Abstract
BACKGROUND
Lipoprotein(a) [Lp(a)] is associated with an increased risk of atherosclerotic cardiovascular disease (ASCVD). However, whether the optimal Lp(a) threshold for risk assessment should differ based on baseline ASCVD status is unknown.
OBJECTIVES
The purpose of this study was to assess the association between Lp(a) and major adverse cardiovascular events (MACE) among patients with and without baseline ASCVD.
METHODS
We studied a retrospective cohort of patients with Lp(a) measured at 2 medical centers in Boston, Massachusetts, from 2000 to 2019. To assess the association of Lp(a) with incident MACE (nonfatal myocardial infarction [MI], nonfatal stroke, coronary revascularization, or cardiovascular mortality), Lp(a) percentile groups were generated with the reference group set at the first to 50th Lp(a) percentiles. Cox proportional hazards modeling was used to assess the association of Lp(a) percentile group with MACE.
RESULTS
Overall, 16,419 individuals were analyzed with a median follow-up of 11.9 years. Among the 10,181 (62%) patients with baseline ASCVD, individuals in the 71st to 90th percentile group had a 21% increased hazard of MACE (adjusted HR: 1.21; P < 0.001), which was similar to that of individuals in the 91st to 100th group (adjusted HR: 1.26; P < 0.001). Among the 6,238 individuals without established ASCVD, there was a continuously higher hazard of MACE with increasing Lp(a), and individuals in the 91st to 100th Lp(a) percentile group had the highest relative risk with an adjusted HR of 1.93 (P < 0.001).
CONCLUSIONS
In a large, contemporary U.S. cohort, elevated Lp(a) is independently associated with long-term MACE among individuals with and without baseline ASCVD. Our results suggest that the threshold for risk assessment may be different in primary vs secondary prevention cohorts.
Keywords: lipoprotein(a) cohort, lipoprotein(a) therapies
Over the last 30 years, lipoprotein(a) [Lp(a)] has emerged as an independent risk factor for incident atherosclerotic cardiovascular disease (ASCVD).1–12 Given its genetic underpinnings, individuals with elevated levels are exposed to this atherogenic particle throughout their lifetimes.13–15 Lp(a) levels >125 nmol/L (w50 mg/dL) correspond to the 80th percentile in population-based studies and has been used in U.S. guidelines to identify higher risk individuals.10,16–18 In addition to its association with incident MI, Lp(a) is an independent risk factor for ischemic stroke and calcific aortic valve disease.4,5,8,19
Until recently, Lp(a) has been considered an unmodifiable cardiovascular risk factor. However, with the advent of small interfering RNAs and antisense oligonucleotides, the landscape of novel therapeutics is showing significant promise.20,21 Currently, there are 2 candidate agents in phase 3 clinical trials, both of which significantly decrease circulating Lp(a) levels while achieving excellent safety and tolerability.20,22 Moreover, proprotein convertase subtilisin/kexin 9 inhibitors have been observed to decrease circulating Lp(a) levels by approximately 20%. Key secondary analyses of the FOURIER (Further Cardiovascular Outcomes Research With PCSK9 Inhibition in Patients With Elevated Risk) and ODYSSEY (ODYSSEY Outcomes: Evaluation of Cardiovascular Outcomes After an Acute Coronary Syndrome During Treatment With Alirocumab) trials demonstrated that individuals with baseline elevated Lp(a) levels in addition to elevated low-density lipoprotein cholesterol (LDL-C) levels received the greatest cardiovascular benefit from proprotein convertase subtilisin/kexin 9 inhibitors.23–25
Although it is clear that Lp(a) is independently associated with ASCVD, the precise population of patients most affected by elevated levels who may benefit from targeted therapeutic interventions is still not fully defined. Furthermore, although ongoing trials are focusing on secondary prevention patients, it is unclear what threshold of elevated Lp(a) may be most useful for identifying higher-risk primary prevention cohorts for potential targeted intervention. Accordingly, we sought to evaluate the association of Lp(a) with major adverse cardiovascular events (MACE) among subjects with and without baseline ASCVD. Additionally, we sought to investigate whether increasing Lp(a) level was associated with sequentially higher ASCVD risk or whether there was a plateau in long-term risk after a certain threshold level.
METHODS
STUDY POPULATION.
The design of the Mass General Brigham Lp(a) Registry has been previously described.26 Briefly, this is a retrospective cohort study from 2 large academic medical centers in Boston, Massachusetts (Brigham and Women’s Hospital and Massachusetts General Hospital), that included all patients who underwent Lp(a) testing as part of routine care from January 1, 2000, to July 1, 2019. The Mass General Brigham Lp(a) Registry was approved by the Institutional Review Board at Mass General Brigham.
All individuals ≥18 years of age with at least 1 Lp(a) result were screened for cohort inclusion. Individuals were excluded from the cohort if they met any of the following criteria: 1) a diagnosis of severe kidney dysfunction defined as stage 5 chronic kidney disease (estimated glomerular filtration rate <15 mL/min/m2), those who have had a kidney transplant, or those on dialysis during the covariate assessment period; and 2) the presence of a diagnostic International Classification of Diseases (ICD) code for a malignant neoplasm during the covariate assessment window, excluding nonmelanoma skin cancers. Finally, for the present study, individuals who died during the covariate assessment window were excluded from the analysis.
LP(A) ASSAYS.
The present study includes 21,410 patients who underwent testing with either the Lp(a)-particle assay (measured in nmol/L) or the Lp(a)-mass assay (measured in mg/dL). All Lp(a) laboratory testing was performed at commercial laboratories over the study period using industry-standard assays. To avoid potential biases caused by possible differences in Lp(a) testing techniques over the study period, percentile distributions for each assay were defined separately. Subsequently, Lp(a) percentiles were then combined across assay types. This methodology has been used previously in other large Lp(a) studies.2,5,9,27,28 For the primary analytic approach, we analyzed all outcomes based on the Lp(a) percentile groups. Given the well-established distribution of Lp(a), a priori percentile groups were determined and used in the present study: 1st to 50th, 51st to 70th, 71st to 90th, and 91st to 100th. Once assays were combined across percentiles, we subsequently converted all Lp(a) values to nmol/L using the following conversion formula to best represent the data in a clinically relevant manner: Lp(a) nmol = (2.18 × Lp(a)-M) − 3.83.27,29 For sensitivity analyses and the restricted cubic spline analysis (as described in the following text), we maintained the raw Lp(a) levels and analyzed using Lp(a) values in the form of nmol/L. See Figure 1 for the distribution of Lp(a) values across the cohort as well as Supplemental Appendix Section I, which demonstrates the relationship between Lp(a) percentile group and Lp(a) values. For patients with multiple Lp(a) levels over the study period, the patient was assigned to their median Lp(a) level and its corresponding percentile.
