, 47) and 24.0% of subjects with levels >50 mg/dL.12 Interestingly, 915 patients from a tertiary cardiovascular/lipid referral center in the same study had a higher mean (SD) Lp(a) of 40 mg/dL (49 mg/dL), median of 19 mg/dL (IQR: 6, 62), and 29.2% with levels >50 mg/dL. Other large cohorts with estimations for Lp(a) levels in the United States include ARIC (median 18.3 mg/dL [IQR: 6.9, 43.8])13 and MESA (median 17.1 mg/dL [IQR: 7.4, 40]).14 Differences between these studies suggests that the proportion of the population with Lp(a) ≥50 mg/dL may differ by patient selection. However, because none of these samples were selected to be nationally representative, the results of these studies may not accurately represent the epidemiology of Lp(a) in the general US population.
Lp(a) was previously reported, albeit without detail, from the National Health and Nutrition Examination Survey III (NHANES III), a cross-sectional cohort designed to be nationally representative. The report showed a median Lp(a) of 23.0 mg/dL (IQR 9.0, 46.0) among a subset of subjects in a meta-analysis (details and methodology of which are not available).8 Other reports from these data describe Lp(a) distributions among children,15 associations of kidney function to Lp(a) levels,16 and genetic association studies between LPA single-nucleotide polymorphisms and Lp(a) serum values.17 However, estimations of Lp(a) for the entire NHANES III adult cohort and consequently the US population were incompletely elucidated. NHANES III is the only nationally representative US cohort to ever measure Lp(a). Thus, although the data are from 1991–1994, detailed evaluations have important epidemiologic and clinical implications.
Therefore, we used the NHANES III to study the epidemiology of Lp(a) and whether Lp(a) is associated with risk for nonfatal CVD among this cross-sectional nationally representative US sample. Our a priori hypothesis was that Lp(a) levels differ from prior estimations and that Lp(a) will be associated with a patient-reported history of nonfatal myocardial infarction (MI) and stroke. We also hypothesized, based on previous reports in the literature, that Lp(a) levels and the associated risk with nonfatal MI and stroke will differ by sex and race/ethnicity.1,7,18–24
Methods
Study population
We performed an observational cross-sectional analysis using data from NHANES III.25 NHANES III was collected in two phases. Each phase comprised a national probability sample.25 Lp(a) was measured among all participants of the second phase (1991–1994) only. We limited our sample to adults (ages 17+) that had Lp(a) measured. NHANES III is a nationally representative cross-sectional cohort of the US population collected between 1988 and 1994 from a multi-stage probability sample of about 40,000 noninstitutionalized civilians aged 2 months and older. The plan and operations for data collection and reporting, including sampling have been previously described.25 The study was approved by the Yale University Institutional Review Board.
Outcomes
The outcomes were respondent-reported personal history of nonfatal heart attack or stroke, asked as “Has a doctor every told you that you had a heart attack?” and “Has a doctor ever told you that you had a stroke?”
Assay
Lp(a) was tested from serum samples. The Lp(a) assay utilized in this study is an enzyme-linked immunosorbent assay directed against apo(a), which is reported in mg/dL (Macra test kit, Strategic Diagnostics Inc., Newark, DE).26 The assay’s plate reader was calibrated using six standards provided by the manufacturer. Quality control was performed in accordance with the methods of the Lipid Research Clinics Program. Additional information on the assays utilized for the laboratory testing have been previously outlined elsewhere.26
Analysis
Because Lp(a) was not normally distributed in the sample, we report Lp(a) distributions as percentiles, including minimum, 5th, 25th, median, 75th, 95th, and maximum. We also report on the proportion of the population with Lp(a) above certain thresholds (≥30, ≥50, ≥70, and ≥90 mg/dL). In the main manuscript, we report results by age, sex, race/ethnicity, and history of MI or stroke for the cut-point of 50 mg/dL. The other cut-points are reported in the supplement. The cut-point of 50 mg/dL was chosen as the primary cut-point because it is the most recent guidelines-recognized reference value for increased risk.10,27 Comparisons for descriptive statistics were made using Somers’ D test. Proportions were compared using χ2 test. LDL-C was calculated using Friedewald28 and Martins-Hopkins methods.29 Non-HDL-C and LDL-C were corrected for Lp(a) mass concentration (LDL-Ccorr = LDL–[Lp(a)*0.3]).30
We then performed univariate and multivariate logistic regression for the outcomes of nonfatal MI and stroke using multiple covariates. The covariates included in our analyses were demographics (age [range: 17 through 90 years], sex, race/ethnicity [non-Hispanic white, non-Hispanic black, Mexican American, or other], marital status, years of education, and income below $20,000 per year), clinical risk factors (respondent reported history of ever being a smoker, diabetes, hypertension, and family history of MI before age 50 years), and biomarkers (non-HDL-C, c-reactive protein [CRP], and estimated glomerular filtration rate [eGFR]). Non-HDL-C was calculated by subtracting HDL-C from total cholesterol and corrected for Lp(a) as described previously. eGFR was calculated using the 6-variable Modification of Diet in Renal Disease equation. Data were evaluated for missing covariates. We planned to impute missing data if >10% of the sample in the final analyses had missing values.
The continuous covariates were mean-centered and interpreted as odds ratio (OR) per one standard deviation (SD) increase. We reported CRP as odds per 1 mg/dL increase. In the final models, we made an a priori decision to exclude covariates with a P-value of > .05, except for Lp(a), age, sex, race/ethnicity, and non-HDL-C. If models had evidence of poor fit, we added the minimal amount of excluded variables, higher-order polynomials of continuous variables, and interactions between covariates (in that order) until postestimation tests indicated there was no variable misspecification. The final models were interpreted using average marginal effects with outputs at representative values for Lp(a) to calculate OR for outcomes at specific values of Lp(a) (ie, 30, 50, 75, 100, and 125 mg/dL) compared with having a Lp(a) level at the median of the entire population.
In secondary analysis, the final multivariate models were run for each sex and race/ethnicity category. We performed two sensitivity tests on the final models. First, we tested Lp(a) as a dichotomous variable (< or ≥30, 50, 70, and 90 mg/dL). Second, to ensure appropriate control for age, we limited our sample to those aged >50 years and repeated the final models.
Univariate and multivariate logistic regression considered a 2-tailed P-value <.05 as statistically significant. All analyses were adjusted for sample weights. We conducted data analysis from March 2019 through May 2020. Data were analyzed using Stata 16 (StataCorp, LLC).
Results
Population characteristics
Unadjusted population characteristics are shown in Table 1. The cohort had a median age of 42 years (IQR: 29 to 64) and 56.5% were women. The unadjusted median Lp(a) was 18 mg/dL (IQR: 4 to 36).
Table 1.
