Plasma Pentraxin 3 Levels Do Not Predict Coronary Events but Reflect Metabolic Disorders in Patients with Coronary Artery Disease in the CARE Trial

Chronic inflammation closely associates with obesity, metabolic syndrome, diabetes mellitus, and atherosclerosis. Evidence indicates that the immunomodulator pentraxin 3 (PTX3) may serve as a biomarker of these cardiometabolic disorders, but whether PTX3 predicts cardiovascular complications is unknown. We examined the association of plasma PTX3 levels with recurrent coronary events via a prospective, nested, case-control design in the CARE trial. Among 4159 patients who had a prior myocardial infarction 3 to 20 months before enrollment and also had total cholesterol levels <240 mg/dL and LDL cholesterol levels between 115 and 175 mg/dL, we measured plasma PTX3 levels at baseline by high-sensitivity ELISA in 413 cases with recurrent myocardial infarction or coronary death during a 5-year follow-up period, and in 366 sex- and age-matched controls. Cases with recurrent coronary events and controls had similar PTX3 levels, and PTX3 did not predict recurrent coronary events — a finding that contrasts with that of C-reactive protein (CRP) and serum amyloid A (SAA) in this cohort. We then associated PTX3 levels with metabolic disorders. Low plasma PTX3 levels correlated with high body-mass index, waist circumference, and triglycerides; and with low HDL cholesterol. Overall, PTX3 levels correlated inversely with the number of metabolic syndrome components. PTX3 levels also correlated inversely with apoCIII and tissue plasminogen activator, but did not associate with CRP. Although the study further links low PTX3 levels with various features associated with metabolic syndrome, the results do not indicate that PTX3 can predict recurrent coronary events among MI survivors.

Evidence suggests that PTX3 serves as a biomarker of inflammatory diseases, including atherosclerosis and metabolic disorders [18,19]. Yamasaki et al. reported that PTX3 levels were inversely associated with the features of metabolic syndrome in a healthy population [20]. But the intriguing possibility that PTX3 is a risk predictor of future cardiovascular complications (e.g., acute myocardial infarction) remains obscure. Other unresolved issues include which metabolic components mainly associate with plasma PTX3 levels; whether plasma PTX3 levels correlate with insulin resistance or diabetes mellitus; and the association of PTX3 with other inflammatory markers. The present study examined plasma PTX3 levels in participants of the Cholesterol and Recurrent Events (CARE) trial (UMIN-CTR; UMIN000012519) to test the hypothesis that plasma PTX3 levels predict future coronary events and associate with features typical of inflammatory cardiometabolic disorders.

Study population
We conducted a prospective, nested, case-control study within the CARE trial (See methods S1). This randomized, double-blind, placebo-controlled trial included 4159 patients with a prior myocardial infarction (MI) that occurred 3 to 20 month before enrollment. The subjects had low-density lipoprotein (LDL) cholesterol levels (total cholesterol levels ,240 mg/dL and LDL cholesterol levels between 115 and 175 mg/dL). Patients underwent random allocation to 40 mg pravastatin or placebo for a median of 5 years [21]. The institutional review board at each participating clinical center approved the study, and individuals gave written informed consent [22]. During the trial, 486 patients experienced the primary endpoint of coronary death or MI. Sufficient plasma for analysis was available from 413 of these cases. We randomly selected control subjects from patients who did not have a primary endpoint, matched to the cases by sex and decade of age. Measurement of PTX3 levels of these blood samples from the CARE trial and this analysis were approved by the committee for the Protection Against Research Risks of Brigham and Women's Hospital, Harvard medical school.

Blood collection and laboratory measurements
A fasting blood sample was taken from each patient at baseline. Serial measurements of PTX3 were performed using a highsensitivity ELISA (Perseus Proteomics Inc., Tokyo, Japan). Plasma lipid profiles, fasting blood glucose and insulin, serum creatinine levels, and tissue plasminogen activator (tPA) were measured as previously described. High-sensitivity assays for CRP and serum amyloid A (SAA) were performed [23].