FIGURE 1.
Lp(a) Distribution Within the Cohort
Among the total cohort of 16,419 individuals with an eligible lipoprotein(a) [Lp(a)] test, the median Lp(a) is 35 nmol/L (IQR: 14–113 nmol/L). The Lp(a) percentile groups and corresponding Lp(a) levels are specified in the figure.
RISK FACTORS AND BASELINE COVARIATES.
The RPDR (Research Patient Data Registry)30 at Mass General Brigham served as the primary source of data for this registry. The RPDR is a centralized clinical data registry that consolidates clinical information from all hospitals within the Mass General Brigham network, including Brigham and Women’s Hospital and Massachusetts General Hospital. This system provides demographic data, laboratory and imaging data, diagnostic and procedural codes, medication data, vital status, and clinical documentation for individuals meeting specified search criteria. The RPDR obtains vital status from the Social Security Administration Death Master File.
For each individual, the presence of cardiovascular (CV) risk factors was ascertained through the use of validated natural language processing (NLP) modules,31 laboratory data, as well as diagnostic and procedural ICD-9, ICD-10, and Current Procedural Terminology codes as previously described.26 Additional details regarding risk factor definitions, ascertainment, and the relative contribution of NLP to risk factor identification are found in the Supplemental Appendix Section II. Given the extensive and varied sources of baseline clinical information, patients who did not have a positive indicator for a given covariate were assumed to not have that baseline risk factor. Therefore, all individuals were considered to have complete covariate data and there were no individuals excluded from the multivariable adjusted models.
The baseline covariate assessment period was defined as 12 months before and 30 days after the date of the Lp(a) test. For individuals with multiple Lp(a) tests over time, the baseline covariate assessment period was assigned to the first chronologic Lp(a) measurement. Accordingly, any clinical covariate (as assessed by NLP, billing codes, or laboratory data) that occurred within this timeframe was attributed to baseline characteristics.
For the present study, a history of ASCVD was defined as a prior history of MI, coronary revascularization, or ischemic (nonhemorrhagic) stroke as assessed during the covariate assessment window using a combination of NLP and ICD diagnostic and procedural codes. Further details regarding ascertainment of ASCVD-related baseline data can be found in the Supplemental Appendix Section III.
OUTCOMES OF INTEREST.
The primary outcome for this analysis was a composite of major adverse cardiovascular events (MACE) which was defined as follows: 1) nonfatal MI; 2) ischemic (nonhemorrhagic) stroke; 3) coronary revascularization; or 4) cardiovascular mortality. As previously described,26 acute MI and acute ischemic (nonhemorrhagic) stroke were defined by the presence of a diagnostic ICD code in the primary hospital discharge position. Coronary revascularization was defined by the presence of a procedural ICD code in the medical record. This methodology has been extensively validated with high specificity, high positive predictive value, and reasonable sensitivity.32–38
To ascertain cause of death, we queried the National Death Index (NDI) and the Massachusetts Office of Vital Statistics for all patients who were recorded as deceased by the Social Security Administration Death Master File. ICD-10 codes were characterized to determine the underlying and proximal causes of death for each patient who expired during the study follow-up period. Deaths were coded as cardiovascular and noncardiovascular. Cardiovascular mortality was determined based on the ICD-coded underlying causes of death39–42 as determined by the NDI or the Massachusetts Office of Vital Statistics. Cause of death was assessed blind to Lp(a) level or any other clinical factor. The ICD codes used for the study’s outcome measures can be found in the Supplemental Appendix Section IV.
STATISTICAL ANALYSIS.
Categorical variables were reported as frequencies and proportions and compared with chi-square or Fisher exact tests, as appropriate. Continuous variables were reported as mean ± SD or median (IQR) and compared with Student’s t-tests or Mann-Whitney U (Wilcoxon) tests, as appropriate. Patients were censored on the date of querying their source of vital statistics. In the analyses of cardiovascular mortality, individuals who died due to noncardiovascular or undetermined causes were conservatively labeled as not experiencing cardiovascular mortality and censored at their respective dates of death.
The proportional hazards assumptions were assessed by analyzing the Schoenfeld residuals. Kaplan-Meier survival curves were compared using the log-rank tests. Cox proportional hazards regressions were used to assess the association of Lp(a) percentile group with MACE and obtain corresponding HRs and 95% CIs. Univariable and multivariable Cox regressions were performed separately for patients with and without a baseline history of ASCVD.
As part of a secondary analytic approach, cubic spline models were developed to assess a potential nonlinear relationship between continuous Lp(a) levels and the primary composite outcome in groups with and without baseline ASCVD. These models were constructed using 3 knots that corresponded to the percentile group cutpoints used throughout this analysis: 41 nmol/L (50th percentile—reference), 111 nmol/L (70th percentile), and 215 nmol/L (90th percentile).
Multivariable models were checked for multi-collinearity using Spearman rank correlations and adjusted for the following covariates: age, sex, self-reported race/ethnicity, hypertension, chronic kidney disease status, non-Lp(a) hyperlipidemia, diabetes, insulin use (in diabetic individuals), and smoking status. In cases where there were missing data on baseline covariates, such individuals were omitted from all relevant multivariable analyses. All analyses were performed using Stata MP version 17 (StataCorp) and RStudio version 2022.12.0 ggplot2 package version 3.4.1.
RESULTS
A total of 21,410 patients underwent Lp(a) testing with a mass- or particle-based assay during the study period. After excluding patients with severe renal disease, those with a malignant neoplasm, and individuals who died during the covariate assessment window, 16,419 subjects (41% women; median age 60 years; IQR: 49–72 years) were eligible for analysis.
Among the 16,419 subjects, 10,181 (62%) had a history of ASCVD. Those with a history of ASCVD were more likely to be older and have a greater burden of traditional modifiable cardiovascular risk factors. See Table 1 for detailed characteristics of subjects with and without baseline ASCVD. Subjects with a history of ASCVD had higher median Lp(a) levels than those without ASCVD (37.8 nmol/L vs 31.1 nmol/L; P < 0.001). See Table 2 for more details about the distribution of Lp(a) stratified by ASCVD status.
TABLE 1.