Population characteristics (n = 9128)
| Median (IQR) or % | Missing, % (n) | |
|---|---|---|
| Lp(a), mg/dL | 18 (4, 36) | 4.4% (406) |
| Age (y) | 42 (29, 64) | 0.0% (0) |
| Female | 56.5% | 0.0% (0) |
| Race/ethnicity | - | 0.0% (0) |
| Non-Hispanic white | 38.1% | - |
| Non-Hispanic black | 30.3% | - |
| Mexican American | 26.8% | - |
| Other | 4.8% | - |
| Marital status | - | 0.2% (19) |
| Never married | 21.4% | - |
| Divorced/separated | 10.9% | - |
| Widowed | 11.0% | - |
| Married/living with partner | 56.8% | - |
| Education (y) | 12 (9, 13) | 0.7% (62) |
| Income <$20,000 | 49.7% | 2.1% (189) |
| Ever smoker | 47.7% | 0.0% (0) |
| Diabetes | 8.1% | 0.0% (2) |
| Hypertension | 27.6% | 0.8% (70) |
| Relative with MI before age 50 | 12.8% | 1.4% (125) |
| Non-HDL-C, mg/dL* | 139 (111, 171) | 5.1% (467) |
| LDL-C (Friedewald method), mg/dL* | 112 (88, 138) | 5.1% (467) |
| LDL-C (Martin-Hopkins method), mg/dL* | 114 (91, 141) | 5.1% (467) |
| HDL-C, mg/dL | 48 (40, 59) | 5.0% (450) |
| Triglycerides, mg/dL | 113 (79, 170) | 4.4% (400) |
| C-reactive protein, mg/dL | 0.21 (0.21, 0.50) | 4.6% (422) |
| eGFR, mL/min/1.73 m2 | 75 (64, 87) | 4.8% (435) |
All variables were not normally distributed. Values are percentage or median (interquartile range). Values are not adjusted for complex survey design. eGFR, estimated glomerular filtration rate; HDL, high-density lipoprotein; IQR, interquartile range; Lp(a), lipoprotein(a); MI, myocardial infarction.
Corrected for Lp(a).
Epidemiology of Lp(a)
Lp(a) was not normally distributed (Fig. 1). The sample adjusted, therefore nationally representative, median was 14 mg/dL (IQR: 3 to 32) with a rightward skewed tail (Table 2). 27.8% (95% confidence interval (CI): 26.3% to 29.3%) had Lp(a) ≥30 mg/dL, 14.7% (95% CI: 13.6% to 15.9%) were ≥50 mg/dL, 5.6% (95% CI 5.0% to 6.3%) were ≥70 mg/dL, and 2.6% (95% CI: 2.2% to 3.1%) were ≥90 mg/dL.
Figure 1.
Distribution of lipoprotein(a) in the sample population overall and by sex and race/ethnicity.
Table 2.
Distribution of Lp(a) in the US population by age, sex, race/ethnicity, and history of MI or stroke
| Lp(a) by percentile | Percentage above ≥50 mg/dL | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Min | 5th % | 25th % | Median | 75th % | 95th % | Max | P-value | Percentage (95% CI) | P-value | |
| Entire population (n = 8722) | 0 | 0 | 3 | 14 | 32 | 71 | 276 | - | 14.7% (13.6%, 15.9%) | - |
| Age | - | - | - | - | - | - | - | .37 | - | .02 |
| <35 (n = 3091) | 0 | 0 | 3 | 14 | 31 | 68 | 276 | - | 13.2% (11.4%, 15.1%) | - |
| 35–54 (n = 2548) | 0 | 0 | 4 | 14 | 31 | 71 | 177 | - | 14.5% (12.6%, 16.5%) | - |
| 55–74 (n = 2092) | 0 | 0 | 3 | 13 | 32 | 76 | 219 | - | 17.1% (15.0%, 19.4%) | - |
| 75+ (n = 991) | 0 | 0 | 4 | 17 | 36 | 73 | 189 | - | 18.0% (15.1%, 21.3%) | - |
| Sex | - | - | - | - | - | - | - | .001 | - | .002 |
| Male (n = 3811) | 0 | 0 | 3 | 13 | 30 | 68 | 219 | - | 12.9% (11.3%, 14.6%)) | - |
| Female (n = 4911) | 0 | 0 | 4 | 14 | 33 | 73 | 276 | - | 16.4% (15.0%, 17.9%) | - |
| Race/ethnicity | - | - | - | - | - | - | - | <.001 | - | <.001 |
| Non-Hispanic white (n = 3356) | 0 | 0 | 3 | 12 | 29 | 66 | 189 | - | 11.9% (10.6%, 13.3%) | - |
| Non-Hispanic black (n = 2585) | 0 | 2 | 21 | 35 | 64 | 114 | 276 | - | 39.0% (36.9%, 41.1%) | - |
| Mexican American (n = 2364) | 0 | 0 | 1 | 8 | 21 | 58 | 134 | - | 7.3% (6.2%, 8.6%) | - |
| Other (n = 417) | 0 | 0 | 3 | 13 | 28 | 68 | 123 | - | 12.6% (8.9%, 17.9%) | - |
| MI | - | - | - | - | - | - | - | .14 | - | <.001 |
| No (n = 8359) | 0 | 0 | 3 | 14 | 31 | 71 | 276 | - | 14.4% (13.3%, 15.5%) | - |
| Yes (n = 358) | 0 | 0 | 2 | 19 | 51 | 118 | 191 | - | 25.4% (18.9%, 33.1%) | - |
| Stroke | - | - | - | - | - | - | - | .40 | - | .09 |
| No (n = 8441) | 0 | 0 | 3 | 14 | 32 | 71 | 276 | - | 14.6% (13.5%, 15.8%) | - |
| Yes (n = 276) | 0 | 0 | 5 | 13 | 36 | 105 | 156 | - | 19.7% (14.0%, 27.1%) | - |
CI, confidence interval; Lp(a), lipoprotein (a); MI, myocardial infarction.
Lp(a) distribution by demographics
Lp(a) levels varied by sex and race/ethnicity. Women had a slightly higher median Lp(a) (14 mg/dL [IQR: 4 to 33]) than men (13 mg/dL [IQR: 3 to 30]) (P = .001) and proportion with Lp(a) ≥50 mg/dL (16.5% [95% CI: 15.0% to 17.9%] vs 12.9% [95% CI: 11.3% to 14.6%], P = .002). Lp(a) also varied by race/ethnicity. Non-Hispanic blacks had the highest median Lp(a) of 35 mg/dL (IQR: 21 to 43) and proportion ≥50 mg/dL (39.0% [95% CI: 36.9% to 41.1%]), and Mexican Americans had the lowest median Lp(a) of 8 mg/dL (IQR: 1 to 21) and proportion ≥50 mg/dL of 7.3% (95% CI: 6.2% to 8.6%); Non-Hispanic whites had median Lp(a) of 12 mg/dL (IQR: 3 to 29) and proportion ≥50 mg/dL of 11.9% (95% CI: 10.6% to 13.3%) (Somer’s D P < .001, χ2 < .001, respectively). Lp(a) distribution did not differ by age, although those above age 55 years tended to have a higher proportion ≥50 mg/dL. See supplement Table 1 for Lp(a) proportions specific populations at threshold of ≥30, ≥70, and ≥90 mg/dL.
