Association of Initial and Serial C-Reactive Protein Levels With Adverse Cardiovascular Events and Death After Acute Coronary Syndrome: A Secondary Analysis of the VISTA-16 Trial

Key Points

Question  Are initial and serial increases in high-sensitivity C-reactive protein levels after acute coronary syndrome in medically optimized patients associated with increased risk of a major cardiac event, cardiovascular death, and all-cause death?

Findings  In this secondary analysis of the VISTA-16 randomized clinical trial that included 5145 patients, baseline and longitudinal high-sensitivity C-reactive protein levels were independently associated with increased risk of a major adverse cardiac event, cardiovascular death, and all-cause death during the 16-week follow-up.

Meaning  Monitoring high-sensitivity C-reactive protein levels in patients after acute coronary syndrome may help better identify patients at greater risk for recurrent cardiovascular events or death.

Importance  Higher baseline high-sensitivity C-reactive protein (hsCRP) levels after an acute coronary syndrome (ACS) are associated with adverse cardiovascular outcomes. The usefulness of serial hsCRP measurements for risk stratifying patients after ACS is not well characterized.

Objective  To assess whether longitudinal increases in hsCRP measurements during the 16 weeks after ACS are independently associated with a greater risk of a major adverse cardiac event (MACE), all-cause death, and cardiovascular death.

Design, Setting, and Participants  Secondary analysis of the double-blind, multicenter, randomized clinical Vascular Inflammation Suppression to Treat Acute Coronary Syndromes for 16 Weeks (VISTA-16) trial conducted between June 1, 2010, and March 7, 2012 (study termination on March 9, 2012), which included 5145 patients from 362 academic and community hospitals in Europe, Australia, New Zealand, India, and North America assigned to receive varespladib or placebo on a background of atorvastatin treatment beginning within 96 hours of presentation with an ACS. The present study evaluated data from patients with available baseline and longitudinal hsCRP levels measured at weeks 1, 2, 4, 8, and 16 after randomization to treatment or placebo. Statistical analysis was performed from June 15, 2018, through September 15, 2018.

Main Outcomes and Measures  Outcomes were MACE (composite of cardiovascular death, myocardial infarction, nonfatal stroke, or unstable angina with documented ischemia requiring hospitalization), cardiovascular death, and all-cause death after adjustment for baseline clinical, treatment, and laboratory characteristics, including baseline hsCRP levels.

Results  Among 4257 patients in this study, 3141 (73.8%) were men and the mean age was 60.3 years (interquartile range [IQR], 53.5-67.8 years). The median 16-week low-density lipoprotein cholesterol level was 64.9 mg/dL (IQR, 50.3-82.3 mg/dL), and the median hsCRP level was 2.4 mg/L (IQR, 1.1-5.2 mg/L). On multivariable analysis, higher baseline hsCRP level (hazard ratio [HR], 1.36 [95% CI, 1.13-1.63]; P = .001) and higher longitudinal hsCRP level (HR, 1.15 [95% CI, 1.09-1.21]; P < .001) were independently associated with MACE. Similar significant and independent associations were shown between baseline and longitudinal hsCRP levels and cardiovascular death (baseline: HR, 1.61 per SD [95% CI, 1.07-2.41], P = .02; longitudinal: HR, 1.26 per SD [95% CI, 1.19-1.34], P < .001) and between baseline and longitudinal hsCRP levels and all-cause death (baseline: HR, 1.58 per SD [95% CI, 1.07-2.35], P = .02; longitudinal: HR, 1.25 per SD [95% CI, 1.18-1.32], P < .001).

Conclusions and Relevance  Initial and subsequent increases in hsCRP levels during 16 weeks after ACS were associated with a greater risk of the combined MACE end point, cardiovascular death, and all-cause death despite established background therapies. Serial measurements of hsCRP during clinical follow-up after ACS may help to identify patients at higher risk for mortality and morbidity.