Baseline Characteristics Stratified by ASCVD
| All Patients (N = 16,419) | ASCVD (n = 10,181, 62.0%) | No ASCVD (n = 6,238, 38.0%) | P Value | |
|---|---|---|---|---|
| Risk factors | ||||
| Hypertension | 9,226 (56.2) | 7,172 (70.4) | 2,054 (32.9) | <0.001 |
| Hyperlipidemia | 10,731 (65.4) | 7,701 (75.6) | 3,030 (48.6) | <0.001 |
| Diabetes mellitus | 3,830 (23.3) | 3,097 (30.4) | 733 (11.8) | <0.001 |
| Coronary artery disease | 5,935 (36.2) | 5,935 (58.3) | 0 (0.0) | <0.001 |
| Myocardial infarction | 5,126 (86.4) | 5,126 (86.4) | 0 (0.0) | |
| Coronary revascularization | 4,129 (69.6) | 4,129 (69.6) | 0 (0.0) | |
| Stroke | 6,216 (37.9) | 6,216 (61.1) | 0 (0.0) | <0.001 |
| Atrial fibrillation | 2,373 (14.5) | 2,010 (19.7) | 363 (5.8) | <0.001 |
| Heart failure | 1,220 (7.4) | 1,113 (10.9) | 107 (1.7) | <0.001 |
| Chronic kidney disease | 907 (5.5) | 753 (7.4) | 154 (2.5) | <0.001 |
| Current smoker | 4,514 (27.5) | 3,125 (30.7) | 1,389 (22.3) | <0.001 |
| Former smoker | 5,012 (30.5) | 3,442 (33.8) | 1,570 (25.2) | <0.001 |
| Demographics | ||||
| Age at time of Lp(a) test, y | 60 (49–72) | 64 (53–75) | 54 (43–64) | <0.001 |
| Female | 6,670 (40.6) | 3,868 (38.0) | 2,802 (44.9) | <0.001 |
| Race | <0.001 | |||
| White | 13,598 (82.8) | 8,302 (81.5) | 5,296 (84.9) | |
| Black | 736 (4.5) | 553 (5.4) | 183 (2.9) | |
| Hispanic | 474 (2.9) | 318 (3.1) | 156 (2.5) | |
| Asian | 412 (2.5) | 241 (2.4) | 171 (2.7) | |
| Othera | 1,199 (7.3) | 767 (7.5) | 432 (6.9) | |
| Laboratory values | ||||
| Triglycerides, mg/dL | 115 (81–166); 14,287 | 116 (83–166); 9,299 | 112 (77–165); 4,988 | <0.001 |
| Total cholesterol, mg/dL | 173 (145–203.3); 14,368 | 163 (138–192); 9,321 | 190 (163–221); 5,047 | <0.001 |
| HDL cholesterol, mg/dL | 46 (37–56.5); 14,362 | 43 (36–53); 9,327 | 51 (41–63); 5,035 | <0.001 |
| LDL cholesterol, mg/dL | 97 (74–123.5); 14,041 | 90 (69–115); 9,145 | 109 (86–137); 4,896 | <0.001 |
| Creatinine level, mg/dL | 1.0 (0.8–1.1); 14,599 | 1.0 (0.8–1.2); 9,864 | 1.0 (0.8–1.1); 4,735 | <0.001 |
| Median Lp(a), nmol/L | 35.4 (13.6–113.0) | 37.6 (13.6–117.0) | 31.1 (11.4–107.4) | <0.001 |
| Medical therapy | ||||
| Statin | 9,854 (60.0) | 7,661 (75.3) | 2,193 (35.2) | <0.001 |
| Nonstatin agent | 1,573 (9.6) | 1,278 (12.6) | 295 (4.7) | <0.001 |
| Insulin use | 1,562 (9.5) | 1,364 (13.4) | 198 (3.2) | <0.001 |
| Noninsulin antidiabetic | 1,858 (11.3) | 1,541 (15.1) | 317 (5.1) | <0.001 |
Values are n (%), median (IQR), or median (IQR); n.
Other includes Indian, Middle Eastern, Native American, other, Pacific Islander, and unknown.
ASCVD = atherosclerotic cardiovascular disease; HDL = high-density lipoprotein cholesterol; LDL = low-density lipoprotein cholesterol.
TABLE 2.
Lp(a) Distribution Stratified by ASCVD Status
| Lp(a) Percentiles | ASCVD (n = 10,181, 62.0%) | No ASCVD (n = 6,238, 38.0%) | P Value |
|---|---|---|---|
| 1st-50th | 5,423 (53.3) | 3,618 (58.0) | <0.001 |
| 51st-70th | 2,086 (20.5) | 1,119 (17.9) | |
| 71st-90th | 1,827 (18.0) | 1,065 (17.1) | |
| 91st-100th | 845 (8.3) | 436 (7.0) |
Values are n (%).
ASCVD = atherosclerotic cardiovascular disease; Lp(a) = lipoprotein(a).
When considering the entire population, over a median follow-up of 11.9 years (IQR: 6.2–14.4 years), 1,067 (6.5%) individuals experienced a nonfatal MI, 1,373 (8.4%) individuals experienced a nonfatal ischemic stroke, 1,362 (8.3%) individuals underwent coronary revascularization, and 2,416 (14.7%) individuals died of cardiovascular causes.
INDIVIDUALS WITH A HISTORY OF ASCVD.
Among the 10,181 individuals with a history of ASCVD, 1,886 (34.8%) in the 1st to 50th Lp(a) percentile, 816 (39.1%) in the 51st to 70th Lp(a) percentile, 749 (41.0%) in the 71st to 90th Lp(a) percentile, and 351 (41.5%) in the 91st to 100th Lp(a) percentile experienced the primary composite outcome of MI, coronary revascularization, ischemic stroke, or cardiovascular death (P < 0.001) (Figure 2, Table 3). In the unadjusted model, individuals in the highest Lp(a) percentile group had a 27% increased hazard of MACE compared with the reference group (P < 0.001). After adjusting for available covariates, these findings remained similar, with individuals in the highest Lp(a) percentile group having a 26% increased hazard of MACE compared with the reference group (P < 0.001) (Table 3). Annual event rates (determined based on the incidence rate per 1,000 patient years) for the primary composite outcome among patients with a history of ASCVD are presented in Figure 3. Notably, among individuals with baseline ASCVD, patients with Lp(a) levels in the 71st to 90th percentile group experienced a similarly increased hazard of MACE (adjusted HR: 1.21; P < 0.001) and annual event rates (5.26%) as did individuals in the 91st to 100th percentile group (adjusted HR: 1.26; P < 0.001 and annual event rates of 5.34%).
FIGURE 2.
Association of Lp(a) Percentile Group With the Primary Composite Outcome
Kaplan-Meier curves demonstrating the association of lipoprotein(a) [Lp(a)] percentile group with the primary composite outcome (myocardial infarction, ischemic stroke, coronary revascularization, cardiovascular mortality) among individuals with a history of atherosclerotic cardiovascular disease (ASCVD) (A) and among those without a history of ASCVD (B). The analysis was adjusted for age, sex, self-reported race and ethnicity, hypertension, chronic kidney disease status, non-Lp(a) hyperlipidemia, diabetes, insulin use (in diabetic individuals), and smoking status. 1st-50th percentile: 0–41 nmol/L; 51st-70th percentile: 42–111 nmol/L; 71st-90th percentile: 112–215 nmol/L; 91st-100th: ≥216 nmol/L.