Lp(a) distribution by history of MI or stroke
For MI there was no difference in distribution by reported history of MI (Table 2). However, there was a larger proportion with Lp(a) ≥50 mg/dL among those with vs without reported history of MI (25.4% [95% CI: 18.9% to 33.1%] and 14.4% [95% CI: 13.3% to 15.5%], respectively P < .001). There was no association between Lp(a) distributions or proportion ≥50 mg/dL in those that reported stroke and no stroke. However, there were a higher proportion above the thresholds of ≥70 (P = .02) and ≥90 mg/dL (P = .02) among those with strokes. See Supplement Table 1 for results at threshold of ≥30, 70 and 90 mg/dL.
Multivariate analyses
In univariate analysis, Lp(a) was associated with MI, but not stroke (Supplement Table 2). In multivariate analyses, Lp(a) remained associated with MI (OR 1.43 [95% CI: 1.16 to 1.77], P = .001), but was not significantly associated with stroke (OR 1.14 [95% CI: 0.90 to 1.44], P = .29) (Table 3). These associations were slightly lower than the odds per SD increase of non-HDL-C. In all models age, diabetes, hypertension, CRP, and income <$20,000 annually were associated with both outcomes. Marital status, education, and eGFR were not associated with any outcomes. Notably, race/ethnicity was not associated with MI and sex was not associated with stroke. Ever being a smoker and having a relative with MI before age 50 years were associated with MI, but not stroke.
Table 3.
Association with covariates and MI or stroke in multivariate analyses (n=8,214)
| Covariate | MI | Stroke | ||
|---|---|---|---|---|
| OR (95% CI) | P-value | OR (95% CI) | P-value | |
| Lp(a) (per SD increase) | 1.43 (1.16, 1.77) | .001 | 1.14 (0.90, 1.44) | .29 |
| Demographics | - | - | - | - |
| Age (per year) | 1.07 (1.06, 1.09) | <.001 | 1.05 (1.04, 1.07) | <.001 |
| Female | 0.40 (0.27, 0.58) | <.001 | 0.73 (0.49, 1.10) | .13 |
| Race/ethnicity | - | .10 | - | .13 |
| Non-Hispanic white | Reference | - | Reference | - |
| Non-Hispanic black | 0.66 (0.40, 1.09) | - | 0.93 (0.56, 1.55) | - |
| Mexican American | 0.65 (0.42, 1.03) | - | 0.61 (0.37, 1.01) | - |
| Other | 1.29 (0.43, 3.91) | - | 1.80 (0.76, 4.23) | - |
| Marital status | - | - | - | - |
| Divorced/ separated | - | - | - | - |
| Widowed | - | - | - | - |
| Never married | - | - | - | - |
| Married/living with partner | - | - | - | - |
| Education (per additional year) | - | - | - | - |
| Income <$20,000 | 1.62 (1.09, 2.39) | .02 | 2.11 (1.33, 3.45) | .001 |
| Clinical risk factors and biomarkers | - | - | - | - |
| Ever smoker | 2.47 (1.71, 3.58) | <.001 | - | - |
| Diabetes | 1.99 (1.25, 3.17) | .004 | 3.35 (2.09, 5.37) | <.001 |
| Hypertension | 1.95 (1.35 (2.83) | <.001 | 1.89 (1.24, 2.86) | .003 |
| Relative with MI before age 50 | 2.43 (1.49, 3.95) | <.001 | - | - |
| Non-HDL-C (per SD increase) | 1.43 (1.22, 1.68) | <.001 | 1.21 (1.00, 1.47) | .051 |
| C-reactive protein (per 1 mg/dL) eGFR (mL/min/ 1.73 m2) | 1.31 (1.16, 1.48) | <.001 | 1.22 (1.07, 1.40) | .004 |
HDL, high-density lipoprotein; Lp(a), lipoprotein(a); MI, myocardial infarction.
Multivariate analyses considering polynomials and interactions
There remained evidence of poor model fit once nonsignificant variables were removed from the models. The MI model required eGFR to remain in the model and age as a quadratic term, and the stroke model required keeping ever being a smoker in the model. Lp(a) as a quadratic term did not improve the models. The sample for the final models consisted of 8214 subjects. Imputation was considered as unnecessary because subjects with missing covariate data for the final models did not exceed 10% (10.0%, n = 913).
Results of these final models were similar to the prior multivariate regression results. Lp(a) was associated with MI (OR 1.41 [95% CI: 1.14 to 1.75], P = .001]), but not stroke (OR 1.14 [95% CI: 0.91 to 1.44], P = .26) (Fig. 2). ORs for outcomes compared with the population median (14 mg/dL) at Lp(a) 5 30 mg/dL, 50 mg/dL, 75 mg/dL, 100 mg/dL, and 125 mg/dL are shown in Figure 2. Average marginal effects for these final multivariate analyses are available in Supplement Table 3.
Figure 2.
Odds ratio for MI or stroke at various Lp(a) values in fully adjusted multivariate models. Lp(a), lipoprotein(a); MI, myocardial infarction. *Sample size was 3234 when limiting to population with age over 50 years.
Lp(a)’ s association with MI or stroke by sex and race/ethnicity
In the final multinomial logistic regression models among men, Lp(a) was associated with MI (OR 1.52 [95% CI: 1.13 to 2.04], P = .006), but not stroke (Fig. 3). Among women, Lp(a) was not significantly associated with either outcome. Lp(a) in non-Hispanic whites and Mexican Americans was associated with MI (OR 1.60 [95% CI: 1.27 to 2.03], P=<.001 and 2.14 [95% CI: 1.29 to 3.55], P = .003, respectively), but not stroke. For non-Hispanic blacks, Lp(a) was not associated with any outcomes. For complete analyses by sex and race/ethnicity see Figure 3.
Figure 3.
Odds ratio for MI or stroke at various Lp(a) values by sex and race/ethnicity in fully adjusted multivariate models. *Note: max Lp(a) among Other was 123 mg/dL, thus predicted odds ratios above this range were not calculated. Lp(a), lipoprotein(a); MI, myocardial infarction.
Sensitivity tests
When we tested Lp(a) as a dichotomous variable (<or ≥50 mg/dL), it remained associated with MI, but not stroke. Other thresholds are shown in Supplement Table 4. When we removed age from the final model and limited the sample to age >50 years (n = 3234), Lp(a) remained significantly associated with MI and not stroke. Sensitivity tests are shown in Figure 2.
