Metalloproteinases have been implicated in the destabilization of atherosclerotic plaque, specifically, degrading the proteins that maintain the integrity of the protective fibrous cap (1- 2). Recognition of their biological proteolytic activity has led to the hypothesis that metalloproteinases might be markers of vulnerable or ruptured atherosclerotic plaque (2- 3). Pregnancy-associated plasma protein-A (PAPP-A) is a high molecular weight, zinc-binding metalloproteinase (1). Originally identified in pregnant women, in whom it is produced in the placenta (4), PAPP-A is also found at lower concentration in nonpregnant women and men. This enzyme has been shown to cleave insulin-like growth factor binding protein-4 from insulin-like growth factor-1, an important regulatory protein in cell proliferation and metabolism (5).

The blood concentration of PAPP-A is increased in patients with acute coronary syndromes (ACS), and PAPP-A is present in vulnerable coronary plaque but not in stable plaques (3,6). Two initial studies observed that PAPP-A concentration is associated with recurrent ischemic events in patients with suspected ACS, including those with undetectable troponin (3,7). However, subsequent results have been mixed, and most of these studies were performed in the setting of less sensitive previous-generation assays for cardiac troponin (8).

We investigated whether PAPP-A, measured using a highly sensitive assay developed to detect low concentrations in nonpregnant individuals, would be useful in conjunction with a contemporary sensitive assay for cardiac troponin for assessing the risk of recurrent ischemic events in patients presenting with non–ST-segment elevation acute coronary syndromes (NSTE-ACS). In addition, we evaluated the relationship of PAPP-A with other biomarkers representing potentially different pathobiological processes.

The MERLIN–TIMI 36 (Metabolic Efficiency With Ranolazine for Less Ischemia in Non-ST Elevation Acute Coronary Syndromes) trial enrolled 6,560 patients with NSTE-ACS and randomized them to treatment with either ranolazine or placebo in a 1:1 ratio and followed for a median of 348 days. The detailed entry criteria have been reported (9- 10) and are summarized in the Online Appendix. The MERLIN–TIMI 36 trial, including this substudy, was approved by the relevant institutional review boards at all participating centers. Written informed consent was obtained from all patients.

PAPP-A was measured in serum in batch (Active cPAPP-A ELISA, Beckman Coulter, Brea, California) by personnel blinded to treatment allocation and clinical events. Analytical sensitivity for the Active cPAPP-A assay is 0.18 μIU/ml, with total imprecision (cardiovascular [CV]) of 10% and 6% at concentrations of 2 and 11 μIU/ml, respectively. Troponin was measured using the AccuTnI (Beckman Coulter using the 99th percentile reference limit (0.04 μg/l). Details of sample collection and other biomarker testing are described in the Online Appendix.

The primary endpoint for this analysis was CV death or myocardial infarction (MI). All endpoints were adjudicated by an independent clinical endpoints committee, blinded to treatment allocation, according to definitions that were published previously (Online Appendix) (9- 10).

Baseline characteristics were compared using the chi-square test for categorical variables and the Wilcoxon rank sum test for continuous variables. Event rates presented are Kaplan-Meier failure rates at 30 days and 12 months. The assessment of the relationship between PAPP-A and outcome was performed using Cox regression. Interaction testing was performed for PAPP-A concentration and heparin administration based on previous studies demonstrating an interaction between heparin and PAPP-A concentration (11). Adjusted analyses included all elements of a well-validated risk model in NSTE-ACS (Thrombolysis In Myocardial Infarction risk score) (12). There was no interaction with the randomized therapy, and therefore treatment allocation was not included in the model.

A pilot study within a development set was performed to develop a prognostic cut point for validation and is described in the Online Appendix. The optimized cut point (6.0 μIU/ml) from the derivation set was then applied in the validation set as well as in the entire cohort. The increased discriminative value of PAPP-A in addition to clinical predictors, and troponin was examined with the method described by Harrell (13) to determine the net reclassification improvement (NRI) (13). NRI was calculated for comparison of the clinical model with cardiac troponin I (cTnI) alone and with PAPP-A using Harrell's technique, as programmed in R, which evaluates the change in estimated risk as a continuous variable and therefore is not dependent on a previous categorization (13). All analyses, unless otherwise specified, were performed using STATA version 9.2 (StataCorp, College Station, Texas).

PAPP-A was measured in all available baseline serum samples (N = 3,782). The baseline characteristics of the study population are listed in (Table 1). Patients with a baseline concentration of PAPP-A ≥6.0 μIU/ml (n = 644, 17%) were older, more frequently male, and more likely to have a creatinine clearance <60 ml/min; however, they were less likely to have diabetes mellitus, hypertension, or dysplipidemia.