TABLE 3.
Composite Endpoint of Major Adverse Cardiovascular Event for Individuals With and Without Baseline ASCVD
| Composite Endpoint: MACE | ASCVD (n = 10,181) | No ASCVD (n = 6,238) | ||
|---|---|---|---|---|
| 1st-50th percentile | 1,886/5,423 (34.8) | 423/3,618 (11.7) | ||
| 51st-70th percentile | 816/2,086 (39.1) | 144/1,119 (12.9) | ||
| 71st-90th percentile | 749/1,827 (41.0) | 152/1,065 (14.3) | ||
| 91st-100th percentile | 351/845 (41.5) | 94/436 (21.6) | ||
| Unadjusted HR (95% CI) | Adjusteda HR (95% CI) | Unadjusted HR (95% CI) | Adjusteda HR (95% CI) | |
| 1st–50th percentile | Reference | – | Reference | – |
| 51st–70th percentile | 1.18 (1.09–1.28); P < 0.001 | 1.14 (1.05–1.24); P = 0.002 | 1.11 (0.92–1.35); P = 0.265 | 1.09 (0.90–1.32); P = 0.354 |
| 71st–90th percentile | 1.26 (1.15–1.37); P < 0.001 | 1.21 (1.11–1.32); P < 0.001 | 1.24 (1.03–1.49); P = 0.024 | 1.17 (0.97–1.41); P = 0.096 |
| 91st–100th percentile | 1.27 (1.14–1.43); P < 0.001 | 1.26 (1.12–1.41); P < 0.001 | 2.04 (1.63–2.55); P < 0.001 | 1.93 (1.54–2.42); P < 0.001 |
Values are n/N (%) unless otherwise indicated.
Adjusted for age, sex, self-reported race/ethnicity, hypertension, chronic kidney disease status, non-Lp(a) hyperlipidemia, diabetes, insulin use (in diabetic individuals), and smoking status.
MACE = major adverse cardiovascular events; other abbreviations as in Table 2.
FIGURE 3.
Annual Event Rates of the Primary Composite Outcome
Annual event rates (incidence rate per 1,000 patient-years) for the composite endpoint (myocardial infarction, ischemic stroke, coronary revascularization, cardiovascular mortality) among individuals with a history of ASCVD (A) and those without ASCVD (B). Error bars represent the 5% of error, as determined by the 95% CI. Abbreviations as in Figure 2.
Among individuals with a history of ASCVD, there was a robust and stepwise increase in the hazard of MI across Lp(a) percentile groups. Specifically, after adjusting for available covariates, individuals in the highest Lp(a) percentile group had a 79% increase in the hazard of MI compared with the reference group (P < 0.001) (Figure 4).
FIGURE 4.
Association of Lp(a) Percentile Group With Incident Myocardial Infarction
Kaplan-Meier curves demonstrating the association of Lp(a) percentile group on incident myocardial infarction among individuals with a history of ASCVD (A) and among those without a history of ASCVD (B). The analysis was adjusted for age, sex, self-reported race and ethnicity, hypertension, chronic kidney disease status, non-Lp(a) hyperlipidemia, diabetes, insulin use (in diabetic individuals), and smoking status. 1st-50th percentile: 0–41 nmol/L; 51st-70th percentile: 42–111 nmol/L; 71st-90th percentile: 112–215 nmol/L; 91st-100th percentile: ≥216 nmol/L. Abbreviations as in Figure 2.
Notably, however, there was no statistically significant difference in the hazard of cardiovascular mortality or ischemic stroke over the follow-up period after adjustment for baseline covariates. See the Supplemental Appendix Section V for further details regarding cardiovascular mortality and ischemic stroke among individuals with a history of ASCVD.
INDIVIDUALS WITHOUT A HISTORY OF ASCVD.
Over the same follow-up period, among 6,238 individuals without a history of ASCVD, 423 (11.7%) with Lp(a) levels in the 1st to 50th Lp(a) percentile, 144 (12.9%) in the 51st to 70th Lp(a) percentile, 152 (14.3%) in the 71st to 90th Lp(a) percentile, and 94 (21.6%) in the 91st to 100th Lp(a) percentile experienced the primary outcome of MI, ischemic stroke, coronary revascularization, or cardiovascular death (P < 0.001) (Figure 2). In the unadjusted model, individuals in the highest Lp(a) percentile group had an HR of 2.04 (95% CI: 1.63–2.55) of MACE compared with the reference group (P < 0.001). These results persisted after adjusting for available covariates (HR of 1.93; 95% CI: 1.54–2.42; P < 0.001) (Table 3). Annual event rates (per 1,000 patient-years) for the primary composite outcome among individuals without a history of ASCVD are presented in Figure 3.
Among individuals without a history of ASCVD, individuals in the highest Lp(a) percentile experienced a significantly higher hazard of incident MI compared with the reference group (HR: 3.24; 95% CI: 2.14–4.90; P < 0.001) (Figure 4). Additionally, individuals in the highest Lp(a) percentile group had an 89% increase in the hazard of cardiovascular death (P < 0.001) (Supplemental Appendix Section V). Finally, individuals in the highest Lp(a) percentile group had a 75% increase in the hazard of ischemic stroke (P = 0.016) (Supplemental Appendix Section V).
RESTRICTED CUBIC SPLINE ANALYSIS.
For the secondary analytic approach using the raw Lp(a) values, the restricted cubic spline analysis demonstrated that higher Lp(a) levels were consistently associated with an increasing risk of the primary composite outcome for those without baseline ASCVD (Figure 5). For individuals with baseline ASCVD, the hazard of the primary composite outcome began to plateau between 150 and 200 nmol/L. Notably, the slope of the model across ranges of elevated Lp(a) was higher among patients without prior ASCVD. This finding suggests that among patients with existing ASCVD, there may be a smaller gain in incremental risk prediction once Lp(a) levels are above the 70th percentile whereas for those without baseline ASCVD there is a continuous increase in risk with higher Lp(a) levels.
FIGURE 5.
The Association of Lp(a) With the Primary Composite Outcome Based on ASCVD Status
Restricted cubic spline modeling the association of continuous lipoprotein(a) levels on the primary composite outcome (myocardial infarction, ischemic stroke, coronary revascularization, cardiovascular mortality) among individuals with a history of ASCVD (A) and among those without a history of ASCVD (B). Knots were set at 41 nmol/L (50th percentile – reference), 111 nmol/L (70th percentile), and 215 nmol/L (90th percentile), and the splines are truncated at the 99th percentile (310 nmol/L). Abbreviations as in Figure 2.