Discussion
Median Lp(a) levels in this nationally representative cohort were found to be slightly lower than previously estimated in non-nationally representative cohorts. Despite lower levels, a substantial portion of the population (about 1 in 7) had elevated Lp(a), defined as ≥50 mg/dL. After accounting for a multitude of demographic characteristics, clinical factors, and biomarkers, Lp(a) remained significantly associated with patient-reported history of nonfatal MI, but not stroke. To our knowledge, this is the first study to fully examine the epidemiology and association with reported clinical outcomes of Lp(a) in a nationally representative US cohort.
Our estimations of the distribution of Lp(a) are slightly lower compared with other large US cohort studies. In ARIC and MESA, two large cohorts of individuals from select US geographies, median Lp(a) was 18.3 mg/dL (IQR: 6.9, 43.8)13 and 17.1 mg/dL (IQR: 7.4, 40),14 respectively. At a national referral laboratory, the median Lp(a) was 17 mg/dL (IQR: 7, 47).12 In this same study, a tertiary cardiovascular/lipid referral center had a median of 19 mg/dL (IQR: 6, 62). Compared with recent findings from the INTERHEART study, Americans tended to have lower levels than Africans, similar to Arabs and South Asians, and higher than Chinese, Europeans, Latin Americans, and Southeast Asians.31 Our median fell within the range of prior studies from European databases, although most European studies tended to be lower (ATTICA 11.4 mg/dL, BRUN 8.8 mg/dL, COPEN 19.1 mg/dL, EAS 9.2 mg/dL, FINRISK 92 12.2 mg/dL, NPHSII 10.9 mg/dL, PRIME 2002 10.0 mg/dL, ULSAM 8.3 mg/dL, WOSCOPS 17.0 mg/dL, and ZUTE 12.3 mg/dL).8
The small differences in the medians between our observations and these other studies is most likely a result of sample selection. Compared with the other cohorts, NHANES III is a lower risk cohort of nonreferred, community-residing participants. The range of Lp(a) may vary greatly by the population sampled.8 Measuring the true epidemiology is only possible using a sample designed to be nationally representative, as was NHANES III. Thus, our estimations are likely closest to the true national median. Of note, the reported median Lp(a) is considerably different from a prior estimation from NHANES III that was included in a meta-analysis; in that study, the median Lp(a) of the NHANES III cohort was 23 mg/dL (IQR 9 to 46).8 Unfortunately, no information is available about the methodology used to assess median values. In our study, we utilized the entire adult sample of eligible individuals with no missing data and, using sample weighting, found a substantially lower median Lp(a) of 14 mg/dL. It is un-clear what accounts for these differences.
The proportion of individuals with the guidelines recognized threshold Lp(a) ≥50 mg/dL was also slightly lower compared with other US cohorts.11 Approximately 1 in 7 individuals (14.7%) in our cohort had an Lp(a) ≥50 mg/dL. The aforementioned study from Varvel found that in a national referral laboratory, 24.0% had levels above 50 mg/dL.12 Notably, 29.2% of samples from the tertiary cardiovascular/lipid referral center were above 50 mg/dL. However, this guideline threshold was primarily obtained from a white population. Other studies suggest that significant thresholds differ by race/ethnicity.21 Because other thresholds may be recognized as the literature evolves, we have included this in the supplement and find that there are differences across race/ethnicity at several different thresholds.
Estimations of the risk conferred from Lp(a) have varied. In recent meta-analysis among 126,634 participants, the risk ratio for CHD was 1.16 (95% CI: 1.11 to 1.22) per SD increase in Lp(a).1,8 These estimates fall within our CIs (OR 1.33 per SD increase[(95% CI: 1.08 to 1.64]). Although in our study no association between Lp(a) and stroke was detected, a recent meta-analysis comparing groups with high vs low Lp(a) showed an OR of 1.29 (95% CI: 1.06 to 1.58) in prospective studies and 1.41 (95% CI: 1.26 to 1.57) in case-control studies.9 Our study not finding a significant association of Lp(a) with stroke may be from the lower power of our study.
Similar to prior studies, sex was identified as a variable with differential distribution of Lp(a). Although previous smaller studies were inconclusive regarding sex differences in Lp(a) level,18 larger studies have suggested significant differences in Lp(a) levels by sex. In a recent meta-analysis (n = 126,634), women had 12% higher Lp(a) than men. Varvel et al. (n = 532,359) also found that women had higher Lp(a) (median 19 mg/dL [IQR: 8, 53]) than men (15 mg/dL [IQR: 7, 42]). Physiologically, Lp(a) levels have been shown to be responsive to estrogens and testosterone.3 Menopausal state was not available in the data used for our study. Further investigations are needed to understand the biologic mechanisms underlying the effect of sex on Lp(a) levels.
Prior studies also suggest that the association between Lp(a) and clinical events differs by sex, with women having slightly higher levels. In a combined cohort from JUPITER, the Women’s Health Study, and Women’s Health Initiative, women had increased risk of CVD from Lp(a) only when accompanied by high total cholesterol, whereas men had risk regardless of total cholesterol.32 In ATTICA, after adjusting for other clinical factors, Lp(a) associated with CHD in men but not women.33 However, in ARIC and the Reykjavik study, both men and women had an elevated risk of CHD associated with Lp(a).20,34 Thus, there may be sex-specific differences, particularly for coronary disease. The sample size may have limited our power to detect additional sex-specific associations with Lp(a) and cardiovascular events.
Lp(a) levels are also known to vary by race/ethnicity. We found that non-Hispanic blacks had higher levels, whereas Mexican Americans had lower levels than non-Hispanic whites. In MESA, blacks had highest median Lp(a) at 35.1 mg/dL, whereas Caucasian and Hispanics had similar Lp(a) (12.9 mg/dL and 13.1 mg/dL, respectively).21 In CARDIA, blacks had higher levels than whites.22 In the Dallas Heart Study, blacks had higher Lp(a) than whites, which were slightly higher than Hispanics.19 In the San Antonio Heart Study, whites also had slightly higher Lp(a) than Mexican Americans.23 These results are consistent with prior postulations that race/ethnicity differences may represent ancient migration patterns of populations out of Africa. Wherein, populations of African descent retain highest Lp(a) levels followed by South Asian, Caucasian, Hispanic, and East Asian.1 These difference may be a result of LPA gene variations.18
We also found that Lp(a) association with risk for events also differed by race/ethnicity. Lp(a) was associated with MI among non-Hispanic whites and Mexican Americans, but not blacks. Again, our findings are congruent with prior studies, although results have been variable. In meta-analysis (n = 34 studies), Lp(a) was associated with risk for CHD among whites but not blacks.8 Notably, in other smaller, but well-known cohorts, results have also varied. In MESA, Lp(a) > 50 mg/dL was associated with CHD in blacks, whites, and Hispanics.21 In ARIC, Lp(a) levels were associated with CHD and ischemic stroke in both African Americans and Caucasians.20 However, in REGARDS, Lp(a) was associated for stroke among blacks, but not whites.24 Future research should focus on understanding how and why race/ethnicity alters the association between Lp(a) and CVD. Risk threshold may be race or ethnicity specific, although debate remains as to the true threshold to consider for specific race/ethnicities.