Residual risk of ischemic cardiovascular events or death remains high following an acute coronary syndrome (ACS) despite optimal guideline-directed coronary revascularization and treatment with antiplatelet and low-density lipoprotein cholesterol lowering agents.^1,2 Identifying biomarkers associated with this residual risk may help guide interventions targeted to lower these levels.

C-reactive protein (CRP) is an inflammatory biomarker that is independently associated with future cardiovascular events across all levels of low-density lipoprotein cholesterol.^3,4 High-sensitivity CRP (hsCRP) has been shown to be a better indicator of outcomes than CRP levels measured through traditional assays.⁵ The effectiveness of statins in decreasing cardiovascular events has been shown to correlate in part with CRP levels during treatment.^6,7 Patients with chronic atherosclerotic cardiovascular disease and elevated hsCRP levels have been shown to benefit from targeted antiinflammatory therapies, and those whose hsCRP level decreases from baseline with such treatment have been shown to have a more favorable prognosis than those with hsCRP levels that remain elevated.⁸ Increasing hsCRP levels measured early after ACS⁹ have been shown to be associated with short-term (including in-hospital) adverse outcomes irrespective of concurrent troponin levels.¹⁰ Despite these observations, it remains unclear whether the subsequent temporal change in hsCRP levels yields additional prognostic information. We assessed serial hsCRP levels for 16 weeks in optimally treated patients with ACS to determine whether baseline and serial hsCRP levels were associated with recurrent cardiovascular events and/or death.

The Vascular Inflammation Suppression to Treat Acute Coronary Syndromes for 16 Weeks (VISTA-16) trial was a double-blind, randomized, multicenter trial of the secretory phospholipase A₂ inhibitor varespladib conducted between June 1, 2010, and March 7, 2012 (study termination on March 9, 2012) (NCT01130246). Study details have been previously described.¹¹ In brief, 5145 patients were randomized to receive varespladib (n = 2572) or placebo (n = 2573) within 96 hours of presentation with an ACS. Documentation of ACS required either (1) biomarker level elevation accompanied by symptoms of acute myocardial ischemia and/or new or presumed new ischemic electrocardiographic abnormalities or (2) symptoms in combination with new or presumed new electrocardiographic changes in patients without elevated biomarker levels. In VISTA-16, 85% of ACS events were biomarker positive for cardiac necrosis. Patients were excluded if low-density lipoprotein cholesterol levels were not at goal (per local guidelines) despite maximal statin therapy or if they had advanced heart failure, hemoglobin A_1c values greater than 11%, malignancy, advanced liver or renal disease, statin intolerance, or hypertriglyceridemia. Treatment was continued daily for 16 weeks in addition to standard background therapy. Written informed consent was provided for the current analysis as part of the original clinical trial consent. Institutional review board approval was obtained from all centers participating in the VISTA-16 trial.

The primary end point was a major adverse cardiac event (MACE), the composite of cardiovascular death, nonfatal myocardial infarction, stroke, or hospitalization for unstable angina at 16 weeks. Secondary end points included individual components of the primary composite end point; the composite of all-cause death, nonfatal myocardial infarction or stroke, or hospitalization for unstable angina; and all-cause death. The trial was terminated early at a prespecified interim analysis because of futility and possible harm in the varespladib group. The primary composite end point was found to be more frequent in the varespladib group compared with the placebo group, with similar observations noted for the secondary end points of cardiovascular death, nonfatal myocardial infarction, and nonfatal stroke.

The hsCRP levels were measured at time of randomization and at 1, 2, 4, 8, and 16 weeks thereafter in 4257 of 5145 patients (82.7%). In this secondary analysis, we assessed the association of increasing hsCRP levels with the 16-week incidence of the primary and secondary end points after adjustment for baseline clinical, laboratory, and treatment characteristics.

Statistical analysis was performed from June 15, 2018, through September 15, 2018. Continuous variables were reported as mean (SD) if normally distributed and as median (interquartile range [IQR]) if nonnormally distributed. Categorical variables were recorded as frequency (percentage).