DISCUSSION
Our findings support the important role of Lp(a) in determining ASCVD risk in both primary and secondary prevention cohorts as demonstrated within a large and well-phenotyped U.S.-based registry. Additionally, this work provides novel insights that have the potential to inform both risk assessment as well as identifying populations for future therapeutic outcomes trials. In particular, we demonstrate the following findings: 1) among individuals with a history of prior ASCVD, there is a meaningful increase in incident ASCVD (particularly for MI and coronary revascularization) with increasing Lp(a), with higher risk predominantly observed once Lp(a) exceeded the 71st percentile (>112 nmol/L or >53 mg/dL), a threshold that is lower than is used in current trials or when using Lp(a) as a risk enhancing feature; 2) among individuals with a history of prior ASCVD, there was no statistically significant association between elevated Lp(a) and ischemic stroke or cardiovascular death; and 3) among individuals without a history of ASCVD, there was a higher Lp(a) threshold for identifying those at elevated risk (ie, >91st percentile; ≥216 nmol/L) and higher risk was observed for all individual components of the composite outcome (ie, MI, coronary revascularization, stroke, and CV death) at this level.
Taken together, our findings have important implications for risk prediction as well as the design of future trials in both primary and secondary prevention populations. Across both primary and secondary prevention groups, there was a meaningful increase in ASCVD risk with increasing Lp(a) levels, with the excess risk being strongest for MI and coronary revascularization, and weakest for stroke and CV death (where it was only observed in our cohort of patients who did not have prior ASCVD). With regard to clinical applicability, our findings suggest that among those with baseline ASCVD, individuals with Lp(a) levels >70th percentile may benefit most from additional aggressive risk factor modification, and possibly in the future, Lp(a)-lowering therapies. However, in those without baseline ASCVD, it appears that those with Lp(a) levels >90th percentile are at the highest attributable risk from elevated Lp(a). In such individuals, aggressive risk factor modification as well as consideration of an antiplatelet agent may be warranted.43,44
Notably, in our secondary prevention cohort, ASCVD risk began to plateau once Lp(a) exceeded the ~70th percentile, whereas in the primary prevention cohort, higher risk was predominantly observed once Lp(a) levels exceeded the >90th percentile (Central Illustration). Although the association between Lp(a) and ASCVD is most often thought of as linear,45,46 the observed plateau in risk for those with background ASCVD is a novel finding that has potentially meaningful implications for risk prediction and the design of future clinical trials. A potential explanation for this leveling of risk is that patients with baseline ASCVD have a higher absolute risk of events and are also more likely to be treated with aggressive preventive therapies, including antiplatelet agents. Accordingly, in an already high-risk population being treated with secondary prevention strategies, the added impact of extremely elevated Lp(a) (above the 90th percentile) may be less potent in terms of future ASCVD risk.
CENTRAL ILLUSTRATION.
Lipoprotein(a), Adverse Cardiovascular Events, and Baseline Atherosclerotic Cardiovascular Disease Status
In this U.S. cohort of 16,419 individuals who underwent lipoprotein(a) testing as part of routine care, elevated lipoprotein(a) was associated with major adverse cardiovascular events—but the risk was different in primary vs secondary prevention cohorts. Specifically, for individuals with baseline atherosclerotic cardiovascular disease (ASCVD), the risk associated with elevated lipoprotein(a) plateaued around the 70th percentile, a finding not observed among individuals without baseline ASCVD where there was a linear association between lipoprotein(a) and risk. Lp(a) = lipoprotein(a).
Accordingly, our data suggest that future trials for Lp(a) lowering in patients who do not have prior ASCVD may benefit from selecting a higher Lp(a) threshold. However, even our patients who did not have prior ASCVD but had an Lp(a) level >90th percentile had an annual event rate of 2.2%, which was lower than the event rates observed among patients with prior ASCVD (which ranged from 4.1% to 5.3%). It is notable that patients without prior ASCVD were approximately 10 years younger and had far fewer risk factors than those with prior ASCVD. Nevertheless, our results suggest that if effective Lp(a)-lowering therapies will ultimately be evaluated in primary prevention populations, further risk enrichment will be necessary to achieve a patient cohort that has sufficient risk. Although not evaluated in our cohort, one potential method to enhance risk assessment among patients with elevated Lp(a) and no prior ASCVD is coronary artery calcium testing,47 because individuals who have elevated Lp(a) and a significant amount of coronary atherosclerosis have a substantially higher risk of future events.
Although our work focuses predominantly on whether a baseline history of ASCVD might modify the association of Lp(a) with MACE, there are a variety of other potential baseline factors that warrant further investigation. In particular, the interaction among biological sex, Lp(a) level, and MACE is important, and there is currently limited data on this subject. When we assessed for effect modification by sex within the full cohort as well as in both the primary prevention and secondary prevention cohorts in our current analysis, there was no statistically significant interaction between sex and MACE. However, further work expanding on this topic is necessary. In addition to biological sex, the interaction of genetic ancestry, race, and ethnicity on the association of Lp(a) with MACE is yet another area that requires careful analysis given conflicting and limited data in this burgeoning field.6,16,45,48
STUDY LIMITATIONS AND STRENGTHS.
Limitations of our work include its retrospective nature and that Lp(a) measurement was performed as part of routine care, which may have led to ascertainment bias. Nevertheless, we believe that our results are generalizable to contemporary patients undergoing routine Lp(a) testing throughout the United States, because universal Lp(a) screening is not endorsed by current U.S. guidelines and is infrequently performed.49 Our results may also be subject to possible residual confounding, although we performed extensive clinical adjustment. Additionally, we did not have blood samples and are not able to provide data on genetic or inflammatory biomarkers. Finally, given that our analysis focused on individuals from 2 large hospitals from within our health care system, with a relatively small proportion of individuals representing ethnic minorities, our findings may be less generalizable to other areas or practice settings.
Additionally, because patients underwent Lp(a) testing as part of routine care over a 19-year period, a number of Lp(a) assays were used during the study period. Despite the different assays used over time, we only incorporated assays that used industry-standard approaches to measure either particle number or Lp(a) concentration. To mitigate the potential issue of multiple assays, we formed internal distributions within each assay type and then combined patient results by generating percentile groups to account for any possible bias related to assay type, a methodology that has been used by other groups as well.2,5,9,27,28 As part of our secondary analytic approach, we maintained the original Lp(a) levels and performed restricted cubic splines on the raw values to further understand the associations of interest, a methodology that demonstrated similar findings as our primary percentile group analytic approach. Additionally, as a further step to ensure the consistency of our results, we analyzed the cohort using different Lp(a) cutpoints aligned with those often used in clinical practice (eg, ≤75 nmol/L, 76–125 nmol/L, 126–175 nmol/L and ≥176 nmol/L), which demonstrated similar findings to our primary analytic approach, as described in Supplemental Appendix Section VII.