There are several strengths for our study. First, the cohort is nationally representative and our analyses adjusted for complex sample design. Because few factors alter Lp(a) levels, it is unlikely that unmeasured factors are responsible for our observed association with clinical outcomes. Furthermore, the study size was large and we were able to look at associations across sex and race/ethnicity. Finally, the associations between Lp(a) and clinical outcomes remained even after the addition of multiple factors to the risk models. This gives credence to Lp(a) being a potent risk factor that is not accounted for by other factors.
Our study has limitations. First, this cohort was collected between 1991 and 1994. Estimations could differ at the present time because of shifts in population demographics. However, it is unlikely that associations of Lp(a) with clinical outcomes has changed over time. Rather, what could have change in the interim are the prevalence of risk factors (eg, obesity, diabetes, hypertension, and lipid levels) and how individuals report a history of MI or stroke. Second, beyond being directed at apo(a), the specific antigen to which the assay directed is not indicated in the study protocol. There was also no common reference material for the assay to be calibrated, although did use the manufacturer’s supplied standards. Modern assays that are isoform independent that have standardized calibrations may offer more accurate estimations of Lp(a) levels.1 Both of these limitations emphasize the need to test Lp(a) in future iterations of NHANES, which should be performed with an isoform independent assay to get an updated estimation of Lp(a) in the United States. Third, patient-report outcomes were the endpoint of this study, thus could have been under- or over-reported. Prior evidence has found variable accuracy of self-reported MI or stroke.35 In NHANES II and III, self-report of MI was found to correlate with Q-wave MI on ECG most strongly among men age 45–74 years and women age ≥55 years. In a sensitivity analysis, we found that among those aged ≥50 years, Lp(a) remained associated with MI. Further limitations include that the study was cross-sectional. Thus, clinical events preceded measurement of Lp(a) and other covariates, which could have been different at or before the clinical event. In addition, the cross-sectional nature by definition excludes fatal CVD events, thus the association between Lp(a) and cardiovascular events could indicate protection against fatal events. However, the plethora of evidence of Lp(a) being causative for CVD makes this exceedingly unlikely. Finally, the race/ethnicities reported in this study were limited. Recent iterations of NHANES provide more detailed race/ethnicity description thus could help to better differentiate race/ethnicity specific effects of Lp(a).
Conclusions
Our results suggest that about 1 in 7 individuals in the United States have an Lp(a) ≥50 mg/dL. Lp(a) was significantly associated with patient-reported history of nonfatal MI, but not stroke. Risk for nonfatal cardiovascular events associated with Lp(a) differed based on sex and race/ethnicity. Future iterations of the NHANES should consider retesting Lp(a) with an isoform independent assay to obtain an updated estimation of Lp(a) in the US population.
Acknowledgments
Authors’ contributions: EJB contributed to formulating the research question, designing the study, carrying it out, analyzing the data, interpreting the data, and writing and revising the article. AM contributed to designing the study, interpreting the data, and writing and revising the article. ESS contributed to designing the study, interpreting the data, and writing and revising the article. ND contributed to interpreting the data and writing and revising the article. KN contributed to formulating the research question, designing the study, interpreting the data, and writing and revising the article.
Appendix
Supplement Table 1.
Proportion of the population and subgroups at various Lp(a) thresholds (≥30, ≥50, ≥70, and ≥90 mg/dL)
| ≥ 30 mg/dL | ≥ 50 mg/dL | ≥ 70 mg/dL | ≥ 90 mg/dL | |||||
|---|---|---|---|---|---|---|---|---|
| Percentage (95% CI) | P-value | Percentage (95% CI) | P-value | Percentage (95% CI) | P-value | Percentage (95% CI) | P-value | |
| Entire population (n = 8722) | 27.8% (26.3%, 29.3%) | - | 14.7% (13.6%, 15.9%) | - | 5.6% (5.0%, 6.3%) | - | 2.6% (2.2%, 3.1%) | - |
| Age | - | .29 | - | .02 | - | .049 | - | .08 |
| <35 (n = 3091) | 27.1% (24.6%,29.7%) | - | 13.2% (11.4%, 15.1%) | - | 4.7% (3.7%, 5.9%) | - | 2.2% (1.6%, 3.2%) | - |
| 35–54 (n = 2548) | 27.5% (25.0%, 30.3%) | - | 14.5% (12.6%, 16.5%) | - | 5.6% (4.6%, 6.9%) | - | 2.4% (1.8%, 3.3%) | - |
| 55–74 (n = 2092) | 27.9% (25.3%, 30.8%) | - | 17.1% (15.0%, 19.4% | - | 7.0% (5.8%, 8.5%) | - | 3.7% (2.8%, 4.9%) | - |
| 75+ (n = 991) | 32.7% (29.0%, 36.7%) | - | 18.0% (15.1%, 21.3%) | - | 6.1% (5.0%, 6.3%) | - | 2.9% (1.9%, 4.3%) | - |
| Sex | - | .003 | - | .002 | - | .02 | - | .07 |
| Male (n = 3811) | 25.4% (23.2%, 27.7%) | - | 12.9% (11.3%, 14.6%) | - | 4.8% (3.9%, 5.8%) | - | 2.2% (1.6%, 2.9%) | - |
| Female (n = 4911) | 30.0% (28.0%, 32.0%) | - | 16.4% (15.0%, 17.9%) | - | 6.3% (5.5%, 7.3%) | - | 3.1% (2.5%, 3.8%) | - |
| Race/ethnicity | - | <.001 | - | <.001 | - | <.001 | - | <.001 |
| Non-Hispanic white(n = 3356) | 24.3% (22.5%, 26.2%) | - | 11.9% (10.6%, 13.3%) | - | 3.9% (3.2%, 4.8%) | - | 1.7% (1.3%, 2.4%) | - |
| Non-Hispanic black (n = 2585) | 60.0% (57.9%, 62.1%) | - | 39.0% (36.9%, 41.1%) | - | 19.7% (18.0%, 21.4%) | - | 9.9% (8.7%, 11.2%) | - |
| Mexican American (n = 2364) | 16.0% (14.4%, 17.8%) | - | 7.3% (6.2%, 8.6%) | - | 2.2% (1.6%, 2.9%) | - | 0.7% (0.4%, 1.1%) | - |
| Other (n = 417) | 23.7% (18.6%, 29.8%) | - | 12.6% (8.9%, 17.7%) | - | 4.0% (2.2%, 7.0%) | - | 2.2% (0.9%, 5.2%) | - |
| MI | - | .04 | - | <.001 | - | <.001 | - | <.001 |
| No (n = 8359) | 27.5% (26.0%, 29.1%) | - | 14.4% (13.3%, 15.5%) | - | 5.3% (4.7%, 6.0%) | - | 2.5% (2.1%, 2.9%) | - |
| Yes (n = 358) | 34.9% (27.9%, 42.7%) | - | 25.4% (189%, 33.1%) | - | 13.1% (8.2%, 20.3%) | - | 7.8% (4.0%, 14.7%) | - |
| Stroke | - | .95 | - | .09 | - | .02 | - | .02 |
| No (n = 8441) | 27.8% (26.3%, 29.3%) | - | 14.6% (13.5%, 15.8%) | - | 5.5% (4.9%, 6.2%) | - | 2.6% (2.1%, 3.1%) | - |
| Yes (n = 276) | 27.5% (20.8%,35.5%) | - | 19.7% (14.0%, 27.1%) | - | 10.5% (6.2%, 17.1%) | - | 6.2% (3.0%, 12.5%) | - |
CI, confidence interval; Lp(a), lipoprotein (a); MI, myocardial infarction.
