Abstract
Background Right ventricular dysfunction and N-terminal proB-type natriuretic peptide (NT-proBNP) are established determinants of prognosis in acute pulmonary embolism (PE). The aim of the study was to investigate the prognostic value of C-reactive protein (CRP) in PE.
Methods Fifty-six patients (mean age, 64.4 ± 14.8years; 22 male subjects) with acute PE were consecutively enrolled and followed for 36 months after discharge. Serum CRP, NT-proBNP, and troponin T levels were determined. Right ventricular function was evaluated by transthoracic echocardiography.
Results Right ventricular dysfunction was present in 31 patients and was more frequent in patients with higher CRP and NT-proBNP levels (P = 0.020 and P = 0.045, respectively). During the 36-month follow-up, there were 15 terminal events (death due to recurrent PE). The mortality rate was 41.2% in patients with NT-proBNP levels greater than 1000 pg/mL, whereas it was 5.9% in patients with less than 500 pg/mL (P = 0.011). Mortality rates also were higher in patients with elevated CRP and troponin T levels, but the differences did not reach clinical significance. The survival rate of acute PE patients with lower NT-proBNP and CRP levels was better than that of patients with higher NT-proBNP and CRP levels. Receiver operating characteristic curve analysis demonstrated cutoff values for NT-proBNP as 1800 pg/mL (sensitivity, 93.3%; specificity, 68.2%; positive predictive values, 66.7%; and negative predictive values, 93.8%) and for CRP as 48mg/L (sensitivity, 72.7%; specificity, 61.9%; positive predictive values, 50.0%; and negative predictive values, 81.3%) to predict mortality in PE patients.
Conclusions C-reactive protein is associated with right ventricular dysfunction, which is a predictor of prognosis in PE and may become a promising biomarker for risk stratification of PE, although CRP is not found superior to NT-proBNP.
Pulmonary embolism (PE) initiates an inflammatory response that includes pulmonary arterial hypertension, ischemia, hypoxia, and thrombus-endothelial interactions.1,2Right ventricular (RV) dysfunction is a predictor of prognosis in PE.3,4Ischemia of the RV myocardium and RV infarction may lead to RV failure and mortality in patients with PE.5Elevated levels of cardiac troponin I or T, or B-type natriuretic peptide (BNP) and N-terminal pro-BNP (NT-proBNP) are associated with an increased risk of complications and death in patients with acute PE.6,7In rats, cardiac inflammation contributes to RV dysfunction after experimental PE. As a consequence, PE with moderately severe pulmonary hypertension results in selective RV dysfunction, which also was associated with inflammation.8To demonstrate the presence and effect of inflammation, certain markers of inflammatory responses, including growth differentiation factor 15, plasma heart-type fatty acid binding protein, C-reactive protein (CRP) and myeloperoxidase, interleukins, and selectins have been studied in PE.9-14
C-reactive protein is an acute phase reactant and is associated with increased risk of cardiovascular and cerebrovascular events.12,15C-reactive protein is a well-known marker of inflammation and tissue damage. Incubation of highly purified CRP with monocytes of peripheral blood may induce tissue factors that increase procoagulant activity significantly. Incubation of human saphenous vein with human recombinant CRP has been shown to increase the tendency to thrombosis through the mediation of interleukin (IL) 6 and monocyte chemoattractant protein 1.16-18Thus, CRP may play a role in the pathophysiology of the vascular wall.
The prognostic value of CRP in acute PE is unknown. The aim of the present study was to investigate the prognostic role of CRP in acute PE.
METHODS
The investigation conforms to the principles outlined in the Declaration of Helsinki. The study was approved by the local ethics committee of Marmara University, and all participants gave written informed consent before participating.