Because elevated Lp(a) levels can contribute to the measured total cholesterol and LDL-C levels, we used relatively high cutpoints for defining laboratory-based definitions of hypercholesterolemia, requiring LDL-C ≥160 mg/dL and total cholesterol ≥240 mg/dL. Therefore, it is unlikely that Lp(a) levels in our cohort would meaningfully contribute to falsely categorizing individuals as having non-Lp(a) hyperlipidemia. Additionally, even if some misclassification were to occur in the definition of baseline non-Lp(a) hyperlipidemia, we would expect this to bias our results toward the null by increasing the prevalence of non-Lp(a) hyperlipidemia, because this was a clinical covariate in our adjusted models.
Despite these limitations, our work has particular strengths. These include its large size, its median follow-up time of approximately 12 years, and the robust ascertainment of baseline covariates using a combination of NLP, billing codes, and laboratory data. Additionally, although our cohort’s ascertainment of outcomes may be subject to some degree of loss to follow-up (eg, when a hospitalization for MI occurred at a different health care system), there is likely no differential follow-up based on Lp(a) level, and the study outcomes were assessed using validated methodologies.32–38
CONCLUSIONS
With multiple targeted Lp(a)-lowering therapies now in development, there is an important need to elucidate populations of patients who may benefit from these promising therapies. Although the 2 ongoing phase 3 clinical outcomes trials are targeting individuals who have the highest risk—prior ASCVD events and Lp(a) levels ~90th percentile (eg, >175 nmol/L for HORIZON [A Randomized Double-blind, Placebo-controlled, Multicenter Trial Assessing the Impact of Lipoprotein (a) Lowering With TQJ230 on Major Cardiovascular Events in Patients With Established Cardiovascular Disease] and >200 nmol/L for OCEAN(a)-Outcomes [Olpasiran Trials of Cardiovascular Events and Lipoprotein(a) Reduction (OCEAN[a])–Outcomes Trial])—our work demonstrates that there will likely be a significant population of individuals with and without baseline ASCVD who remain at increased cardiovascular risk from Lp(a) who will not be included in these trials. Thus, in addition to ongoing clinical trials, additional studies are needed to further elucidate how Lp(a) can affect risk in various populations, and whether the excess risk attributable to Lp(a) can be effectively lowered.
Through a large, contemporary U.S.-based cohort with extended follow-up, this study demonstrates the independent association between elevated Lp(a) and MACE for individuals with and without baseline ASCVD. Additionally, this study provides meaningful insights as to the optimal thresholds for risk assessment and how such thresholds may be different for primary and secondary prevention cohorts. These insights can guide both current clinical risk assessment as well as future trials for Lp(a)-lowering therapies as we have identified populations of patients (both primary and secondary prevention) who would not be included in current Lp(a) trials but have significant residual Lp(a) attributable risk.
Supplementary Material
APPENDIX For supplemental sections as well as supplemental figures and tables, please see the online version of this paper.
PERSPECTIVES.
COMPETENCY IN MEDICAL KNOWLEDGE:
Elevated plasma Lp(a) is associated with an increased risk of subsequent MACE, though the threshold of Lp(a) and attributable risk differ based on the presence or absence of baseline ASCVD.
TRANSLATIONAL OUTLOOK:
Future trials of Lp(a) lowering therapies for primary and secondary prevention of ischemic cardiovascular events should incorporate risk estimates to identify individuals most likely to gain benefit.
FUNDING SUPPORT AND AUTHOR DISCLOSURES
This study was funded, in part, by Amgen Inc. Dr Januzzi is a Trustee of the American College of Cardiology; has received grant support from Abbott, Applied Therapeutics, HeartFlow Inc, Innolife, and Roche Diagnostics; has received consulting income from Abbott, AstraZeneca, Bayer, Beckman, Boehringer Ingelheim, Janssen, Novartis, Merck, and Roche Diagnostics; and participates in clinical endpoint committees/data safety monitoring boards for Abbott, AbbVie, Bayer, CVRx, Intercept, Pfizer, and Takeda. Drs Booth III, López, and Kent are employees and shareholders of Amgen Inc. Dr Bhatt has served on the Advisory Board of Angiowave, Bayer, Boehringer Ingelheim, CellProthera, Cereno Scientific, Elsevier Practice Update Cardiology, High Enroll, Janssen, Level Ex, McKinsey, Medscape Cardiology, Merck, MyoKardia, NirvaMed, Novo Nordisk, PhaseBio, PLx Pharma, and Stasys; has served on the Board of Directors of American Heart Association New York City, Angiowave (stock options), Bristol Myers Squibb (stock), DRS.LINQ (stock options), and High Enroll (stock); has served as a consultant for Broadview Ventures, Hims, SFJ, and Youngene; has served on Data Monitoring Committees for Acesion Pharma, Assistance Publique-Hôpitaux de Paris, Baim Institute for Clinical Research (formerly Harvard Clinical Research Institute, for the PORTICO trial, funded by St. Jude Medical, now Abbott), Boston Scientific (Chair, PEITHO trial), Cleveland Clinic, Contego Medical (Chair, PERFORMANCE 2), Duke Clinical Research Institute, Mayo Clinic, Mount Sinai School of Medicine (for the ENVISAGE trial, funded by Daiichi-Sankyo; for the ABILITY-DM trial, funded by Concept Medical; and for ALLAY-HF, funded by Alleviant Medical), Novartis, Population Health Research Institute, and Rutgers University (for the National Institutes of Health-funded MINT Trial); has received honoraria from the American College of Cardiology (Senior Associate Editor, Clinical Trials and News, ACC.org; Chair, ACC Accreditation Oversight Committee), Arnold and Porter law firm (work related to Sanofi/Bristol Myers Squibb clopidogrel litigation), Baim Institute for Clinical Research (formerly Harvard Clinical Research Institute; RE-DUAL PCI clinical trial steering committee funded by Boehringer Ingelheim; AEGIS-II executive committee funded by CSL Behring), Belvoir Publications (Editor-in-Chief, Harvard Heart Letter), Canadian Medical and Surgical Knowledge Translation Research Group (clinical trial steering committees), CSL Behring (American Heart Association lecture), Cowen and Company, Duke Clinical Research Institute (clinical trial steering committees, including for the PRONOUNCE trial, funded by Ferring Pharmaceuticals), HMP Global (Editor-in-Chief, Journal of Invasive Cardiology), Journal of the American College of Cardiology (Guest Editor; Associate Editor), K2P (Co-Chair, interdisciplinary curriculum), Level Ex, Medtelligence/ReachMD (CME steering committees), MJH Life Sciences, Oakstone CME (Course Director, Comprehensive Review of Interventional Cardiology), Piper Sandler, Population Health Research Institute (for the COMPASS operations committee, publications committee, steering committee, and USA national co-leader, funded by Bayer), WebMD (CME steering committees), and Wiley (steering committee); served as Deputy Editor of Clinical Cardiology; is named on a patent for sotagliflozin assigned to Brigham and Women’s Hospital who assigned to Lexicon (neither he nor Brigham and Women’s Hospital receive any income from this patent); has received research funding from Abbott, Acesion Pharma, Afimmune, Aker Biomarine, Alnylam, Amarin, Amgen, AstraZeneca, Bayer, Beren, Boehringer Ingelheim, Boston Scientific, Bristol Myers Squibb, Cardax, CellProthera, Cereno Scientific, Chiesi, CinCor, Cleerly, CSL Behring, Eisai, Ethicon, Faraday Pharmaceuticals, Ferring Pharmaceuticals, Forest Laboratories, Fractyl, Garmin, HLS Therapeutics, Idorsia, Ironwood, Ischemix, Janssen, Javelin, Lexicon, Lilly, Medtronic, Merck, Moderna, MyoKardia, NirvaMed, Novartis, Novo Nordisk, Otsuka, Owkin, Pfizer, PhaseBio, PLx Pharma, Recardio, Regeneron, Reid Hoffman Foundation, Roche, Sanofi, Stasys, Synaptic, The Medicines Company, Youngene, and 89Bio; has received royalties from Elsevier (Editor, Braunwald’s Heart Disease); has served as site co-investigator for Abbott, Biotronik, Boston Scientific, CSI, Endotronix, St. Jude Medical (now Abbott), Philips, SpectraWAVE, Svelte, and Vascular Solutions; is a Trustee of the American College of Cardiology; and has performed unfunded research for FlowCo. Dr Blankstein has received research support and consulting fees from Amgen Inc and Novartis Inc. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
ABBREVIATIONS AND ACRONYMS
- ASCVD
atherosclerotic cardiovascular disease
- Lp(a)
lipoprotein(a)
- MACE
major adverse cardiovascular event(s)
Footnotes
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the Author Center.