Given that hsCRP was a time-dependent repeated measures covariate, the counting process method was applied.¹² The trajectory of hsCRP levels was defined as the dynamic longitudinal hsCRP levels, and the association of these levels with clinical events was explored using the survival model with the counting process style data. When formulating the counting process style as the input data format for the survival model, the dynamic changes for a patient were identified using a series of time intervals (eg, T1 or T2) for hsCRP level measurement, and 2 elements were defined for each individual: a dichotomous variable that defined the censoring status at time T2 (coded 1 for event and 0 for censoring) and a set of covariates (including the time-dependent covariates) for the patient during that interval. For a specific interval, the values of the time-dependent covariates remained unchanged and the patient remained at risk, and in the next interval, the values of the time-dependent covariates could change and the patient remained at risk; in the last interval, the values of the time-dependent covariates could change and the patient either experienced an event or was censored. The model for this method is shown in the eMethods in the Supplement.

The association of longitudinal hsCRP levels with various cardiovascular outcomes was examined using a Cox proportional hazards regression model and was reported as a hazard ratio (HR) with 95% CI and P value. Stepwise model selection (with the 0.20 entry criterion and 0.05 retention criterion) was used to create a covariate set (longitudinal hsCRP level, baseline hsCRP level, demographic characteristics, medical history, and drug treatment), and a multivariable Cox proportional hazards regression model that adjusted for the set of baseline clinical characteristics was implemented to further investigate the association of longitudinal hsCRP level with MACE, all-cause mortality, and cardiovascular mortality. As an additional approach, a joint model for the survival data and the longitudinal hsCRP data was created using the NLMIXED procedure in SAS, version 9.4 (SAS Institute, Inc), in which the 95% CI of the correlation coefficient was calculated using the Fisher transformation.

To better understand longitudinal hsCRP, baseline covariates including log-transformed baseline hsCRP measures were examined for their associations with the longitudinal hsCRP levels, each in a repeated-measures linear mixed model. Those with a P < .10 were selected into a multivariable repeated-measures linear mixed model to identify risk factors of longitudinal hsCRP levels. In these mixed model repeated-measures analyses, the patients were set as being nested in the treatment arms and the unstructured covariance structure was applied.

In addition to the modeling completed using longitudinal hsCRP measurements, additional analyses were performed after dichotomizing study participants as having increased or not increased hsCRP levels based on time-weighted average levels. Intraclass correlations were calculated for hsCRP level, lipid levels, and blood pressure to track and compare longitudinal changes (eMethods in the Supplement). A 2-sided P < .05 was considered to be statistically significant. Analyses were performed using SAS, version 9.4 (SAS Institute, Inc).

Of the 5135 patients, 4309 patients had both baseline and follow-up hsCRP levels measured (eTable 1 in the Supplement); 52 of these patients were excluded because the first follow-up hsCRP level was obtained after their MACE, leaving 4257 patients for the analyses. There were originally 247 events observed during the study, with 200 events occurring among patients with at least 1 follow-up hsCRP measurement. Among these patients, 40 had time to event before week 1 (ie, ≤7 days). Three additional events were excluded because the patients did not have baseline hsCRP measurements. The remaining 145 events were studied.

