Inflammation, Cholesterol, Lipoprotein(a) and 30-Year Cardiovascular Outcomes in Women

Abstract

Background

High-sensitivity C-reactive protein (hsCRP), low-density lipoprotein cholesterol (LDL-C), and lipoprotein(a) [Lp(a)] contribute to 5-year and 10-year predictions of cardiovascular risk and represent distinct pathways for pharmacologic intervention. We addressed the utility of these biomarkers over longer periods of time in women as early life intervention represents an important risk reduction method.

Methods

In the Women’s Health Study, levels of hsCRP, LDL-C, and Lp(a) were measured at baseline among 27,939 initially healthy American women who were followed over 30 years for the primary end point of first major adverse cardiovascular event, a composite of myocardial infarction, coronary revascularization, stroke, or cardiovascular death. Adjusted hazard ratios and 95% confidence intervals (CI) were calculated across quintiles of each biomarker, along with 30-year cumulative incidence curves adjusted for competing risks.

Results

The mean baseline age of participants was 54.7 years. During 30-year follow-up, 3662 first major cardiovascular events accrued. Increasing quintiles of hsCRP, LDLC, and Lp(a) all predicted 30-year risks. Adjusted hazard ratios for the primary end point comparing the top to bottom quintile were 1.70 (95% CI, 1.52 to 1.90) for hsCRP, 1.36 (95% CI, 1.23 to 1.52) for LDL-C, and 1.33 (95% CI, 1.21 to 1.47) for Lp(a). Findings appeared consistent for coronary heart disease and stroke. Each biomarker displayed independent contributions to overall risk. The greatest spread for risk was obtained in models using all three biomarkers.

Conclusions

A single bundled measure of hsCRP, LDL-C, and Lp(a) among initially healthy American women is associated with incident cardiovascular events over 30-years. These data support efforts to extend strategies for primary prevention of atherosclerotic events beyond traditional 10-year estimates of risk. (Funding: National Institutes of Health, National Heart, Lung and Blood Institute, USA. ClinicalTrials.gov: NCT00000479)

Introduction

Blood biomarkers can be instrumental for understanding biology and for targeting cardiovascular interventions as demonstrated for the measurement and pharmacologic reduction of low-density lipoprotein cholesterol (LDL-C)(1). At a time when the clinical community is moving beyond evaluation and reduction of LDL-C alone, these issues have become particularly important for high-sensitivity C-reactive protein (hsCRP), a biomarker of low-grade vascular inflammation, and lipoprotein(a) (Lp(a)), a genetically determined lipid fraction. To date, three randomized placebo-controlled trials demonstrate that reducing inflammation can significantly lower cardiovascular event rates (2–4) and one anti-inflammatory agent, low-dose colchicine, was recently approved by the US FDA for atherosclerosis event reduction (5). Several outcome trials of agents that lower Lp(a) are ongoing (6–9).

Current guidelines for the primary and secondary prevention of atherosclerotic disease are shifting to include a broader assessment of hsCRP and Lp(a) along with LDL-C, evidence consistent with contemporary studies based on short-term (3 to 5 year) follow-up among individuals taking and not taking statin therapy (10–14). However, data are scarce on the long-term (25 to 30 year) risks associated with these biomarkers alone and in combination. As atherosclerotic disease develops over decades, yet early life interventions represent an important method for risk reduction, these issues are a major concern particularly among women for whom cardiovascular disease remains underdiagnosed and undertreated.

As Lp(a) levels are genetically determined and stable over time, measurement is recommended once without need for repeat evaluation. Consequently, we hypothesized that bundling together hsCRP, LDL-C, and Lp(a) at a single time point might provide a useful method for lifetime biomarker risk assessment. We addressed this issue in the Women’s Health Study (WHS), a prospective cohort of 27,939 initially healthy American women who had all three biomarkers measured at baseline and who have been followed over a period of 30 years for incident myocardial infarction, coronary revascularization procedures, stroke, and cardiovascular death.