REFERENCES
- 1.Erqou S, Kaptoge S, Perry PL, et al. for the Emerging Risk Factors Collaboration. Lipoprotein(a) concentration and the risk of coronary heart disease, stroke, and nonvascular mortality. JAMA. 2009;302(4):412–423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Kamstrup PR, Benn M, Tybjaerg-Hansen A, Nordestgaard BG. Extreme lipoprotein(a) levels and risk of myocardial infarction in the general population: the Copenhagen City Heart Study. Circulation. 2008;117(2):176–184. [DOI] [PubMed] [Google Scholar]
- 3.Kamstrup PR, Tybjaerg-Hansen A, Steffensen R, Nordestgaard BG. Genetically elevated lipoprotein(a) and increased risk of myocardial infarction. JAMA. 2009;301(22):2331–2339. [DOI] [PubMed] [Google Scholar]
- 4.Thanassoulis G, Campbell CY, Owens DS, et al. Genetic associations with valvular calcification and aortic stenosis. N Engl J Med. 2013;368(6):503–512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kamstrup PR, Tybjaerg-Hansen A, Nordestgaard BG. Elevated lipoprotein(a) and risk of aortic valve stenosis in the general population. J Am Coll Cardiol. 2014;63(5):470–477. [DOI] [PubMed] [Google Scholar]
- 6.Pare G, Caku A, McQueen M, et al. Lipoprotein(a) levels and the risk of myocardial infarction among 7 ethnic groups. Circulation. 2019;139(12):1472–1482. [DOI] [PubMed] [Google Scholar]
- 7.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(26):2518–2528. [DOI] [PubMed] [Google Scholar]
- 8.Nave AH, Lange KS, Leonards CO, et al. Lipoprotein (a) as a risk factor for ischemic stroke: a meta-analysis. Atherosclerosis. 2015;242(2):496–503. [DOI] [PubMed] [Google Scholar]
- 9.Langsted A, Nordestgaard BG, Kamstrup PR. Elevated lipoprotein(a) and risk of ischemic stroke. J Am Coll Cardiol. 2019;74(1):54–66. [DOI] [PubMed] [Google Scholar]
- 10.Tsimikas S A Test in Context: Lipoprotein(a): diagnosis, prognosis, controversies, and emerging therapies. J Am Coll Cardiol. 2017;69(6):692–711. [DOI] [PubMed] [Google Scholar]
- 11.Berman AN, Blankstein R. Current and future role of lipoprotein(a) in preventive cardiology. Curr Opin Cardiol. 2019;34(5):514–518. [DOI] [PubMed] [Google Scholar]
- 12.Berman AN, Blankstein R. Optimizing dyslipidemia management for the prevention of cardiovascular disease: a focus on risk assessment and therapeutic options. Curr Cardiol Rep. 2019;21(9):110. [DOI] [PubMed] [Google Scholar]
- 13.Marston NA, Gurmu Y, Melloni GEM, et al. The Effect of PCSK9 (proprotein convertase subtilisin/kexin type 9) inhibition on the risk of venous thromboembolism. Circulation. 2020;141(20):1600–1607. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Koschinsky ML, Marcovina SM. Structure-function relationships in apolipoprotein(a): insights into lipoprotein(a) assembly and pathogenicity. Curr Opin Lipidol. 2004;15(2):167–174. [DOI] [PubMed] [Google Scholar]
- 15.Thanassoulis G Screening for high lipoprotein(a). Circulation. 2019;139(12):1493–1496. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Nordestgaard BG, Chapman MJ, Ray K, et al. Lipoprotein(a) as a cardiovascular risk factor: current status. Eur Heart J. 2010;31(23):2844–2853. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.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(11):2239–2245. [DOI] [PubMed] [Google Scholar]
- 18.Tsimikas S, Marcovina SM. Ancestry, lipoprotein(a), and cardiovascular risk thresholds: JACC review topic of the week. J Am Coll Cardiol. 2022;80(9):934–946. [DOI] [PubMed] [Google Scholar]
- 19.Hsieh G, Rizk T, Berman AN, Biery DW, Blankstein R. The current landscape of lipoprotein(a) in calcific aortic valvular disease. Curr Opin Cardiol. 2021;36(5):542–548. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Tsimikas S, Karwatowska-Prokopczuk E, Gouni-Berthold I, et al. Lipoprotein(a) reduction in persons with cardiovascular disease. N Engl J Med. 2020;382(3):244–255. [DOI] [PubMed] [Google Scholar]
- 21.Tsimikas S, Moriarty PM, Stroes ES. Emerging RNA therapeutics to lower blood levels of Lp(a): JACC focus seminar 2/4. J Am Coll Cardiol. 2021;77(12):1576–1589. [DOI] [PubMed] [Google Scholar]
- 22.O’Donoghue ML, Rosenson RS, Gencer B, et al. Small interfering RNA to reduce lipoprotein(a) in cardiovascular disease. N Engl J Med. 2022;387(20):1855–1864. [DOI] [PubMed] [Google Scholar]
- 23.O’Donoghue ML, Fazio S, Giugliano RP, et al. Lipoprotein(a), PCSK9 inhibition, and cardiovascular risk. Circulation. 2019;139(12):1483–1492. [DOI] [PubMed] [Google Scholar]
- 24.Bittner VA, Szarek M, Aylward PE, et al. Effect of alirocumab on lipoprotein(a) and cardiovascular risk after acute coronary syndrome. J Am Coll Cardiol. 2020;75(2):133–144. [DOI] [PubMed] [Google Scholar]
- 25.Szarek M, Bittner VA, Aylward P, et al. Lipoprotein(a) lowering by alirocumab reduces the total burden of cardiovascular events independent of low-density lipoprotein cholesterol lowering: ODYSSEY OUTCOMES trial. Eur Heart J. 2020;41(44):4245–4255. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Berman AN, Biery DW, Ginder C, et al. Study of lipoprotein(a) and its impact on atherosclerotic cardiovascular disease: design and rationale of the Mass General Brigham Lp(a) Registry. Clin Cardiol. 2020;43(11):1209–1215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Langsted A, Kamstrup PR, Nordestgaard BG. High lipoprotein(a) and high risk of mortality. Eur Heart J. 2019;40(33):2760–2770. [DOI] [PubMed] [Google Scholar]