Definition of metabolic syndrome
We defined metabolic syndrome using the modified 2005 National Cholesterol Education Program Adult Treatment Panel III (NCEP-ATP III) definition as 3 or more of the following: waist

Statistical analysis
For analysis comparing cases and controls, continuous variables are shown as means (standard deviation) or medians (inter quartile range), and the unpaired t test or the Wilcoxon unpaired rank sum test compared cases and controls. Proportions were compared using the x 2 test. Spearman correlations assessed the relation between continuous variables. We used unconditional logistic regression to calculate the odds ratio and 95% confidence interval (CI) of coronary heart disease (CHD). Tests for linear trend, for both the linear regression and logistic regression models, used the

Results
Plasma PTX3 levels associate with features typical of metabolic syndrome, but do not predict recurrent coronary events Table 1 presents the mean values for selected characteristics among the controls, according to quartiles of plasma PTX3 concentration. Participants with higher PTX3 concentration generally had lower prevalence of metabolic syndrome, as well as favorable levels of various components associated with metabolic syndrome -including lower BMI, waist circumference, triglycerides, apoCIII and tPA, and higher levels of high-density lipoprotein (HDL) cholesterol. Participants with higher PTX3 concentrations also had higher levels of the inflammatory marker SAA, but not of CRP. High PTX3 concentration was not associated with fasting glucose, insulin or the HOMA index. Compared with controls, cases had higher BMI and waist circumference, and more smoked cigarettes currently, had diabetes and hypertension, and took cardiovascular medications ( Table 2). The cases had higher levels of total cholesterol, LDL cholesterol, triglycerides, and plasma glucose, and lower HDL cholesterol levels than the controls. No significant difference in plasma PTX3 levels occurred between the cases and controls. PTX3 did not predict CHD risk (nonfatal MI and fatal CHD), even after adjustment for metabolic syndrome components, other coronary risk factors, and medication use ( Table 3).

PTX3 levels associate with components of metabolic syndrome
Individuals with metabolic syndrome had lower PTX3 levels (mean = 3.89 ng/mL, SE = 0.10, n = 397) than those without metabolic syndrome (4.22 ng/mL, SE = 0.10, P = 0.02, n = 382), after adjustment for age, sex, and smoking. Additionally, plasma PTX3 levels correlated inversely with variables associated with metabolic syndrome (Figure 1), including BMI, waist circumference, and triglycerides, while positively correlating with HDL cholesterol after adjustment for age. Plasma PTX3 remained statistically significant in association with BMI (P trend ,0.001) and waist circumference (P trend ,0.001) even after adjustment for age, sex, smoking, history of hypertension, triglycerides, HDL cholesterol, and insulin.

Lower PTX3 levels in obese, non-diabetic individuals
Obese patients had lower levels of adjusted mean PTX3 (mean = 3.77 ng/mL, SE = 0.13) than did non-obese patients (mean = 4.15 ng/mL, SE = 0.08, Figure 2A). Conversely, patients with diabetes (mean = 4.53 ng/mL, SE = 0.11) had higher adjusted mean PTX3 concentrations than did controls without diabetes (mean = 3.96 ng/mL, SE = 0.07, Figure 2A). To probe further the association of PTX3 and obesity, we stratified the control subjects by diabetic status. Compared to patients with a healthy weight (BMI ,25), both overweight patients (25# BMI ,30) and obese patients (BMI $30) had lower mean plasma PTX3 levels among those without diabetes (n = 655), after adjustment for confounders ( Figure 2B). Patients with diabetes (n = 124) had no significant difference in mean plasma PTX3 levels in different categories of BMI ( Figure 2B). Of the 124 patients with diabetes mellitus, 88 patients (71%) took medications for this condition.

PTX3 levels associate inversely with components of metabolic syndrome in all individuals and non-diabetic individuals
We further explored the association between PTX3, metabolic syndrome, and diabetes mellitus. PTX3 associated inversely with the total number of metabolic syndrome components in all individuals (n = 779) after adjustment for age, sex, and smoking ( Figure 2C). Non-diabetic patients (n = 655) showed a stronger inverse association between PTX3 and total number of metabolic syndrome components ( Figure 2C).

PTX3 levels do not correlate with CRP levels
PTX3 levels did not correlate with levels of the short pentraxin CRP ( Figure 3A), and very weakly and positively correlated with another inflammatory marker, SAA ( Figure 3B). CRP and SAA correlated with each other strongly and positively ( Figure 3C). After exclusion of participants with CRP above 10 mg/L (n = 687), PTX3 showed no significant correlations with CRP or SAA ( Figure 3D and 3E), whereas CRP and SAA correlated positively with each other (Figure 3F), suggesting that different mechanisms regulate levels of the long and short pentraxin in these MI survivors.