Methods

Participants, End Points, and Biomarker Assays

The WHS enrolled 39,876 healthy American female health professionals between 1992 and 1995 (15). At baseline, participants provided information on behavioral, lifestyle, and demographic risk factors. All participants have systematically been followed prospectively through January 2023 with maximal follow-up time curtailed at 30 years. The primary end point for these analyses was the occurrence of first major incident cardiovascular events, inclusive of: incident myocardial infarction that was confirmed if the reported event was associated with myocardial damage biomarkers or diagnostic electrocardiographic criteria; incident stroke that was confirmed if the participant had new neurologic defects that persisted for greater than 24 hours and were classified as hemorrhagic or ischemic based upon computed tomography or magnetic resonance scanning; coronary revascularizations confirmed by hospital reports; and deaths from cardiovascular disease confirmed with additional data from autopsy reports and death certificates.

At enrollment, participants were offered the opportunity to provide a blood sample collected in ethylenediaminetetraacetic acid, which was processed and stored centrally in a liquid nitrogen biobank. Baseline samples were transferred to a certified core laboratory for assay. Levels of hsCRP were measured with a validated high-sensitivity assay (Denka Seikan), LDL-C was measured with a direct-measurement assay (Roche Diagnostics), and Lp(a) was measured by an assay independent of apolipoprotein(a) isoform size (Denka Seikan)(16,17). Of 28,345 study participants who elected to provide a baseline blood sample, assays were conducted in 27,939.

Statistical Analysis

Spearman correlation coefficients were used to discern baseline relationships between hsCRP, LDL-C, and Lp(a). The study population was then divided according to increasing quintiles of each biomarker and hazard ratios for incident cardiovascular events assessed in cause-specific Cox proportional hazard models comparing quintiles 2 through 5 with the lowest (referent) quintile, including other deaths as a competing risk. Estimates for hazard ratios were obtained in models adjusted for age, in models further adjusted for blood pressure, smoking, and diabetes, and finally in models additionally including the other biomarkers simultaneously. The WHS was initially designed as a 2×2 factorial trial of aspirin and vitamin E which had minimal effects on incident vascular disease (18,19). However, per WHS protocols, all analyses additionally controlled for baseline randomization status. Post-hoc analyses further adjusted for use of hormone replacement therapy, body mass index, and estimated glomerular filtration rate (eGFR). Cumulative incidence function curves adjusting for age and the competing risks of death from other causes were constructed for each biomarker using cause-specific Cox models. We examined proportionality of the hazard ratio using an interaction with time, performed landmark analyses partitioning follow-up time into periods of up to and after 15 years of follow-up, tested the continuous effect of each marker, and conducted analyses using alternative Fine-Gray modeling (Supplemental Methods).

To assess potential joint effects between any two of the biomarkers, risk factor adjusted analyses were conducted assessing event rates according to whether paired baseline biomarker levels were above or below the respective cohort medians. To increase clinical interpretation, similar joint effect analyses were also conducted according to clinically relevant thresholds for each biomarker.

To assess the potential combined effects of all three biomarkers, we repeated the above analysis after classifying participants based upon having 0, 1, 2, or 3 levels of baseline hsCRP, LDL-C, or Lp(a) in the 5th quintile.

Finally, to assess for any potential bias that might have been introduced due to increasing statin use over time, we performed a sensitivity analysis in which follow-up was censored at the time of first reported statin prescription.

In all instances, separate analyses were performed for the endpoints of total cardiovascular events, coronary heart disease events, and stroke events. Confidence intervals are calculated at the 95% level. Confidence interval widths have not been adjusted for multiplicity and should not be used in place of hypothesis testing.

All authors contributed to study design and data gathering, vouch for the analyses, and agreed to publish the paper. Dr. Ridker wrote all the drafts of the manuscript.

Results

Baseline Characteristics and Biomarker Correlations

At cohort assembly, the mean age of the 27,939 participants was 54.7 years, 25% had hypertension, 12% were current smokers, 2.5% had diabetes, the mean body mass index was 25.9 kg/m², 14.4% had a parental history of myocardial infarction before age 65, and 94% were Caucasian (Table 1 and Table S1). There was minimal correlation between biomarkers; the Spearman correlation coefficient between hsCRP and LDL-C was 0.08, between LDL-C and Lp(a) was 0.17, and between hsCRP and Lp(a) was 0.01.

Table 1.

