Prognostic Value of Placental Growth Factor in Patients With Acute Chest Pain FREE

Recent work has established a fundamental role for inflammation in mediating initiation and progression of atherosclerosis, as well as its acute thrombotic complications. Elevated levels of inflammatory markers such as high-sensitivity C-reactive protein (hsCRP), serum amyloid A, and interleukin 6 not only commonly accompany acute coronary syndromes (ACS),^1- 3 but also appear to predict risk of adverse outcomes.^2,4- 7 Although the classic acute-phase protein hsCRP emerged as the most promising inflammatory biomarker for clinical purposes, there is substantial heterogeneity with respect to the prevalence of elevated hsCRP levels.⁸ More than 30% of patients with severe unstable angina do not present with elevated hsCRP levels.^2,5 Moreover, individual differences in the degree of response to given inflammatory stimuli might affect the levels of downstream acute-phase reactants such as hsCRP.^9- 10 Thus, there may be value for identifying proximal stimuli of vascular inflammation for use as predictive biomarkers in patients with suspected ACS.

Placental growth factor (PlGF), a member of the vascular endothelial growth factor (VEGF) family of growth factors, was recently shown to be up-regulated in early and advanced atherosclerotic lesions.¹¹ Originally identified in the placenta,¹² PlGF stimulates vascular smooth muscle cell growth, recruits macrophages into atherosclerotic lesions, up-regulates production of tumor necrosis factor α and monocyte chemotactic protein 1 by macrophages, and stimulates pathological angiogenesis.^11,13 Inhibition of PlGF effects by the blocking of its receptor, Fms-like tyrosine kinase, in an animal model suppressed both atherosclerotic plaque growth and vulnerability via inhibition of inflammatory cell infiltration.¹¹ These data suggest that PlGF may act as a primary inflammatory instigator of atherosclerotic plaque instability. Accordingly, we aimed to characterize the potential prognostic value of PlGF in addition to the well-established biomarkers troponin T (TnT), soluble CD40 ligand (sCD40L), and hsCRP in patients with unstable coronary heart disease. We investigated patients with angiographically documented ACS enrolled in the CAPTURE (c7E3 Fab Anti-Platelet Therapy in Unstable Refractory Angina) trial,¹⁴ as well as a prospective cohort of patients presenting with acute chest pain to an emergency department.

The European multicenter CAPTURE trial included patients with recurrent resting chest pain associated with electrocardiographic changes during treatment with intravenous heparin and nitroglycerin.¹⁴ All patients underwent prerandomization coronary angiography and had a culprit lesion measuring 70% or larger suitable for percutaneous transluminal coronary angioplasty (PTCA). Heparin was administered from before randomization until at least 1 hour after the PTCA procedure, which was scheduled between 18 and 24 hours after beginning study treatment. Patients were randomly assigned to receive either the glycoprotein IIb/IIIa receptor antagonist abciximab or placebo. Because other markers such as TnT^15- 16 and sCD40L¹⁷ have been shown to interact with the treatment effect of abciximab, the present analysis was restricted to patients receiving placebo and having available blood samples (n = 547; 86% of patients receiving placebo). Blood samples were collected a mean of 8.7 (SD, 4.9) hours after onset of symptoms but prior to PTCA. Each center in the CAPTURE trial received institutional review board approval for the blood drawings, and all patients provided written informed consent.

We separately analyzed a heterogeneous group of 626 consecutive patients with acute chest pain (161 women and 465 men; mean age, 61 [range, 38-82] years) presenting between December 1996 and March 1999 to the emergency department at the University of Hamburg, Hamburg, Germany, with acute chest pain lasting less than 12 hours. Patients with characteristic ST-segment elevations were excluded. The presence of coronary artery disease was documented by 1 of the following criteria: electrocardiographic evidence of myocardial ischemia (new ST-segment depression or T-wave inversion), a history of coronary heart disease (myocardial infarction or coronary revascularization, a positive exercise stress test, or narrowing of at least 50% of the luminal diameter of a major coronary artery on a current angiogram). Patients without coronary heart disease had to have normal coronary angiogram results. Blood samples were collected at the time of arrival in the emergency department (a mean of 5.1 [SD, 3.4] hours after onset of symptoms but prior to initiation of anticoagulation and antiplatelet treatment) and a second blood sample was drawn 4 hours later. The chest pain study was approved by the ethics committee of the General Medical Council, Hamburg. All patients provided written informed consent.