Discussion
This analysis from the CARE trial tested the hypothesis that plasma levels of the immunomodulator PTX3 predict risk of acute complications of coronary atherosclerosis in a population of MI survivors and also associated with cardiometabolic disorders. Although our results demonstrate that plasma PTX3 levels associate inversely with components of metabolic syndrome, the study does not support a relation between this pentraxin and recurrent CHD risk, unlike other pentraxin family members such as CRP.
Several clinical studies have reported elevated plasma PTX3 levels in patients following acute MI or unstable angina [5][6][7]. In patients with heart failure, elevated plasma PTX3 levels associated independently with cardiac events but did not specifically correlate with coronary events [8,14]. Other clinical studies indicated that PTX3 levels associate with cardiovascular disease and all-cause death among apparently healthy older individuals and patients with CHD [11,16]. These studies led to our intriguing hypothesis that PTX3 serves as a predictor of cardiovascular risk. In the population studied here with prior history of MI, however, plasma PTX3 levels did not predict subsequent events. The difference in the study population (e.g., primary vs. secondary prevention) may pertain to this discrepancy, although other risk factors including CRP have similar associations with cardiovascular disease in primary and secondary prevention. The pro-inflammatory or antiinflammatory role of PTX3 may also depend on the relative contribution of producing cell types, the disease context, and/or the stage of disease. Although PTX3 may associate with coronary events in other populations, the present study does not support its role as a risk predictor.
Obesity and metabolic syndrome represent chronic inflammatory states [33][34][35]. Whereas plasma levels of the short pentraxin CRP are higher in obese individuals and those with metabolic syndrome [36], evidence linking PTX3 with obesity and metabolic syndrome remains scant. The present study using MI survivors, and another clinical study in a healthy Japanese population, found that PTX3 is inversely correlated with obesity and metabolic syndrome [20]. In addition, patients with end-stage renal disease show inverse correlations of plasma PTX3 levels with BMI and waist circumference [15]. Another study reported decreased PTX3 mRNA expression in mature adipocytes derived from human adipose tissue, compared to that in preadipocytes [37]. These divergent results for CRP and PTX3 results may reflect different sources and biologic roles of the short and long pentraxins. Previous studies, including our own, have reported that various cell types associated with atherosclerosis and metabolic syndrome produce PTX3 in response to inflammatory signals, while the liver elaborates most circulating CRP [1,27,29,30,38,39]. HDL induces PTX3 mRNA expression and protein release in human endothelial cells through activation of the PI3K/Akt pathway [32], which may contribute to low plasma PTX3 levels in individuals with metabolic syndrome who have low HDL levels. PTX3 correlated inversely with triglycerides and with apoCIII, which interferes with clearance of triglyceride-rich lipoproteins from plasma [40]. We recently established that apoCIII induces cell activation and inflammation in vitro and in vivo [41]. Interaction between apoCIII and PTX3 requires further study. While several previous reports demonstrated significant positive associations between PTX3 and CRP [42], others showed no significant correlation [5,6]. PTX3 did not correlate significantly with CRP in our population. The distinct behaviors of PTX3 compared with those of CRP also suggest that long and short pentraxins have different relationships with the pathogenesis of obesity and metabolic syndrome.
In this study population, individuals with diabetes had higher plasma PTX3 levels than non-diabetic individuals although PTX3 did not correlate with fasting glucose, insulin or the HOMA-IR. More than 70% of diabetic patients took medications for diabetes mellitus, suggesting that these treatments affected the association between plasma PTX3 levels and diabetes mellitus in our study population. To clarify the effect of obesity on plasma PTX3 levels independent of diabetes mellitus, we divided study participants into diabetic and non-diabetic groups. Among individuals without diabetes, obese individuals had lower plasma PTX3 levels than lean individuals. Moreover, non-diabetic individuals showed a significant inverse correlation between PTX3 and the number of metabolic syndrome components.
This study has a few limitations. We did not measure PTX3 levels in patients who received pravastatin. The present study tested the specific hypothesis that plasma PTX3 levels predict risk of recurrent cardiovascular events in MI survivors. We thus measured PTX3 levels only at the baseline in case and control subjects. Our results indicated that PTX3 levels did not associate with future recurrent events in the CARE study cohort. As the present study did not to identify PTX3 as a biomarker of cardiovascular risk, we did not measure PTX3 levels during the follow-up period in control and pravastatin-treated subjects. Several studies have linked higher PTX3 levels with heart failure [8,14]. Because CARE did not adjudicate heart failure endpoints, we did not examine the association between PTX3 levels and heart failure.
The present study indicates that PTX3 does not predict future cardiac events, unlike CRP, although it associates with reduced levels of cardiometabolic risk factors. Various features associated with metabolic syndrome correlate negatively with PTX3 levels but positively with CRP. The underlying mechanisms may involve complex processes regulating PTX3 production in various cell types activated in the pro-inflammatory microenvironment of patients with cardiometabolic syndrome. Although the present study offers clinical evidence that further links PTX3 with metabolic syndrome, understanding the biology of this immunomodulator in cardiometabolic disorders requires additional multidisciplinary investigations.

Supporting Information
Methods S1 Supplemental data for detailed methods. Supplemental information of study population, blood collection and laboratory measurements, definition of the metabolic syndrome, and statistical analysis. (DOC)