Table 1 summarizes baseline demographic and clinical characteristics of all patients included in the study. Patients had a mean age of 60.3 years (IQR, 53.5-67.8 years), and 3141 of 4257 patients (73.8%) were men. Body mass index (calculated as weight in kilograms divided by height in meters squared) was in the overweight range (median, 29.3 [IQR, 26.3-32.5]); 3249 patients had hypertension (77.1%), 2127 had hypercholesterolemia (50.5%), and 2674 had metabolic syndrome (64.8%) (total number of patients for each variable differed because of missing data). Previous myocardial infarction was present in 1268 patients (30.0%), 763 (18.0%) had previously undergone percutaneous coronary intervention, and 279 (6.6%) had coronary artery bypass surgery with 1520 patients (35.8%) receiving lipid-modifying therapy before their index ACS event. At the time of randomization, most patients were receiving optimal medical therapies, including dual antiplatelet therapy, β-blockers, angiotension-converting enzyme inhibitors or angiotensin receptor blockers, and statins; 2025 patients (47.6%) presented with ST-segment elevation myocardial infarction. Low-density lipoprotein cholesterol levels and non–high-density lipoprotein cholesterol levels decreased substantially compared with baseline levels in the overall cohort. In the overall cohort, median baseline hsCRP level was 10.5 mg/L (IQR, 4.2-30.3 mg/L) (to convert hsCRP to nanomoles per liter, multiply by 9.524). The mean level of hsCRP change in the varespladib group was –20.73 mg/L (95% CI, –22.25 to –19.22 mg/L) and in the placebo group was –19.72 mg/L (95% CI, –21.22 to –18.22) (P = .35).

Analyses using time-weighted average hsCRP levels are shown in the eResults, eTable2, and eFigures 1 and 2 in the Supplement. Intraclass modeling is shown in eTable3 in the Supplement.

Table 2 shows an analysis of cardiovascular outcomes based on longitudinal hsCRP levels, adjusted for baseline hsCRP level. All-cause death (HR, 1.25; 95% CI, 1.19-1.32; P < .001), cardiovascular death (HR, 1.26; 95% CI, 1.20-1.32; P < .001), myocardial infarction (HR, 1.16; 95% CI, 1.08-1.25; P < .001), and several composite end points were significantly higher in patients with longitudinal increases in hsCRP level.

Table 3 shows multivariable Cox proportional hazards regression models for the combined MACE end point, all-cause death, and cardiovascular death. Baseline (HR, 1.36; 95% CI, 1.13-1.63; P = .001) and longitudinal hsCRP (HR, 1.15; 95% CI, 1.09-1.21; P < .001) levels were independently associated with MACE after adjustment for drug treatment. Baseline hsCRP level was independently associated with all-cause death (HR, 1.58 per SD; 95% CI, 1.07-2.35; P = .02) and cardiovascular death (HR, 1.61 per SD; 95% CI, 1.07-2.41; P = .02). Similarly, significant associations were noted between longitudinal hsCRP level and all-cause death (HR, 1.25 per SD; 95% CI, 1.18-1.32; P < .001) and cardiovascular death (HR, 1.26 per SD; 95% CI, 1.19-1.34; P < .001). A joint model of the survival data and the longitudinal hsCRP measurement data showed that longitudinal hsCRP level was significantly correlated with MACE (r, 0.053; 95% CI, 0.023-0.083).

Table 4 shows a multivariable analysis of factors associated with changes in longitudinal hsCRP levels. Baseline hsCRP level, age, body mass index, hypertension, congestive heart failure, and active smoking had significant positive associations with longitudinal hsCRP level, whereas male sex, baseline antiplatelet use, baseline high-density lipoprotein cholesterol level, varespladib treatment, and high-intensity statin therapy use had significant negative associations with longitudinal hsCRP levels.

The present study supports previous evidence that hsCRP levels measured at the time of ACS are associated with future events and provides new data suggesting that an increase in hsCRP levels after ACS is associated with risk for MACE, cardiovascular death, and all-cause death. Each SD increment in longitudinal hsCRP concentration was associated with a 15% increased risk of MACE, 25% increased risk of all-cause death, and 26% increased risk of cardiovascular death. The associations between longitudinal hsCRP levels and adverse outcomes were identified with adjustment for baseline hsCRP level and assigned treatment in the VISTA-16 trial, as well as a history of intensive treatment with statins and other evidence-based treatments for ACS. A better understanding of the implications of serial hsCRP level measurements after an ACS may help to further improve risk stratification and attenuate residual cardiovascular risk after ACS.