Baseline distributions of age, hypertension, diabetes, and smoking status according to increasing quintiles of hsCRP, LDL-C, and Lp(a) among 27,939 initially healthy American women participating in the Women's Health Study

Quintile of Baseline LDLC
	Quintile 1	Quintile 2	Quintile 3	Quintile 4	Quintile 5
Range, mg/dL	<96.1	96.1-<113.5	113.5-<129.7	129.7-<150.7	≥ 150.7
No.	5616	5576	5583	5580	5584
Age, years median— IQR	51 (48–57)	52 (48–57)	53 (49–59)	54 (50–60)	55 (50–61)
Hypertension — %	20.5	22.2	25.5	27.7	29.8
Diabetes — %	2.1	2.2	2.3	2.6	3.1
Current Smokers — %	9.7	10.5	10.8	13.0	14.3
Quintile of Baseline Lp(a)
	Quintile 1	Quintile 2	Quintile 3	Quintile 4	Quintile 5
Range, mg/dL	<3.6	3.6-<7.6	7.6-<15.5	15.5-<44.1	≥ 44.1
No.	5694	5366	5619	5524	5545
Age, years median— IQR	53 (49–59)	53 (49–59)	53 (49–59)	53 (49–59)	53 (49–50)
Hypertension — %	26.9	23.2	23.6	26.1	25.7
Diabetes—%	3.1	2.4	2.1	2.3	2.4
Current Smokers — %	11.9	11.6	11.4	11.4	11.6

Primary End Point

During the 30-year follow-up (median 27.4 years; interquartile range, 22.6 to 28.5 years), 3662 confirmed first major cardiovascular events accrued. In age-adjusted and covariable-adjusted models, the risks for first major cardiovascular events across increasing quintiles for all three biomarkers are shown in Table 2. In the full cohort, the covariable adjusted hazard ratio for this primary end point comparing the top to bottom quintile was 1.70 (95% CI, 1.52 to 1.90) for hsCRP, 1.36 (95% CI, 1.23 to 1.52) for LDL-C, and 1.33 (95% CI, 1.21 to 1.47) for Lp(a). In analysis further adjusting simultaneously for each of the other biomarkers, on a per quintile basis the observed hazard ratios were 1.14 (95% CI, 1.11 to 1.17) for hsCRP, 1.08 (95% CI, 1.05 to 1.10) for LDL-C, and 1.06 (95% CI, 1.03 to 1.08) for Lp(a). On a per standard deviation basis, the covariable and biomarker adjusted hazard ratio for the primary end point was 1.23 (95% CI, 1.18 to 1.27) for hsCRP, 1.10 (95% CI 1.06 to 1.14) for LDL-C, and 1.09 (95% CI 1.05–1.13) for Lp(a)(Table S2). Post-hoc additional adjustment for use of hormone replacement therapy, body mass index, and eGFR did not appear to have an impact on effect estimates. As also shown in Table 2, effects for hsCRP and LDL-C marginally attenuated over time. For hsCRP, the per quintile hazard ratio for the primary end point was 1.17 (95% CI 1.13 to 1.21) in years 0 to 15 and 1.12 (95% CI 1.09 to 1.16) in years 15 to 30. For LDL-C, the per quintile hazard ratio for the primary end point was 1.13 (95% CI 1.09 to 1.17) in years 0 to 15 and 1.06 (95% CI 1.02 to 1.09) in years 15 to 30. No attenuation over time was observed for Lp(a).

Table 2.

Primary End Point: 30-year risks for first major cardiovascular events according to baseline levels of hsCRP, LDL-C, and Lp(a)