Fifty-six patients with acute PE were recruited consecutively into the study. The diagnosis of PE was made by using spiral computed tomography (CT)-angiography in 31 patients and by ventilation-perfusion scan in 22 patients. Ventilation-perfusion (VQ) scans were performed instead of CT angiography in these 22 patients on the basis of their potential risk of developing contrast nephropathy after CT angiography (including a history of diabetes mellitus, old age [older than 65 years], and borderline creatinine level). All PE patients diagnosed by VQ scans have a high clinical probability score with an unequivocally positive VQ scan. The diagnosis of the remaining 3 patients, who were hemodynamically too unstable to undergo CT angiography or a VQ scan, was based on a combination of clinical presentation, including acute onset dyspnea, hypotension, tachypnea or shock, and abnormal echocardiographic findings. In addition to clinical, echocardiographic and radiological data, the diagnosis of PE was further encouraged by a positive D-dimer (upper limit of normal, 0.5 ug/mL), arterial blood gas, and electrocardiographic findings. Pulmonary embolism was graded as small (nonmassive; when neither RV dysfunction nor hypotension was present), moderate (submassive; when RV dysfunction was present without hypotension), or severe (massive; when both RV dysfunction and hypotension were present).19,20Right ventricular function was evaluated by transthoracic echocardiography, and RV dysfunction was diagnosed in the presence of any of the following: dilatation of the RV (diastolic diameter, >30 mm), abnormal motion of the interventricular septum, hypokinesis of RV, or tricuspid valve regurgitation (jet velocity, >2.5 m/s).21
We collected all data regarding clinical symptoms and signs of the patients on hospital admission. Radiological examinations, including posteroanterior chest radiographs, CT, VQ scans, and echocardiographic examination, were all performed within 6 hours of admission. Blood samples were obtained within 3 hours of admission to determine serum CRP (upper limit of normal, 5 mg/L), NT-proBNP (upper limit of normal, 500 pg/mL), and troponin T levels (upper limit of normal, 0.1 ng/mL). C-reactive protein was measured by latex-enhanced nephelometry (N high-sensitivity CRP assay) on a BNII nephelometer (Dade Behring, Inc., BN Prospect, Marburg, Germany). As defined by the manufacturer, detectable CRP concentration using this assay ranged from 0.175 to 1100 mg/L.
After the diagnosis of PE, all patients received standard anticoagulant treatment including low molecular-weight heparin (enoxaparin, 100 IU/kg twice daily), followed by warfarin therapy with a target international normalized ratio of 2.0-3.0. The patients were closely followed during hospitalization and in the pulmonary medicine outpatients clinics at the first, second, third, and sixth months after discharge and then every 6 months unless the patient had any complaints. Patients were followed prospectively for 36 months after discharge from the hospital to evaluate prognosis and mortality.
Patients with known renal failure or decompensated left heart failure, which might result in elevated NT-proBNP levels were excluded from statistical analysis considering NT-proBNP (3 patients). None of the remaining patients had chronic obstructive pulmonary disease associated with pulmonary hypertension, which also might cause elevated levels of NT-proBNP. Patients with any disease associated with elevated CRP levels, such as active rheumatologic disease, active infection, chronic in`flammatory conditions, recent myocardial infarction, and cerebrovascular diseases were excluded from the analysis considering CRP (8 patients).
Statistical Analysis
The statistical analysis was made with the use of a commercially available statistical package SPSS for Windows, Version 15.0. Continuous variables were expressed as mean ± SD, whereas categorical variables were expressed as ratios. Categorical and discrete variables were compared using the χ 2 test (either 2 × 2 or 3× 2). The method of life table analysis was used for the survival analysis. Cumulative survival curves were obtained by the Gehan (generalized Wilcoxon) test. Receiver operating characteristic curve analysis was performed to determine the cutoff levels of NT-proBNP and CRP to predict mortality. Logistic regression analysis was performed to explore the independent predictors of mortality at 6 months and odds ratios and 95% confidence intervals. Probability values of less than 0.05 were considered significant.
RESULTS
Fifty-six consecutive patients with a diagnosis of acute PE were included in the study. The mean age of the patients was 64.4 ± 14.8 years, and 22 of them were male patients. According to previously defined criteria from Goldhaber, 1923 (41.1%) of the patients had small PE, whereas 29 patients (51.8%) had moderate, and 4 (7.1%) had massive PE. Clinical features and laboratory parameters of the patients are listed in Table 1.
Among the 56 patients with PE, 31 patients (55.4%) had RV dysfunction. Among these 31 patients, 1 patient was excluded from statistical analysis of NT-proBNP because he had renal failure, and 5 patients were excluded from analysis of CRP because of their comorbidities associated with elevated CRP levels. The RV dysfunction was significantly associated with elevated levels of NT-proBNP (P = 0.020). The frequency of RV dysfunction was higher in patients with elevated NT-proBNP levels (Table 2). Similarly, RV dysfunction was more frequent in acute PE patients with elevated levels of CRP (Table 2). This relation also was statistically significant (P = 0.045). However, there was not any significant difference in the frequency of RV dysfunction between the patients with elevated troponin T levels and those with normal troponin T levels (Table 2).