Although the clinical implications of serial hsCRP level measurements in patients after ACS have not been previously appreciated to our knowledge, the expected time course of hsCRP levels after an ACS event have been characterized. C-reactive protein levels increase after an ACS event and peak approximately 50 hours after the onset of pain.¹³ A significant correlation between peak hsCRP levels and peak creatine kinase–MB fraction levels has been reported,¹³ and serum concentrations of baseline hsCRP levels have been shown to be associated with risk of mortality independent of troponin levels.¹⁰ In uncomplicated cases, normalization of hsCRP levels occurs in approximately 1 week.¹³

Medications used to treat patients after ACS or patients at high risk for cardiovascular events can modulate hsCRP levels. High-dose statins can accelerate the decrease in hsCRP levels after ACS, with moderate- or high-dose statins reducing hsCRP levels at discharge (62% vs 11% with placebo), 30 days (84% vs 30% with placebo), and 16 weeks (83% vs 74% with placebo).^14,15 These decreases in hsCRP levels, in conjunction with lowering of low-density lipoprotein cholesterol levels, have been shown to be associated with reduced risk of death or major cardiovascular events.² The present analysis showed that use of antiplatelet agents (clopidogrel, ticlopidine hydrochloride, or prasugrel) was associated with stable or decreasing hsCRP levels. Aspirin has been shown to possess antiinflammatory properties in vitro; however, it has been shown that aspirin does not reduce CRP levels in healthy patients.¹⁶ In contrast, aspirin has been shown to reduce CRP levels in patients with chronic stable angina¹⁷ and acute myocardial infarction.¹⁸ For patients undergoing percutaneous coronary intervention who received clopidogrel in addition to aspirin, a 65% reduction in the mean CRP level increase was noted at 24 hours after the procedure compared with those who received aspirin alone.¹⁹

The Canakinumab Antiinflammatory Thrombosis Outcome Study trial compared effects of targeted antiinflammatory therapy with canakinumab, a monoclonal antibody targeting IL-1β, with placebo in patients with previous myocardial infarction and elevated hsCRP levels and showed a reduced number of MACEs with canakinumab administration.⁸ Furthermore, patients who achieved hsCRP levels less than 2 mg/L had significant reductions in MACEs, cardiovascular mortality, and all-cause mortality, whereas patients whose hsCRP levels remained greater than 2 mg/L during treatment had no significant reductions in these end points.²⁰ These findings suggest that changes in hsCRP levels are associated with cardiovascular outcomes in chronic coronary heart disease, although it remains to be determined whether hsCRP level represents a surrogate marker for residual inflammation after ACS or whether it should be considered as a direct biologic target. The present analysis extended the observations in the Canakinumab Antiinflammatory Thrombosis Outcome Study trial, with similar conclusions in the intermediate term after ACS. Our findings suggest that, in addition to targeting cholesterol measurements to reduce residual risk in patients after ACS, modulating residual inflammation may be an important factor associated with decreasing morbidity and mortality after ACS.

Epidemiologic studies^21,22 in initially healthy populations have shown that serial changes in CRP level may be associated with cardiovascular morbidity and death. In a study of 5811 urban residents,²¹ changes from normal (≤3 mg/L) to elevated (>3 mg/L) levels of CRP within 1 year were associated with an HR of 6.7 (95% CI, 2.42-18.59) for all-cause mortality during the subsequent 4 years, compared with CRP levels that remained normal. In the Atherosclerosis Risk in Communities study,²² more than 10 000 patients had 2 hsCRP measurements within a period of 6 to 8 years. Those with increasing hsCRP levels had an increased risk of incident coronary heart disease (HR, 1.43), fatal coronary heart disease (HR, 1.98), and mortality (HR, 1.34) after adjustment for traditional risk factors. These studies suggest, that even in chronically stable, relatively well individuals, serial measurements showing large increases in levels of CRP may have a significant adverse prognostic value and should be considered in the clinical setting. The results of our study extend these findings to patients in the short-term period after ACS.