Measurement of cardiac marker levels was performed blinded to the patients' histories. Levels of PlGF, VEGF, and sCD40L were measured by enzyme-linked immunosorbent assay (R&D Systems, Wiesbaden, Germany). Total imprecision (expressed as coefficient of variation) for PlGF was 7.3%. For sCD40L, a diagnostic threshold level of 5.0 µg/L was used.¹⁷ Levels of TnT were determined using a 1-step enzyme immunoassay based on electrochemiluminescence technology (Roche Diagnostics, Mannheim, Germany). A diagnostic threshold value of 0.01 µ/L was used.¹⁸ Levels of hsCRP were measured using the Behring BN II Nephelometer (Dade Behring, Deerfield, Ill). A diagnostic threshold value of 10 mg/L was used.^5,19

All results for continuous variables are expressed as mean (SD). Comparisons between groups were analyzed by Mann-Whitney U test. Comparison of categorical variables was generated by the Pearson χ² test. The primary end points were mortality or nonfatal myocardial infarction during 30 days of follow-up. Secondary end points for the CAPTURE trial were mortality or nonfatal myocardial infarction up to 6 months.¹⁴ The distribution of hsCRP and PlGF blood levels was nonnormal, but logarithmic transformation resulted in normal distribution of the values. The Cox proportional hazards regression model was used to estimate the relative risk for cardiovascular events in relation to the logarithmically transformed variables.²⁰ The assumption that the hazards are proportional over time was validated by defining a time-dependent covariate as a function of time and PlGF level. Analysis of quintiles of PlGF levels based on nontransformed values was performed using the Cox proportional hazards regression model with the lowest quintile serving as the reference group. Analysis of the receiver operating characteristics (ROC) curve over the dynamic range of the PlGF assay was used to identify the threshold level for PlGF providing highest predictive value. The effect of baseline characteristics (with P = .10 necessary to enter a variable into the model) and other biochemical markers on any observed associations between PlGF levels and cardiovascular events was analyzed using stepwise multivariable Cox proportional hazards models adjusted for age, sex, history of diabetes, ST-segment changes, and the measured biochemical markers. All included variables were significant predictors in a univariate model. Reported P values are 2-sided; P<.05 was used to determine statistical significance. All statistical analyses were performed with SPSS version 11.5 (SPSS Inc, Chicago, Ill).

In baseline blood samples of CAPTURE patients receiving placebo treatment, PlGF was detectable in 95.6% of the samples (mean [SD] for all patients: 27.95 [9.83] ng/L [range, 7.0–181.29 ng/L]). During 30 days of follow-up, there were 49 events, including 10 deaths (9.0% event rate). After logarithmic transformation, levels of both hsCRP (hazard ratio [HR], 1.73; 95% confidence interval [CI], 1.20-2.49; P = .003) and PlGF (HR, 9.29; 95% CI, 4.39-19.67; P<.001) were predictive of events. In a multivariable analysis, however, PlGF level was the more powerful predictor of the patients' outcome (HR, 9.19; 95% CI, 3.91-21.59; P<.001), whereas hsCRP level no longer independently predicted patients' outcome (HR, 1.01; 95% CI, 0.68-1.50; P = .97). Analysis of ROC curves revealed that the association of PlGF level with the outcome of the patients was significantly stronger than for hsCRP level, with areas under the curve of 0.72 (95% CI, 0.54-0.81) for PlGF and 0.57 (95% CI, 0.40-0.65) for hsCRP (P = .005). Analysis of ROC curves verified a threshold PlGF level of 27.0 ng/L for maximized predictive value. Based on this threshold value, 223 patients (40.8%) had PlGF blood levels greater than 27.0 ng/L and 324 patients had PlGF blood levels of 27.0 ng/L or less. As illustrated in Table 1, there were few significant differences in the baseline characteristics between the 2 groups. Patients with elevated PlGF levels were more likely to have diabetes or hypertension, respectively, and had significantly higher hsCRP levels (Table 1).

Patients were stratified into quintiles according to measured nontransformed PlGF levels: 13.3 ng/L or less (n = 109), 13.4-19.2 ng/L (n = 110), 19.3-27.3 ng/L (n = 110), 27.4-40.0 ng/L (n = 109), and greater than 40.0 ng/L (n = 109), respectively. For the 72-hour, 30-day, and 6-month follow-up, event rates showed significant differences according to the quintiles of PlGF levels (Figure 1). At 72 hours, there were 43 events (7.9% event rate); at 6 months, there were 64 events (11.7% event rate). The increased event rate in patients with elevated PlGF levels was not only related to a higher incidence of nonfatal myocardial infarction, but also driven by an increased mortality (0%, 0%, 1.8%, 3.7%, and 4.6% for the 5 quintiles, respectively; P<.001). After 72 hours (including coronary angioplasty in all patients), 12.1% of patients with high PlGF levels had an adverse event compared with 4.9% for patients with low PlGF levels (HR, 2.65; 95% CI, 1.39-5.05; P = .002) (Figure 2) and the crude event rates for 30 days were 14.8% for patients with high PlGF levels and 4.9% for patients with low PlGF levels (HR, 3.34; 95% CI, 1.79-6.24; P<.001). During 6 months of follow-up, event-rate curves continued to diverge (19.7% vs 6.2% for high and low PlGF levels, respectively; HR, 3.74; 95% CI, 2.13-6.54; P<.001). The single end point mortality at 6-month follow-up also significantly differed between both groups (4.0% vs 0.6%; HR, 6.75; 95% CI, 1.44-31.55; P = .009). A total of 11 patients died during 6 months of follow-up.