Supplement Table 2.
Association of covariates with MI and stroke in univariate analyses
| Covariate | MI | Stroke | ||
|---|---|---|---|---|
| OR (95% CI) | P-value | OR (95% CI) | P-value | |
| Lp(a) (per SD increase) | 1.36 (1.14, 1.62) | .001 | 1.19 (0.98, 1.45) | .09 |
| Demographics | - | - | - | - |
| Age (per year) | 1.07 (1.06, 1.08) | <.001 | 1.07 (1.06, 1.08) | <.001 |
| Female | 0.50 (0.37, 0.69) | <.001 | 1.07 (0.74, 1.56) | .72 |
| Race/ethnicity | - | <.001 | - | <.001 |
| Non-Hispanic white | Reference | - | Reference | <.001 |
| Non-Hispanic black | 0.73 (0.54, 0.99) | - | 1.21 (0.87, 1.69) | - |
| Mexican American | 0.36 (0.25, 0.52) | - | 0.45 (0.30, 0.67) | - |
| Other | 0.80 (0.35, 1.83) | - | 1.15 (0.50, 2.61) | - |
| Marital status | - | <.001 | - | <.001 |
| Never married | Reference | - | Reference | - |
| Divorced/separated | 5.37 (2.22, 13.00) | - | 5.57 (2.14, 14.53) | - |
| Widowed | 14.15 (6.16, 32.48) | - | 18.3 (7.71, 43.38) | - |
| Married/living with partner | 4.79 (2.13, 10.76) | - | 3.22 (1.36, 7.61) | - |
| Education (per additional year) | 0.88 (0.85, 0.91) | <.001 | 0.85 (0.82, 0.88) | <.001 |
| Income <$20,000 | 2.22 (1.61, 3.07) | <.001 | 3.13 (2.10, 4.69) | <.001 |
| Clinical risk factors and biomarkers | - | - | - | - |
| Ever smoker | 3.05 (2.21, 4.20) | <.001 | 1.62 (1.13, 2.33) | .01 |
| Diabetes | 4.66 (3.14, 6.91) | <.001 | 7.08 (4.66, 10.75) | <.001 |
| Hypertension | 4.54 (3.27, 6.30) | <.001 | 4.69 (3.17, 6.93) | <.001 |
| Relative with MI before age 50 | 1.58 (1.04, 2.39) | .03 | 1.12 (0.69, 1.81) | .64 |
| Non-HDL-C (per SD increase) | 1.69 (1.51, 1.91) | <.001 | 1.59 (1.37, 1.86) | <.001 |
| C-reactive protein (per 1 mg/dL) | 1.53 (1.35, 1.73) | <.001 | 1.56 (1.34, 1.82) | <.001 |
| eGFR (mL/min/1.73 m2) | 0.94 (0.93, 0.95) | <.001 | 0.95 (0.93, 0.96) | <.001 |
eGFR, estimated glomerular filtration rate; HDL, high-density lipoprotein; Lp(a), lipoprotein(a); MI, myocardial infarction.
Supplement Table 3.
Association of covariates with MI and stroke in multivariate analyses with polynomials and interactions included, shown as average marginal effects
| Covariate | MI | Stroke | ||
|---|---|---|---|---|
| dy/dx (standard error) | P-value | dy/dx (standard error) | P-value | |
| Lp(a) (per SD increase) | 0.0092 (0.0030) | .002 | 0.0025 (0.0022) | .26 |
| Demographics | - | - | - | - |
| Age (per year) | 0.0014 (0.0002) | <.001 | 0.0009 (0.0001) | <.001 |
| Female | −0.0258 (0.0055) | <.001 | −0.0031 (0.0041) | .46 |
| Race/ethnicity | - | .18 | - | .17 |
| Non-Hispanic white | Reference | - | Reference | - |
| Non-Hispanic black | −0.0085 (0.0059) | - | − 0.0009 (0.0044) | - |
| Mexican American | −0.0093 (0.0052) | - | −0.0066 (0.0036) | - |
| Other | 0.0067 (0.0175) | - | 0.0142 (0.0126) | - |
| Marital status | - | - | - | - |
| Never married | - | - | - | - |
| Divorced/separated | - | - | - | - |
| Widowed | - | - | - | - |
| Married/living with partner | - | - | - | - |
| Education (per additional year) | - | - | - | - |
| Income <$20,000 | 0.0148 (0.0055) | .008 | 0.0146 (0.0047) | .002 |
| Clinical risk factors and biomarkers | - | - | - | - |
| Ever smoker | 0.0120 (0.0046) | <.001 | 0.0071 (0.0043) | .10 |
| Diabetes | 0.0192 (0.0084) | .023 | 0.0321 (0.0094) | .001 |
| Hypertension | 0.0162 (0.0055) | .003 | 0.0119 (0.0040) | .003 |
| Relative with MI before age 50 | 0.0298 (0.0100) | .003 | - | - |
| Non-HDL-C (per SD increase) | 0.0083 (0.0022) | <.001 | 0.0035 (0.0018) | .06 |
| C-reactive protein (per 1 mg/dL) | 0.0065 (0.0015) | <.001 | 0.0034 (0.0013) | .01 |
| eGFR (mL/min/1.73 m2) | −0.0003 (0.0002) | .08 | - | - |
eGFR, estimated glomerular filtration rate; HDL, high-density lipoprotein; Lp(a), lipoprotein(a); MI, myocardial infarction.