Although it remains unclear whether CRP level is a marker or a participant in the atherothrombotic process, identification of increasing CRP levels may help determine which patients may benefit from more intensive treatment. From a mechanistic perspective, CRP has been shown to have a reciprocal relationship with both low-density lipoprotein cholesterol levels and high-density lipoprotein cholesterol levels. In addition, CRP-plaque interactions have been shown to promote a prothrombotic milieu by influencing many steps in the pathogenesis of atherosclerosis.^23,24 Further elucidating the pathogenic mechanisms underlying the associations between inflammation, lipid levels, and atherosclerosis is important because they may help explain why increasing CRP levels associate with greater mortality.

Several limitations of the present analysis warrant consideration. The follow-up period after ACS was only 16 weeks, and the association of increasing hsCRP levels early after an ACS with longer-term outcomes remain to be determined. Varespladib treatment was found to be potentially harmful in the VISTA-16 trial. Therefore, the association of increasing hsCRP levels with adverse outcomes in the entire study cohort might have been a surrogate, at least to some extent, for an association of varespladib treatment with adverse outcomes. We performed sensitivity analyses limited to the placebo group that confirmed the adverse association of increasing hsCRP levels with MACE and death. Limitations in our time-weighted analyses are presented in the eDiscussion in the Supplement.

The present study suggests that the initial and subsequent levels of hsCRP after ACS are associated with the risk of recurrent MACE and death. These associations were identified despite the use of optimal evidence-based medical therapies. Further studies will be required to determine whether initial and serial hsCRP measurements can help guide the use of targeted antiinflammatory therapies after ACS to help further reduce residual cardiovascular risk in this vulnerable population.

Accepted for Publication: January 11, 2019.

Corresponding Author: Rishi Puri, MBBS, PhD, Department of Cardiovascular Medicine and Cleveland Clinic Coordinating Center for Clinical Research, Cleveland Clinic, 9500 Euclid Ave, Cleveland, OH 44195 (purir@ccf.org).

Published Online: March 6, 2019. doi:10.1001/jamacardio.2019.0179

Author Contributions: Drs Mani and Puri contributed equally to this manuscript. Dr Nicholls had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Mani, Puri, Kastelelein, Nicholls.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript:Mani, Puri, Menon.

Critical revision of the manuscript for important intellectual content: Mani, Puri, Schwartz, Nissen, Shao, Kastelein, Lincoff, Nicholls.

Statistical analysis: Mani, Shao, Nicholls.

Obtained funding: Nicholls.

Administrative, technical, or material support: Nissen, Kastelein, Menon, Lincoff, Nicholls.

Supervision: Puri, Nicholls.

Conflict of Interest Disclosures: Dr Schwartz reported receiving grants from Anthera during the conduct of the study and grants from Cerenis, Resverlogix, Roche, Sanofi, and The Medicines Company outside the submitted work. Dr Nissen reported grants from Anthera during the conduct of the study. Dr Kastelein reported receiving grants from Anthera during the conduct of the study and personal fees from AstraZeneca, Staten Biotech, Regeneron, Amgen, CSL-Behring, Esperion, Gemphire, and The Medicines Company. Dr Lincoff reported receiving personal fees from Novartis outside the submitted work. Dr Nicholls reported receiving grants from Anthera during the conduct of the study; grants from AstraZeneca, Amgen, Anthera, Eli Lilly and Company, Novartis, Cerenis, The Medicines Company, Resverlogix, InfraReDx, Roche, Sanofi-Regeneron, and LipoScience; and personal fees from Akcea, AstraZeneca, Eli Lilly and Company, Anthera, Omthera, Merck & Co., Takeda, Resverlogix, Sanofi-Regeneron, CSL Behring, Esperion, and Boehringer Ingelheim outside the submitted work. No other disclosures were reported.

Funding/Support: The Vascular Inflammation Suppression to Treat Acute Coronary Syndromes for 16 Weeks trial and the secondary analysis was funded by Anthera Pharmaceuticals.

Role of the Funder/Sponsor: Anthera Pharmaceuticals had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.