Patients with elevated baseline levels were more likely to have had an index presentation of non–ST-segment elevation MI and to have a TIMI risk score of ≥4. Nevertheless, PAPP-A was only weakly correlated with the concentrations of cTnI (r = 0.17), B-type natriuretic peptide (BNP) (r = 0.093), myeloperoxidase (MPO) (r = 0.39), and high-sensitivity C-reactive protein (r = −0.13) (all p < 0.001). There was no significant difference in left ventricular ejection fraction (p = 0.68) or severity of coronary artery disease (p = 0.66) among patients with and without elevated PAPP-A.

The derivation cohort of 543 patients was randomly selected for cut-point derivation. Baseline characteristics and the distribution of PAPP-A were similar between the derivation and validation sets and are shown separately in (6), with the exception of troponin elevation. A cut point of 6.0 μIU/ml showed the strongest relationship with CV death or MI and was prospectively tested in the validation cohort (6). The prognostic performance of PAPP-A was highly similar in the derivation and validation cohorts (6), and therefore the results of the overall cohort are presented from this point forward.

In unadjusted analyses, patients with a baseline PAPP-A ≥6.0 μIU/ml had an approximately 2-fold higher risk of CV death or MI at 30 days compared with those with PAPP-A concentrations below the cut point (HR: 2.01; 95% CI: 1.43 to 2.82) (Table 2). The adverse CV risk associated with an elevated PAPP-A persisted through 12 months (HR: 1.63; 95% CI: 1.29 to 2.05) ((Figure 1) and Table 2). Moreover, patients with an elevated baseline PAPP-A had significantly higher rates individually of 12-month recurrent ischemic events (MI, p = 0.0049 and severe recurrent ischemia, p = 0.032), as well as CV death (p = 0.0028) compared with patients with a baseline PAPP-A <6.0 μIU/ml (Table 2). There was no significant relationship between PAPP-A and the risk of hospitalization for heart failure (Table 2).

The risk of CV death or MI associated with baseline PAPP-A concentration was not modified by troponin status as measured using a contemporary sensitive assay both at 30 days (p interaction = 0.93) and 12 months (p interaction = 0.87) (Figure 2). A similar relationship between PAPP-A and CV death or MI was seen among patients with elevated BNP ((6) and Figure 3).

When applied in conjunction with established clinical risk predictors in ACS (age, sex, ST-segment deviation, diabetes, smoking, hypertension, and history of coronary artery disease) as well as established biomarkers (cTnI), PAPP-A was independently associated with the risk of both 30-day CV death/MI (adjusted HR: 1.62; 95% CI: 1.15 to 2.29; p = 0.006) and 30 day CV death, MI, or severe recurrent ischemia (adjusted HR: 1.52; 95% CI: 1.16 to 1.98; p = 0.002) (Figure 4). Moreover, when other inflammatory biomarkers, MPO, and high-sensitivity C-reactive protein, and BNP were added to the model in addition to TnI, only PAPP-A remained a predictor of CV death or MI at 30 days (adjusted HR: 1.56; 95% CI: 1.09 to 2.24; p = 0.015) and 1 year (adjusted HR: 1.37; 95% CI: 1.07 to 1.75; p = 0.011) (Figure 5). There was no significant interaction between PAPP-A and heparin use for CV death/MI or CV death, MI, or severe recurrent ischemia at 30 days or 1 year (p > 0.65 for each interaction).

The incremental improvement in risk stratification beyond clinical risk predictors was evaluated using NRI as a measure of reclassification. The addition of PAPP-A dichotomized at 6.0 μIU/ml significantly improved reclassification for cardiovascular death (CVD)/MI (p = 0.003) and for CVD/MI/severe recurrent ischemia (p = 0.021) at 12 months. Furthermore, when cTnI was added to clinical predictors, the NRI remained significant for both CVD/MI (p = 0.0028) and for CVD/MI/severe recurrent ischemia (p = 0.021). In contrast to reclassification, there was no significant improvement in the c-statistic for CVD/MI by adding PAPP-A to clinical predictors and TnI (0.706 vs. 0.709, p = 0.23).

We found that an elevated PAPP-A concentration in patients with NSTE-ACS identified individuals at significantly higher short and long-term risk of adverse CV events when used independently and in conjunction with clinical characteristics and contemporary sensitive troponin and natriuretic peptide assays. Importantly, PAPP-A was predictive of the risk of recurrent MI rather than new or worsening heart failure. Consistent with the hypothesized pathobiology, these findings support PAPP-A specifically as a candidate marker of risk of recurrent atherothrombosis and differentiates PAPP-A from other novel markers such as natriuretic peptides and MPO, which are markers of all-cause mortality and heart failure but show variable association with recurrent ischemic events (14- 16).