Supplement Table 4.
Association of Lp(a) with MI and stroke in multivariate analyses with polynomials and interactions included for the entire population and by sex or race/ethnicity at various dichotomous cut-points
| Covariate | MI | Stroke | ||
|---|---|---|---|---|
| OR (95% CI) | P-value | OR (95% CI) | P-value | |
| Entire population | - | - | - | - |
| ≥30 mg/dL | 1.63 (1.09, 2.45) | .02 | 0.89 (0.58, 1.37) | .61 |
| ≥50 mg/dL | 2.22 (1.36, 3.61) | .001 | 1.26 (0.77, 2.06) | .37 |
| ≥70 mg/dL | 3.06 (1.51, 6.21) | .002 | 1.80 (0.88, 3.70) | .11 |
| ≥90 mg/dL | 3.04 (1.20, 7.69) | .02 | 2.12 (0.83, 5.39) | .11 |
| Male | - | - | - | - |
| ≥30 mg/dL | 1.75 (1.00, 3.09) | .052 | 0.84 (0.46, 1.54) | .57 |
| ≥50 mg/dL | 2.96 (1.47, 5.95) | .002 | 1.01 (0.50, 2.06) | .98 |
| ≥70 mg/dL | 4.27 (1.58, 11.53) | .004 | 1.18 (0.36, 3.85) | .78 |
| ≥90 mg/dL | 4.03 (1.06, 15.3) | .04 | 1.03 (0.46, 2.30) | .95 |
| Female | - | - | - | - |
| ≥30 mg/dL | 1.42 (0.87, 2.33) | .16 | 0.93 (0.49, 1.75) | .81 |
| ≥50 mg/dL | 1.46 (0.84, 2.54) | .18 | 1.42 (0.70, 2.87) | .33 |
| ≥70 mg/dL | 1.81 (0.80, 4.10) | .16 | 2.36 (0.93, 6.03) | .07 |
| ≥90 mg/dL | 1.95 (0.63, 6.07) | .25 | 3.14 (0.89, 11.0) | .08 |
| Non-Hispanic white | - | - | - | - |
| ≥30 mg/dL | 1.94 (1.22, 3.08) | .005 | 0.98 (0.57, 1.71) | .96 |
| ≥50 mg/dL | 2.95 (1.68, 5.21) | <.001 | 1.34 (0.68, 2.66) | .39 |
| ≥70 mg/dL | 4.99 (2.17, 11.49) | <.001 | 2.63 (0.99, 6.98) | .052 |
| ≥90 mg/dL | 4.93 (1.60, 15.23) | .006 | 3.29 (0.93, 11.64) | .06 |
| Non-Hispanic black | - | - | - | - |
| ≥30 mg/dL | 0.70 (0.41, 1.22) | .21 | 1.09 (0.57, 2.11) | .79 |
| ≥50 mg/dL | 0.65 (0.37, 1.14) | .13 | 1.38 (0.76, 2.54) | .29 |
| ≥70 mg/dL | 0.84 (0.44, 1.60) | .60 | 1.33 (0.65, 2.73) | .44 |
| ≥90 mg/dL | 0.88 (0.34, 1.77) | .54 | 0.93 (0.42, 2.22) | .93 |
| Mexican American | - | - | - | - |
| ≥30 mg/dL | 3.59 (1.65, 7.82) | .001 | 1.10 (0.42, 2.88) | .85 |
| ≥50 mg/dL | 8.04 (3.52, 18.36) | <.001 | 1.36(0.31, 5.88) | .68 |
| ≥70 mg/dL | 3.25 (0.63, 16.74) | .16 | 0.37 (0.05, 2.59) | .31 |
| ≥90 mg/dL | 8.09 (0.77, 85.10) | .08 | 0.79 (0.11, 5.57) | .82 |
CI, confidence interval; Lp(a), lipoprotein (a); MI, myocardial infarction; OR, odds ratio
Footnotes
Conflicts of interest: Dr Desai reported receiving grants and personal fees from Amgen, Boehringer Ingelheim, and Relypsa; receiving personal fees from Cytokinetics, Novartis, and scPharmaceuticals; having a contract with the Centers for Medicare & Medicaid Services; and receiving funding from Johnson and Johnson and Medtronic outside the submitted work. Dr. Spatz receives support from the Centers for Medicare & Medicaid Services to develop performance measures used in public reporting programs, the Food and Drug Administration to support projects within the Yale-Mayo Clinic Center of Excellence in Regulatory Science and Innovation (CERSI), the National Institute on Minority Health and Health Disparities (U54MD010711–01) to study precision-based approaches to hypertension, and from the National Institute of Biomedical Imaging and Bioengineering (R01 EB028106–01) to study a cuff-less blood pressure device. Dr. Mani receives is supported by grants from the National Institutes of Health (NIH) (RHL135767A). The remaining authors have nothing to disclose.
Financial disclosures
This publication was made possible by CTSA Grant Number TL1 TR001864 from the National Center for Advancing Translational Science (NCATS), a component of the National Institutes of Health (NIH). Its contents are solely the responsibility of the authors and do not necessarily represent the official view of the NIH.