During the follow-up period of 36 months, there were 15 terminal events, and the cause of death was recurrent PE in all patients. Among these 15 patients, 4 patients were excluded from analysis of CRP because their CRP levels were considered to have been due to comorbidities associated with elevated CRP levels. Table 3 demonstrates mortality rates in PE patients based on admission NT-proBNP, CRP, and troponin T levels. There was a significant difference in the mortality rate only in relation to NT-proBNP levels.
The life table analysis showed that the survival of acute PE patients with low NT-proBNP values was better than the survival of patients with high NT-proBNP levels. The cumulative survival difference between groups with low and high NT-proBNP was found to be statistically significant with the Gehan (generalized Wilcoxon) test (P < 0.05) (Fig. 1). The life table analysis demonstrated that the acute PE patients with low CRP values also had better 36-month survival rate than those with high CRP levels. There was only 1 terminal event noted among the patients with CRP levels less than 10 mg/L during the follow-up, whereas 10 terminal events were observed in patients with CRP levels higher than 10 mg/L. The cumulative proportion surviving at the end of 36 months was lower in patients with CRP levels higher than 10 mg/L compared with those with CRP levels less than 10 mg/L (cumulative survival ratio [Si], 0.64 ± 0.09 vs 0.86 ± 0.13). Also, the hazard rate (H) of the patients with CRP levels less than 10 mg/L within the 36 months was lower than the hazard rate of the patients with CRP levels higher than 10 mg/L (0.05 ± 0.05 vs 0.11 ± 0.04). However, there was no statistically significant difference in the cumulative survival rates between groups with low and high CRP levels with Gehan (generalized Wilcoxon) test (Fig. 2).
Receiver operating characteristic curve analysis demonstrated cutoff values for NT-proBNP as 1800pg/mL and for CRP as 48mg/L to predict mortality rate (Fig. 3). Table 4 demonstrates the sensitivity, specificity, and positive and negative predictive values for NT-proBNP and CRP.
We modeled a logistic regression analysis to explore the independent predictors of mortality. Age, sex, comorbidities (hypertension, congestive heart failure, and coronary artery disease), and NT-proBNP and CRP levels were included into the model. The adjusted R2 of the model was 0.695 (P < 0.001). The NT-proBNP levels were still associated with mortality when adjusted by age, sex, and comorbidities (P = 0.038). Logistic regression analysis revealed no other significant association with mortality. Table 5 demonstrates the biomarkers combined with echocardiography for prediction of mortality in PE patients and indicates that NT-proBNP levels were the strongest predictor of mortality in our cohort.
DISCUSSION
The approach of risk stratification in acute PE has been developing recently. Right ventricular dysfunction is an evidence-based echocardiographic finding associated with a prognosis in PE.22,23However, echocardiography and other radiological methods, such as CT angiography, are not always available to determine the prognosis in acute PE, especially in hemodynamically unstable patients and may not give additional information about long-term prognosis. The use of biochemical biomarkers has been suggested in risk stratification and determination of prognosis in acute PE. Cardiac troponin T and NT-proBNP have been recently studied for risk stratification in acute PE. Although elevated levels of NT-proBNP are associated with adverse outcome,7controversy exists regarding the usefulness of troponin as a test for the risk stratification of patients with acute PE.24A combination of elevated biomarkers and echocardiographic RV dysfunction seems to be a better predictor of mortality. Whereas RV dysfunction assessed by echocardiography provides a 5% positive predictive value for mortality in hemodynamically stable PE patients,25cardiac troponin T and BNP increase this ratio to 35%, 40%, and 17%, respectively.26,27Binder et al.7showed that an adverse outcome increased by 10 to 12 times when elevated NT-proBNP and troponin levels were combined with echocardiographic RV dysfunction. Despite these strategies for the risk stratification of PE, there is still a need for a simple and easily applicable method to determine prognosis in PE.