	Quintile of Baseline hsCRP
	Quintile 1	Quintile 2	Quintile 3	Quintile 4	Quintile 5	Per Quintile
Range, mg/L	<0.65	0.65-<1.47	1.47-<2.75	2.75-<5.18	≥ 5.18
no./total no.	455/5659	586/5575	695/5539	841/5582	1085/5584
HRage adjusted	1.0	1.16	1.33	1.63	2.22	1.22
95% CI	referent	1.03–1.31	1.18–1.49	1.45–1.82	1.99–2.47	1.19–1.25
HRcovariable adjusted	1.0	1.11	1.19	1.38	1.70	1.14
95% CI	referent	0.98–1.25	1.05–1.34	1.22–1.55	1.52–1.90	1.12–1.17
HRcovariable adjusted, years 0 to 15	1.0	1.28	1.39	1.58	1.97	1.17
95% CI	referent	1.05–1.56	1.15–1.68	1.31–1.90	1.64–2.36	1.13–1.21
HRcovariable adjusted, years 16 to 30	1.0	1.01	1.07	1.26	1.52	1.12
95% CI	referent	0.86–1.18	0.92–1.25	1.09–1.47	1.32–1.77	1.09–1.16
HRbiomarker adjusted	1.0	1.09	1.16	1.33	1.68	1.14
95% CI	referent	0.96–1.23	1.02–1.30	1.18 –1.50	1.50–1.88	1.11–1.17
	Quintile of Baseline LDLC
	Quintile 1	Quintile 2	Quintile 3	Quintile 4	Quintile 5	Per Quintile
Range, mg/L	<96.1	96.1-<113.5	113.5-<129.7	129.7-<150.7	≥ 150.7
no./total no.	550/5616	594/5576	689/5583	842/5580	987/5584
HRage adjusted	1.0	1.02	1.10	1.31	1.48	1.11
95% CI	referent	0.91–1.14	0.99–1.24	1.17–1.46	1.33–1.64	1.09–1.40
HRcovariable adjusted	1.0	1.00	1.07	1.23	1.36	1.09
95% CI	referent	0.89–1.12	0.95–1.19	1.10–1.37	1.23–1.52	1.07–1.12
HRcovariable adjusted, years 0 to 15	1.0	1.00	1.14	1.35	1.56	1.13
95% CI	referent	0.83–1.20	0.96–1.36	1.14–1.59	1.33–1.83	1.09–1.17
HRcovariable adjusted, years 16 to 30	1.0	1.00	1.01	1.14	1.21	1.06
95% CI	referent	0.86–1.16	0.87–1.18	0.99–1.32	1.05–1.40	1.02–1.09
HRbiomarker adjusted	1.0	0.99	1.06	1.19	1.31	1.08
95% CI	referent	0.88–1.11	0.94–1.19	1.07–1.33	1.18–1.46	1.05–1.10
	Quintile of Baseline Lp(a)
	Quintile 1	Quintile 2	Quintile 3	Quintile 4	Quintile 5	Per Quintile
Range, mg/L	<3.6	3.6-<7.6	7.6-<15.5	15.5-<44.1	≥ 44.1
no./total no.	706/5694	631/5366	643/5619	716/5524	900/5545
HRage adjusted	1.0	0.97	0.89	1.01	1.30	1.06
95% CI	referent	0.87–1.07	0.80–0.99	0.91–1.12	1.18–1.44	1.04–1.09
HRcovariable adjusted	1.0	1.01	0.94	1.04	1.33	1.07
95% CI	referent	0.91–1.12	0.85–1.05	0.93–1.15	1.21–1.47	1.04–1.09
HRcovariable adjusted, years 0 to 15	1.0	0.99	0.92	0.89	1.42	1.07
95% CI	referent	0.85–1.16	0.79–1.09	0.76–1.05	1.23–1.64	1.04–1.11
HRcovariable adjusted, years 16 to 30	1.0	1.02	0.96	1.17	1.26	1.06
95% CI	referent	0.88–1.18	0.83–1.11	1.02–1.34	1.10–1.45	1.03–1.10
HRbiomarker adjusted	1.0	1.02	0.93	1.02	1.29	1.06
95% CI	referent	0.91–1.13	0.84–1.04	0.92–1.14	1.17–1.43	1.03–1.08

All models adjusted for age and initial randomization treatment group. Covariable models additionally adjusted for smoking (current, past, never), presence of diabetes, and Framingham blood pressure categories. Biomarker model additionally adjusted for the other two biomarkers. Confidence interval widths have not been adjusted for multiplicity and may not be used in place of hypothesis testing.

Age and competing risk adjusted cumulative incidence curves for the probability of an incident major adverse cardiovascular event was graded across quintiles of hsCRP and LDL-C. By contrast, risk was increased for Lp(a) primarily among individuals in the top quintile (>44 mg/dL)(Figure 1 and Figures S1 and S2). When evaluated using alternative Fine-Gray models, the estimated subdistribution hazard ratios appeared similar (Table S3).

Figure 1.

30-year age and competing risk adjusted cumulative incidence of first major cardiovascular events among initially healthy women according to baseline levels of hsCRP, LDL-C, and Lp(a). Quintile 1 (Blue), Quintile 2 (Red), Quintile 3 (Green), Quintile 4 (Black), Quintile 5 (Purple).

Coronary Heart Disease and Stroke

Individual levels of hsCRP, LDL-C, and Lp(a) each predicted 30-year risks for the individual endpoints of coronary heart disease and stroke (Table S4). Supplementary Figures 1 and 2 present 30-year age and competing risk adjusted cumulative incidence curves for coronary heart disease events and stroke events, respectively, among initially healthy women according to baseline levels of hsCRP, LDL-C, and Lp(a).