Hospitalization for recurrent ischemic events occurs in as many as 1 in every 5 patients during the first year after NSTE-ACS (17). Although clinical models for the prediction of CV mortality after ACS have been developed and are used clinically, prediction of recurrent ischemic events with similarly high discrimination has been difficult (18), as evident by the significant decrement in the c-statistic of the TIMI risk score for patients with NSTE-ACS when predicting MI (0.66) versus mortality (0.74) (12) and for the Global Registry of Adverse Coronary Events score when predicting death or MI (0.70) compared with death alone (0.82) (19). Cardiac troponin is strongly and reproducibly associated with the risk of recurrent ischemic events after ACS. However, other newer established or emerging biomarkers have proven to have stronger associations with heart failure events (20). As we have shown previously in this population, BNP is an independent predictor of death and heart failure; however, it is not strongly predictive of recurrent ischemic events (21). In this context, our findings with PAPP-A are intriguing, both for its potential role as a clinical risk predictor and for therapeutic guidance.

Smaller previous studies have found mixed results regarding the association between PAPP-A and recurrent ischemic events independent of TnI or in patients with undetectable TnI (3,7). Lund et al. (7) found that increasing PAPP-A was an independent predictor of future ischemic events in 136 consecutive patients presenting to the emergency department with suspected ACS found to be cTnI negative. Heeschen et al. (22) found similar results in a larger cohort of 644 patients with chest pain. These studies, however, used insensitive TnI assays by current standards or cut points above those currently recommended, leading to recommendations for assessment using contemporary assays and cut points (8,23). Notably, our current study prospectively demonstrated a significant relationship between PAPP-A and CV death or recurrent ischemic events in 3,782 patients in conjunction with a contemporary sensitive assay for cTnI. Moreover, our large sample size adds substantially, more than doubling, the previous experience with PAPP-A for risk assessment and also permitted us to build a separate derivation set for development and internal validation of a prognostic decision limit.

The potential complementary nature of PAPP-A to other biomarkers that reflect distinct pathways underlying the development of recurrent events is further supported by the weak correlation between PAPP-A and markers of myonecrosis, hemodynamic stress, and even nonspecific markers of inflammation. Although each of these dimensions of performance that we observed with PAPP-A are interesting, further research is required to establish clear therapeutic ties (24).

Although the prognostic implications of increased PAPP-A appear reproducible and a plausible role in plaque destabilization exists, the precise pathways and contribution, if any, of PAPP-A relative to other metalloproteinases in plaque instability have yet to be definitively established. Our findings support additional mechanistic investigation of these pathways. Alternative interpretations of the association of PAPP-A and outcomes have also been proposed including a protective effect mediated through insulin-like growth factor-1 that is elicited in response to ischemic injury (6,25). Such an investigation is also likely to be valuable in elucidating the potential for PAPP-A as a therapeutic target (26). Previous work has shown that PAPP-A concentrations are not influenced by statin therapy, at least in stable patients with hypercholesterolemia (27), and the current analysis nested in MERLIN–TIMI 36 trial found no interaction with ranolazine.

This study was limited to patients selected for participation in a clinical trial who presented with suspected ACS and had samples available for analysis. As such, these results do not address the question of prognostic significance in the broader group of patients presenting with nontraumatic chest pain, including those patients with symptoms atypical of myocardial ischemia or presenting with other acute illnesses that may be inflammatory in nature. Although samples were handled using standardized procedures, it is possible that pre-analytical issues (i.e., storage) could have affected the stability of PAPP-A. Any such effect would be expected to have led us to underestimate the magnitude of the risk relationship. In addition, external validation of our proposed cut point in separate datasets would be valuable. Finally, we used an available contemporary sensitive troponin assay in this investigation, and different results might be observed if repeated using a third-generation highly sensitive assay.

PAPP-A was independently associated with the short- and long-term risk of CV death and recurrent ischemic events in patients with NSTE-ACS, along with clinical predictors and cTnI. The relationship with severe recurrent ischemic events differentiates PAPP-A from other novel markers that reliably predict only all-cause mortality and heart failure. Our findings add to emerging evidence supporting the metalloproteinase PAPP-A as a candidate prognostic marker of recurrent ischemia and CV death in patients with ACS and supports continued investigation of its potential therapeutic implications.

For supplemental material, please see the online version of this article.