- 28.Szarek M, Reijnders E, Jukema JW, et al. Relating lipoprotein(a) concentrations to cardiovascular event risk after acute coronary syndrome: a comparison of three tests. Circulation. 2024;149:192–203. 10.1161/CIRCULATIONAHA.123.066398 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Madsen CM, Kamstrup PR, Langsted A, Varbo A, Nordestgaard BG. Lipoprotein(a)-lowering by 50 mg/dL (105 nmol/L) may be needed to reduce cardiovascular disease 20% in secondary prevention: a population-based study. Arterioscler Thromb Vasc Biol. 2020;40(1):255–266. [DOI] [PubMed] [Google Scholar]
- 30.Research Patient Data Registry Mass General Brigham. RPDR Data Request. Accessed January 5, 2023. https://rc.partners.org/research-apps-services/identify-subjects-request-data#research-patient-data-registry
- 31.Berman AN, Biery DW, Ginder C, et al. Natural language processing for the assessment of cardiovascular disease comorbidities: the cardio-Canary comorbidity project. Clin Cardiol. 2021;44(9):1296–1304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Kiyota Y, Schneeweiss S, Glynn RJ, Cannuscio CC, Avorn J, Solomon DH. Accuracy of Medicare claims-based diagnosis of acute myocardial infarction: estimating positive predictive value on the basis of review of hospital records. Am Heart J. 2004;148(1):99–104. [DOI] [PubMed] [Google Scholar]
- 33.Selby J, Reichman ME, Graham D, et al. Mini-Sentinel Medical Product Assessment: a protocol for active surveillance of acute myocardial infarction in association with use of anti-diabetic agents. Published 2016. Updated January 27, 2016. Accessed October 16, 2019. https://www.sentinelinitiative.org/sites/default/files/Drugs/Assessments/Mini-Sentinel_AMI-and-Anti-Diabetic-Agents_Protocol_0.pdf
- 34.Yeh RW, Sidney S, Chandra M, Sorel M, Selby JV, Go AS. Population trends in the incidence and outcomes of acute myocardial infarction. N Engl J Med. 2010;362(23):2155–2165. [DOI] [PubMed] [Google Scholar]
- 35.Andrade SE, Harrold LR, Tjia J, et al. A systematic review of validated methods for identifying cerebrovascular accident or transient ischemic attack using administrative data. Pharmacoepidemiol Drug Saf. 2012;21(Suppl 1):100–128. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Birman-Deych E, Waterman AD, Yan Y, Nilasena DS, Radford MJ, Gage BF. Accuracy of ICD-9-CM codes for identifying cardiovascular and stroke risk factors. Med Care. 2005;43(5):480–485. [DOI] [PubMed] [Google Scholar]
- 37.Patorno E, Goldfine AB, Schneeweiss S, et al. Cardiovascular outcomes associated with canagliflozin versus other non-gliflozin antidiabetic drugs: population based cohort study. BMJ. 2018;360:k119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Cozzolino F, Montedori A, Abraha I, et al. A diagnostic accuracy study validating cardiovascular ICD-9-CM codes in healthcare administrative databases. The Umbria Data-Value Project. PLoS One. 2019;14(7):e0218919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Coady SA, Sorlie PD, Cooper LS, Folsom AR, Rosamond WD, Conwill DE. Validation of death certificate diagnosis for coronary heart disease: the Atherosclerosis Risk in Communities (ARIC) Study. J Clin Epidemiol. 2001;54(1):40–50. [DOI] [PubMed] [Google Scholar]
- 40.Ives DG, Samuel P, Psaty BM, Kuller LH. Agreement between nosologist and cardiovascular health study review of deaths: implications of coding differences. J Am Geriatr Soc. 2009;57(1):133–139. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Chen Y, Freedman ND, Albert PS, et al. Association of cardiovascular disease with premature mortality in the United States. JAMA Cardiol. 2019;4(12):1230–1238. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Harriss LR, Ajani AE, Hunt D, et al. Accuracy of national mortality codes in identifying adjudicated cardiovascular deaths. Aust N Z J Public Health. 2011;35(5):466–476. [DOI] [PubMed] [Google Scholar]
- 43.Lacaze P, Bakshi A, Riaz M, et al. Aspirin for primary prevention of cardiovascular events in relation to lipoprotein(a) genotypes. J Am Coll Cardiol. 2022;80(14):1287–1298. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Patel Siddharth M, Bonaca Marc P, Morrow David A, et al. Lipoprotein(a) and benefit of antiplatelet therapy. JACC: Adv. 2023;2(9):100675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Kronenberg F, Mora S, Stroes ESG, et al. Lipoprotein(a) in atherosclerotic cardiovascular disease and aortic stenosis: a European Atherosclerosis Society consensus statement. Eur Heart J. 2022;43(39):3925–3946. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Patel AP, Wang M, Pirruccello JP, et al. Lp(a) (Lipoprotein[a]) concentrations and incident atherosclerotic cardiovascular disease: new insights from a large national biobank. Arterioscler Thromb Vasc Biol. 2021;41(1):465–474. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Mehta A, Vasquez N, Ayers CR, et al. Independent association of lipoprotein(a) and coronary artery calcification with atherosclerotic cardiovascular risk. J Am Coll Cardiol. 2022;79(8):757–768. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.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(4):996–1001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Bhatia HS, Hurst S, Desai P, Zhu W, Yeang C. Lipoprotein(a) testing trends in a large academic health system in the United States. J Am Heart Assoc. 2023;12(18):e031255. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.