Inflammation has long been known to be involved in the pathophysiology of arterial and venous thrombosis by the means of activation of coagulation cascades with the formation of thrombin and fibrin deposition. Inflammatory mediators, particularly CRP, IL-6, IL-8, and tumor necrosis factor α, play a role in arterial and venous thrombosis.28-30It has been shown in limited animal studies that inflammation after PE contributes to RV damage, dysfunction, and cardiac inflammation.8Pulmonary embolism can cause vascular inflammatory reactions by thromboembolism of the pulmonary artery and also can cause pulmonary parenchymal inflammation through a possible pulmonary infarction mimicking pneumonia. Pulmonary embolism can precipitate RV dysfunction through the cardiac inflammation, which contributes to myocyte damage.8
Recently, several studies have proposed a relation between inflammatory markers and venous thromboembolism (VTE) and have recognized their role in the cause and prognosis of venous thrombosis including PE.10,31C-reactive protein, erythrocyte sedimentation rate, and IL-6 at diagnosis of venous thromboembolic disease were suggested to be useful to identify patients with higher risks of death and postphlebitic syndrome during the first year after diagnosis.32
C-reactive protein can be used to assess the risk and prognosis of myocardial ischemia and ischemic stroke that have pathophysiological mechanisms based on arterial thrombosis. In the literature, the relation between RV dysfunction and CRP was investigated in only one study that was not dealing with PE.33According to our knowledge, a relation between CRP and RV dysfunction and prognosis has not been studied in acute PE patients before. We explored the prognostic value of CRP in patients with acute PE and found that RV dysfunction was more frequent among the patients with elevated CRP levels when the patients were divided into groups according to CRP levels of less than 10 mg/L, 10 to 100 mg/L, and greater than 100 mg/L. The prognostic and diagnostic studies in the literature about CRP levels also has used the cutoff values of 10 and 100 mg/L for different cardiopulmonary diseases including myocardial infarction, ischemic stroke, VTE, and pneumonia.30,34,35Most of these studies explored the association between elevated plasma CRP levels (mostly defined as >10 mg/L) and VTE but not PE.35-39They failed to conclude that CRP could either predict future VTE or be important in the diagnosis of VTE.35
The present study demonstrated that high levels of NT-proBNP and CRP in acute PE patients might predict adverse outcomes during 36 months of follow-up. High levels of CRP and NT-proBNP were associated with RV dysfunction and lower survival rates. Although the association with RV dysfunction was statistically significant for both NT-proBNP and CRP, the association with mortality was significant only for NT-proBNP. We think that the small sample size might be a reason for the statistical insignificance for CRP. As vascular inflammatory reactions such as seen in myocardial ischemia and pulmonary parenchymal inflammation play a role in cardiac inflammation in PE, it is logical to expect a prognostic value for CRP in acute PE. Higher CRP levels indicate a higher extent of cardiac inflammation, and thus more susceptibility to RV dysfunction, which also is a prognostic determinant in these patients. In our study, CRP correlated with presence of RV strain but did not seem to outperform BNP for prediction of outcome. However, CRP is a more widely available biomarker, and further prospective evaluation of its value in severity prediction in larger independent cohorts is now warranted.
There are certain limitations to our study. We did not study other inflammatory cytokines in parallel with CRP levels. We only included CRP and NT-proBNP levels determined at the time of admission; CRP and NT-proBNP levels repeated outside the acute setting were not included in the study. Also, certain drugs including statins might interfere with CRP levels. Because the amount, duration, and type of statin therapy differed greatly between the patients with hyperlipidemia, we could not provide any interpretation about statin effects on CRP analysis. The small sample size is another study limitation, and further exclusion of the patients with other comorbidities that might cause elevated levels of NT-proBNP and CRP also might affect the results. Yet the present study is important because it has shown an association between CRP levels and RV dysfunction, which is known as a main predictor of mortality and prognosis in acute PE. Survival curves showed better survival with CRP levels lower of 10 mg/ml. However, the difference did not reach statistical significance. This might be due to the small sample size, and we suggest our result should be confirmed in larger samples.
CONCLUSIONS
Acute PE patients with higher levels of CRP had more frequently RV dysfunction. Survival at 36 months was lower in patients with higher levels of CRP, although the difference did not reach statistical significance. C-reactive protein may help in the risk stratification of patients with acute PE and provide a guidance of the appropriate therapeutic management of acute PE as being an easily applicable biochemical marker. C-reactive protein levels behave as a marker of RV dysfunction. However, with the results as our basis, elevated NT-proBNP levels were significantly associated with RV dysfunction and mortality in PE. C-reactive protein was not superior to NT-proBNP in predicting mortality in acute PE, and its value as predictor of mortality, separated of BNP, is probably low. We suggest that the prognostic value of CRP in acute PE be explored in larger patient samples before any definitive conclusion is drawn.
ACKNOWLEDGMENT
The authors thank Prof. R. W. Guillery from the University of Oxford for the English correction of the manuscript.