Joint Effects

The results of joint effects analyses addressing paired baseline biomarker levels above or below the cohort medians on 30-year risk is shown in Table S5. Figure 2 presents age and competing risk adjusted cumulative incidence curves in joint effect analyses for individuals with levels of hsCRP, LDL-C, and Lp(a) less than or greater than or equal to the thresholds of 2 mg/L, 130 mg/dL, and 40 mg/dL, respectively. Alternative use of Lp(a) thresholds of 30 or 50 mg/dL had minimal impact on these observations.

Figure 2.

Joint effects of hsCRP, LDL-C, and Lp(a) on 30-year age and competing risk adjusted cumulative incidence of first major cardiovascular events among initially healthy women. Data are shown for values of hsCRP ≥ or < 2 mg/L; for LDLC ≥ or < 130 mg/dL, and Lp(a) ≥ or < 40 mg/dL. Alternative use of Lp(a) thresholds of 30 or 50 mg/dL had minimal impact on these observations.

Combined Effects

Levels of hsCRP, LDL-C, and Lp(a) displayed independent contributions to risk and the greatest spread for risk was obtained in models using all three biomarkers in combination. For the primary end point, the covariable adjusted hazard ratios for individuals with 0, 1, 2, or 3 biomarker levels in the 5th quintile were 1.0 (referent), 1.27 (95% CI, 1.19 to 1.37), 1.66 (95% CI, 1.51 to 1.83), and 2.63 (95% CI, 2.16 to 3.19)(Table S6). Similar combined effects were observed for the individual end points of stroke and coronary heart disease where the hazard ratios for those with all three biomarkers in the top quintile were 1.68 (95% CI, 1.14 to 2.48) and 3.71 (95% CI, 2.94 to 4.68), respectively (Figure S3).

Figure 3 presents age and competing risk adjusted cumulative incidence curves for the probability of first major cardiovascular events, coronary heart disease events, and stroke events over the 30-year follow-up period in women according to the number of baseline biomarker levels in the top quintile (0, 1, 2, or 3).

Figure 3.

Combined effects of hsCRP, LDL-C, and Lp(a) on 30-year age and competing risk adjusted cumulative incidence of first major cardiovascular events, coronary heart disease events, and stroke events among initially healthy women according to the number of baseline biomarker levels in the top quintile (0, 1, 2, or 3).

Sensitivity Analyses with Follow-up Censored at Time of Statin Prescription

Statin use was rare at cohort initiation but became increasingly frequent over time (particularly after year 15) such that by year 30, 16,053 women (57.5%) reported having received at least one prescription for statin therapy. As the WHS lacks high-level data on statin compliance or duration of use, we performed a conservative sensitivity analysis in which follow-up time was censored at the time of first reported statin prescription. Of 3662 incident first major cardiovascular events observed in the full cohort, 2151 occurred prior to a report of statin prescription.

As observed in the full cohort, in this sensitivity analysis, the age-adjusted and covariable-adjusted risks for first major cardiovascular events increased across increasing quintiles for all three biomarkers. Censoring follow-up at time of statin initiation, the covariable-adjusted hazard ratio for the top versus bottom quintile was 1.65 (95% CI, 1.43 to 1.90) for hsCRP, 1.62 (95% CI, 1.41 to 1.86) for LDL-C, and 1.42 (95% CI, 1.25 to 1.62) for Lp(a) (Table S7). In a manner almost identical to that observed in the total cohort, age and competing risk adjusted cumulative incidence curves constructed for these sensitivity analyses continued to demonstrate strong predictive effects for all three biomarkers over time (Figure S4).

Finally, as in the total cohort, the greatest spread for long-term risk was again obtained in models using all three biomarkers. In these sensitivity analyses, the covariable adjusted hazard ratios for those with all three biomarkers in the top quintile were 3.21 (95% CI, 2.41 to 4.27) for the primary end point, 2.87 (95% CI, 1.71 to 4.84) for stroke, and 4.08 (95% CI, 2.88 to 5.77) for coronary heart disease (Table S8 and Figure S5).