References
- 1.Tsimikas S A Test in Context: Lipoprotein(a): Diagnosis, Prognosis, Controversies, and Emerging Therapies. J Am Coll Cardiol 2017; 69:692–711. [DOI] [PubMed] [Google Scholar]
- 2.Nordestgaard BG, Chapman MJ, Ray K, et al. Lipoprotein(a) as a cardiovascular risk factor: current status. Eur Heart J 2010;31: 2844–2853. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kronenberg F, Utermann G. Lipoprotein(a): resurrected by genetics. J Intern Med 2013;273:6–30. [DOI] [PubMed] [Google Scholar]
- 4.Gencer B, Kronenberg F, Stroes ES, Mach F. Lipoprotein(a): the revenant. Eur Heart J 2017;38:1553–1560. [DOI] [PubMed] [Google Scholar]
- 5.Clarke R, Peden JF, Hopewell JC, et al. Genetic variants associated with Lp(a) lipoprotein level and coronary disease. N Engl J Med 2009;361:2518–2528. [DOI] [PubMed] [Google Scholar]
- 6.Kamstrup PR, Tybjaerg-Hansen A, Steffensen R, Nordestgaard BG. Genetically elevated lipoprotein(a) and increased risk of myocardial infarction. JAMA 2009;301:2331–2339. [DOI] [PubMed] [Google Scholar]
- 7.Emdin CA, Khera AV, Natarajan P, et al. Phenotypic Characterization of Genetically Lowered Human Lipoprotein(a) Levels. J Am Coll Cardiol 2016;68:2761–2772. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Erqou S, Kaptoge S, Perry PL, et al. Lipoprotein(a) concentration and the risk of coronary heart disease, stroke, and nonvascular mortality. JAMA 2009;302:412–423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Nave AH, Lange KS, Leonards CO, et al. Lipoprotein (a) as a risk factor for ischemic stroke: a meta-analysis. Atherosclerosis 2015;242: 496–503. [DOI] [PubMed] [Google Scholar]
- 10.Arnett DK, Blumenthal RS, Albert MA, et al. 2019 ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. J Am Coll Cardiol 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Grundy SM, Stone NJ, Bailey AL, et al. 2018 AHA/ACC/AACV-PR/AAPA/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA Guideline on the Management of Blood Cholesterol: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. J Am Coll Cardiol 2018. [Google Scholar]
- 12.Varvel S, McConnell JP, Tsimikas S. Prevalence of Elevated Lp(a) Mass Levels and Patient Thresholds in 532 359 Patients in the United States. Arterioscler Thromb Vasc Biol 2016;36:2239–2245. [DOI] [PubMed] [Google Scholar]
- 13.Sharrett AR, Ballantyne CM, Coady SA, et al. Coronary heart disease prediction from lipoprotein cholesterol levels, triglycerides, lipoprotein(a), apolipoproteins A-I and B, and HDL density subfractions: The Atherosclerosis Risk in Communities (ARIC) Study. Circulation 2001;104:1108–1113. [DOI] [PubMed] [Google Scholar]
- 14.Steffen BT, Thanassoulis G, Duprez D, et al. Race-Based Differences in Lipoprotein(a)-Associated Risk of Carotid Atherosclerosis. Arterioscler Thromb Vasc Biol 2019;39:523–529. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Obisesan TO, Aliyu MH, Adediran AS, Bond V, Maxwell CJ, Rotimi CN. Correlates of serum lipoprotein (A) in children and adolescents in the United States. The Third National Health Nutrition and Examination Survey (NHANES-III). Lipids Health Dis 2004;3: 29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kovesdy CP, Astor BC, Longenecker JC, Coresh J. Association of kidney function with serum lipoprotein(a) level: The Third National Health and Nutrition Examination Survey (1991–1994). Am J Kidney Dis. 2002;40:899–908. [DOI] [PubMed] [Google Scholar]
- 17.Dumitrescu L, Glenn K, Brown-Gentry K, et al. Variation in LPA is associated with Lp(a) levels in three populations from the Third National Health and Nutrition Examination Survey. PLoS One 2011;6: e16604. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Enkhmaa B, Anuurad E, Berglund L. Lipoprotein (a): impact by ethnicity and environmental and medical conditions. J Lipid Res 2016;57:1111–1125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tsimikas S, Clopton P, Brilakis ES, et al. Relationship of oxidized phospholipids on apolipoprotein B-100 particles to race/ethnicity, apolipoprotein(a) isoform size, and cardiovascular risk factors: results from the Dallas Heart Study. Circulation 2009;119:1711–1719. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Virani SS, Brautbar A, Davis BC, et al. Associations between lipoprotein(a) levels and cardiovascular outcomes in black and white subjects: the Atherosclerosis Risk in Communities (ARIC) Study. Circulation 2012;125:241–249. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Guan W, Cao J, Steffen BT, et al. Race is a key variable in assigning lipoprotein(a) cutoff values for coronary heart disease risk assessment: the Multi-Ethnic Study of Atherosclerosis. Arterioscler Thromb Vasc Biol 2015;35:996–1001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Marcovina SM, Albers JJ, Wijsman E, Zhang Z, Chapman NH, Kennedy H. Differences in Lp[a] concentrations and apo[a] polymorphs between black and white Americans. J Lipid Res 1996;37: 2569–2585. [PubMed] [Google Scholar]
- 23.Haffner SM, Gruber KK, Morales PA, et al. Lipoprotein(a) Concentrations in Mexican Americans and Non-Hispanic Whites: The San Antonio Heart Study. Am J Epidemiol 1992;136:1060–1068. [DOI] [PubMed] [Google Scholar]
- 24.Arora P, Kalra R, Callas PW, et al. Lipoprotein(a) and Risk of Ischemic Stroke in the REGARDS Study. Arterioscler Thromb Vasc Biol 2019;39:810–818. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Plan and operation of the Third National Health and Nutrition Examination Survey, 1988–94. Series 1: programs and collection procedures. Vital Health Stat 1994;1:1–407. [PubMed] [Google Scholar]
- 26.Gunter E, Lewis B, Koncikowski S. Laboratory Procedures Used for the Third National Health and Nutrition Examination Survey (NHANES III), 1988–1994. U.S. Department of Health and Human Services; 1996. [Google Scholar]
- 27.Catapano AL, Graham I, De Backer G, et al. 2016 ESC/EAS Guidelines for the Management of Dyslipidaemias. Eur Heart J 2016;37: 2999–3058. [DOI] [PubMed] [Google Scholar]
- 28.Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972;18:499–502. [PubMed] [Google Scholar]
- 29.Martin SS, Blaha MJ, Elshazly MB, et al. Comparison of a Novel Method vs the Friedewald Equation for Estimating Low-Density Lipoprotein Cholesterol Levels From the Standard Lipid Profile. JAMA. 2013;310:2061–2068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Tsimikas S, Karwatowska-Prokopczuk E, Gouni-Berthold I, et al. Lipoprotein(a) Reduction in Persons with Cardiovascular Disease. N Engl J Med 2020;382:244–255. [DOI] [PubMed] [Google Scholar]
- 31.Pare G, Caku A, McQueen M, et al. Lipoprotein(a) Levels and the Risk of Myocardial Infarction Among 7 Ethnic Groups. Circulation 2019;139:1472–1482. [DOI] [PubMed] [Google Scholar]
- 32.Cook NR, Mora S, Ridker PM. Lipoprotein(a) and Cardiovascular Risk Prediction Among Women. J Am Coll Cardiol 2018;72:287. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Kouvari M, Panagiotakos DB, Chrysohoou C, et al. Lipoprotein (a) and 10-year Cardiovascular Disease Incidence in Apparently Healthy Individuals: A Sex-based Sensitivity Analysis from ATTICA Cohort Study. Angiology 2019;70:819–829. [DOI] [PubMed] [Google Scholar]
- 34.Bennet A, Di Angelantonio E, Erqou S, et al. Lipoprotein(a) Levels and Risk of Future Coronary Heart Disease: Large-Scale Prospective Data. JAMA Intern Med 2008;168:598–608. [DOI] [PubMed] [Google Scholar]
- 35.Moran A, Shen A, Turner-Lloveras D, et al. Utility of self-reported diagnosis and electrocardiogram Q-waves for estimating myocardial infarction prevalence: an international comparison study. Heart 2012;98:1660–1666. [DOI] [PubMed] [Google Scholar]