The predictive value of PlGF level was independent of myocardial necrosis as evidenced by TnT level (Table 2). High PlGF levels indicated increased cardiac risk in patients with undetectable TnT levels (4.5% vs 0.3% for high and low PlGF levels, respectively; P = .01), in patients with low TnT levels (21.0% vs 3.6%; P<.001), and in patients with high TnT levels (31.4% vs 10.4%; P = .003). Also, PlGF level predicted adverse outcome independent of sCD40L level: in patients negative for TnT and with low sCD40L levels, high PlGF levels remained predictive for increased cardiac risk (10.8% vs 2.8%; HR, 4.29; 95% CI, 1.28-14.39; P = .01) (Figure 3). In a multivariable analysis including baseline characteristics and 4 biochemical markers (PlGF, TnT, hsCRP, sCD40L), PlGF remained an independent and powerful predictor of increased cardiac risk at 30 days of follow-up (adjusted HR, 3.03; 95% CI, 1.54-5.95; P<.001) (Table 2) and at 6 months of follow-up (adjusted HR, 3.37; 95% CI, 1.84-6.17; P<.001).

A second blood sample, drawn before discharge (a mean of 7.2 [SD, 4.5] days after randomization), was available for 489 (89.4%) of the 547 patients receiving placebo. Levels of PlGF decreased only slightly from a mean (SD) of 27.95 (9.83) ng/L at baseline to 25.04 (11.29) ng/L at discharge (–10.4%; P = .01), whereas hsCRP levels had increased by 52.6% (P<.001). For patients with PlGF levels above 27.0 ng/L at discharge, the incidence of mortality and nonfatal myocardial infarction was significantly higher compared with patients with low PlGF levels, both at 30-day follow-up (4.6% vs 0.8%; P = .02) and at 6-month follow-up (7.4% vs 2.2%; P = .005).

Despite similar clinical presentation at the time of arrival, patients in the emergency department cohort were classified as follows at the time of discharge: 308 patients with ACS (117 patients had non–ST-elevation myocardial infarction), 91 patients with stable angina, 10 patients with pulmonary embolism, 11 patients with congestive heart failure, 7 patients with myocarditis, and 199 patients with no evidence of heart disease.

Mean (SD) PlGF levels were significantly higher in patients with unstable angina or non–ST-elevation myocardial infarction (n = 308) (28.3 [7.3] ng/L) compared with patients with stable angina (16.2 [3.6] ng/L) (P<.001), and with patients having no evidence of heart disease (9.6 [4.2]) (P<.001), respectively (Figure 4). Mean PlGF levels in patients with non–ST-elevation myocardial infarction did not significantly differ from mean PlGF levels in patients with unstable angina (30.5 [15.2] vs 28.3 [9.8] ng/L; P = .42). The 97.5th percentile upper reference limit in patients without evidence for heart disease was 24.9 ng/L and the 99th percentile upper reference limit was 27.3 ng/L. In 308 patients with ACS, 44.8% of the patients had PlGF levels above the 99th percentile upper reference limit. Consistent with the findings derived from the CAPTURE patients, PlGF levels did not correlate with markers of necrosis (eg, TnT; r = 0.07) or platelet activation (eg, sCD40L; r = 0.12), but did significantly correlate with markers of inflammation (eg, hsCRP; r = 0.43) (P<.001).