Discussion

In this prospective cohort of 27,939 initially healthy American women initiated in 1992, a single bundled measure of hsCRP, LDL-C, and Lp(a) provided strong evidence of increased cardiovascular risk over a subsequent 30-year period. Each biomarker provided additive information to the others such that the combination of all three provided the greatest magnitude of spread for long-term risk stratification. Consistent with prior work in cohorts with 5-to-10-year follow-up data, 30-year risk was graded across quintiles for hsCRP and LDL-C but was evident for Lp(a) predominantly at the highest levels.

These data may have multiple implications for cardiovascular prevention. First, while traditional cardiovascular risk prediction models are based on 10-year risks, there has been considerable interest in the prediction of lifetime risk and in cost-effective methods to assess and intervene on risk over the lifespan (20). In this context, the current data demonstrates that a bundled assessment of three simple blood markers has predictive efficacy well beyond traditional 10-year estimates. Second, the observation that a single measure of hsCRP strongly predicted risk over a 30-year period should provide reassurance for clinicians who do not routinely measure this inflammatory biomarker due to concerns regarding variability over time (21). The finding here of at least similar long-term predictive value for hsCRP and LDL-C is consistent with direct comparisons of hyperlipidemia and inflammation in short-term studies inclusive of contemporary patients receiving guideline directed medical care as well as those who are statin intolerant (10,11). Third, the current data have implications for lifestyle and pharmacologic interventions designed to reduce cardiovascular risk. Prevention guidelines addressing diet, exercise, smoking cessation, and stress reduction consistently show greater benefit when behavioral interventions start earlier in life. While behavioral modifications can reduce hsCRP and LDL-C, they do not typically reduce Lp(a) which is largely genetically determined.

With respect to pharmacologic interventions, LDL-C reduction clearly lowers cardiovascular risk and is our most important pharmacologic tool for risk reduction beyond lifestyle change (1). However, as shown in our cumulative incidence curves, major cardiovascular events continued to accrue at substantial rates over time even though our cohort was comprised of individuals with quality access to care for whom prevalent statin use exceeded 50% at 30-years. Our data thus reinforce the continued broad need for LDL-C lowering both with statins and adjunctive lipid lowering agents.

Moreover, clinicians today have data from three randomized trials indicating that inflammation inhibition in addition to lipid lowering also reduces cardiovascular risk (2–4), and one anti-inflammatory agent, low-dose colchicine (0.5 mg daily), has been approved for secondary prevention and high-risk primary prevention by the US FDA (5). Other anti-inflammatory agents including interleukin-6 inhibitors are currently under investigation in large scale trials (22), and both SGLT2 inhibitors and GLP-1 receptor agonists lower hsCRP and lower vascular event rates. There are also several outcome trials that are ongoing with novel agents that substantially reduce Lp(a). As such, simultaneous assessment of LDL-C, hsCRP, and Lp(a) may assist clinicians in selecting the most appropriate pharmacologic agents for long-term atherosclerosis protection.

Finally, our joint effects and combined effects models provide 30-year prospective epidemiologic evidence that multiple pathways underlying atherosclerotic disease interact with each other to drive potentially catastrophic events. Thus, these data are consistent with the hypothesis that adjunctive interventions addressing a diverse set of biologic targets may ultimately be needed for optimal atherosclerotic protection.

Limitations of our study merit consideration. To increase the likelihood of long-term protocol adherence, the WHS was designed for efficiency in 1992 to include female health professionals. However, the proportion of non-Caucasian women in the WHS is 5.1%, lower than in cohorts of NIH-funded studies recruiting today. Second, while we focused attention on women for whom cardiovascular disease remains underdiagnosed and undertreated, long-term data in men are needed to generalize our findings. Third, our confidence interval widths have not been adjusted for multiplicity and thus should not be used in place of hypothesis testing and the hazard ratios may not be indicative of high predictive accuracy. Last, we do not have repeated biomarker measures. However, the fact that a cohort with access to care and high rates of statin prophylaxis nonetheless remains at substantial risk underscores the need for clinicians to consider adjunctive therapies as well as continued aggressive lipid lowering and behavioral interventions throughout the lifespan.

In summary, a single bundled measure of hsCRP, LDL-C, and Lp(a) among initially healthy American women predicts incident cardiovascular events over a 30-year period. Beyond implications for diagnostics, wellness interventions, and the selection of targeted therapy, these data strongly support efforts to extend strategies for the primary prevention of atherosclerotic events well beyond traditional 10-year estimates of risk.