A total of 65 events (including 18 fatal events) were recorded during 30 days of follow-up. Using the threshold value for PlGF level of 27.0 ng/L derived from the CAPTURE cohort, patients with high PlGF levels were at significantly increased risk compared with patients with low PlGF levels (21.2% vs 5.3%; univariate analysis: HR, 4.80; 95% CI, 2.81-8.21; P<.001; multivariable analysis: adjusted HR, 4.52; 95% CI, 2.23-9.17; P<.001) (Figure 5). Levels of TnT (HR, 7.37; 95% CI, 4.10-13.26; P<.001), sCD40L (HR, 2.74; 95% CI, 1.54-4.86; P = .001), and PlGF (HR, 3.01; 95% CI, 1.68-5.38; P<.001) emerged as independent powerful predictors for cardiovascular events during 30 days of follow-up. Patients who were negative for all 3 markers were at very low cardiac risk (7 days: no event; 30 days: 2.1% event rate [95% CI, 1.5%-2.6%]) (Figure 6). A second blood sample collected 4 hours after arrival in the emergency department did not increase the predictive value of PlGF for 30-day outcome (area under the ROC curve: 0.70 [95% CI, 0.62-0.78] vs 0.71 [95% CI, 0.65-0.77] for the second vs the baseline blood samples, respectively; P = .84).

We found that PlGF blood levels at presentation are of prognostic value of clinical outcome in patients with known or suspected ACS. The predictive value of PlGF levels is independent of evidence for myocardial necrosis as determined by TnT levels¹⁵ and platelet activation as evidenced by sCD40L levels.¹⁷ Moreover, elevated PlGF levels did not identify only those patients with acute chest pain who developed ACS, but also those patients with an increased risk of recurrent instability after hospital discharge.

The role of PlGF as a primary inflammatory instigator of atherosclerotic lesion instability is supported by its proinflammatory effects in animal models of atherosclerosis or arthritis.¹¹ Although PlGF belongs to the VEGF family of growth factors, its pathophysiological role appears to be more related to vascular inflammation than to angiogenesis.¹¹ Whereas VEGF is activated by hypoxia and elevation of VEGF levels is regarded as an early adaptation of the myocardium to deprivation of blood flow,²¹ PlGF is not affected or even down-regulated by hypoxia.^22- 23 We did not find any correlation between PlGF levels and VEGF levels (data not shown) as a marker of myocardial ischemia or between PlGF levels and TnT levels as a marker of myocardial necrosis. Thus, PlGF levels do not appear to be confounded by myocardial necrosis, whereas VEGF levels are linked to elevated levels of TnT, impaired TIMI flow, and clinical evidence of myocardial ischemia.²⁴ The lack of PlGF levels being sensitive to minor myocardial injury might be specifically important in patients with ACS, of whom approximately one third are positive for TnT and troponin I at the time of arrival in the hospital.²⁵ In contrast, myocardial injury appears to compromise the value of hsCRP levels for predicting outcome in patients with ACS. As a classic unspecific downstream acute-phase marker, hsCRP levels can rise substantially after acute myocardial ischemia (TnT ≤0.01 µg/L: 14.1 [SD, 19.6] mg/L; TnT >0.01 µg/L: 23.6 [SD, 28.9] mg/L; P<.001) such that determining an individual's underlying basal level is difficult and may result in misclassification.²⁶

By multivariable proportional hazards analysis, levels of PlGF, sCD40L, and TnT all emerged as independent predictors of adverse outcome. Combining PlGF and sCD40L was especially revealing in patients negative for TnT, suggesting that both markers reflect distinct pathways that eventually contribute to a proinflammatory and procoagulant milieu in the coronary circulation. Given the superiority of PlGF level over hsCRP level for predicting cardiovascular events in patients with ACS, the identification of PlGF as a primary inflammatory instigator of coronary lesion instability will substantially enhance our diagnostic armamentarium for the diagnosis and risk stratification of patients with ACS.

Potential limitations of the current study merit consideration. First, because our blood samples were stored at −80°C until analysis, we cannot exlude the possiblity of protein degredation. However, the samples for this study were only thawed once and we have noted measured PlGF values to be similar in fresh and frozen samples (data available from authors upon request). Second, because of the selected nature of the CAPTURE cohort and the relatively small number of 30-day events, we were concerned that our results might not be generalizable. We therefore prospectively tested the prognostic value of PlGF and other markers in a "real-world" cohort of patients presenting with chest pain. Third, we could not determine from this cohort whether PlGF can reliably identify patients who will benefit from aggressive management, as has been shown for example with troponin T^15- 16,27 and interleukin 6.²⁸

In summary, PlGF plasma levels represent a potentially powerful clinical biomarker of vascular inflammation and adverse outcome in patients with ACS. Measuring PlGF levels may extend the predictive and prognostic information gained from traditional inflammatory markers in patients with ACS. Since the proinflammatory effects of PlGF can be specifically inhibited by blocking its receptor, Fms-like tyrosine kinase, these findings may also provide a rationale for a novel anti-inflammatory therapeutic target in patients with coronary artery disease.²⁹