Comparison of 8 biomarkers for prediction of right ventricular hypokinesis 6 months after submassive pulmonary embolism

Congestive Heart Failure

Comparison of 8 biomarkers for prediction of right ventricular hypokinesis 6 months after submassive pulmonary embolism Jeffrey A. Kline, MD, a Raghid Zeitouni, BS, a Michael R. Marchick, MD, a Jackeline Hernandez-Nino, MD, a and Geoffrey A. Rose, MD b Charlotte, NC

Background Elevated blood concentrations of troponin proteins or brain natriuretic peptide (BNP) worsen the prognosis of patients with pulmonary embolism (PE). Novel biomarkers that reflect mechanisms of right ventricle (RV) damage from PE may provide additional prognostic value. We compare the prognostic use of BNP, troponin I, D-dimer, monocyte chemoattractant protein-1, matrix metalloproteinase, myeloperoxidase, C-reactive protein, and caspase 3 as biomarkers of RV damage and adverse outcomes in submassive PE. Methods This article used a prospective cohort study of normotensive (systolic blood pressure always >100 mm Hg) patients with computed tomographic angiography-diagnosed PE. All patients underwent echocardiography and phlebotomy at diagnosis, and survivors had another echocardiography 6 months later. We tested each biomarker for prognostic significance, requiring a lower limit 95% CI >0.50 for the area under the receiver operating characteristic curve (AUROC) with a reference standard positive of RV hypokinesis on either echocardiogram. Biomarkers with prognostic significance were dichotomized at the concentration that yielded highest likelihood ratio positive and mortality rates compared (Fisher exact test). Results We enrolled 152 patients with complete data. Thirty-seven (24%, 95% CI 18%-32%) had RV hypokinesis. Only BNP and troponin had significant AUROC values as follows: 0.71 (95% CI 0.60-0.81) and 0.71 (95% CI 0.62-0.82), respectively. Overall mortality was 13/153 (8.5%); mortality rate for BNP >100 versus ≤100 pg/mL was 23% versus 3% (P = .003), respectively. Mortality rate for troponin I >0.1 versus ≤0.1 ng/mL was 13% versus 6% (P = .205), respectively. Conclusions Of 8 mechanistically plausible biomarkers, only BNP and troponin I had significant prognostic use with BNP having an advantage for predicting mortality. (Am Heart J 2008;156:308-14.) Submassive acute pulmonary embolism (PE), defined as PE that never produces arterial hypotension, can permanently damage the right ventricle (RV).1 Prospective studies have shown that patients with PE and RV hypokinesis observed on echocardiography at diagnosis have a significantly increased probability of short-term mortality and cardiopulmonary functional limitations.1 A substantial body of literature has demonstrated the value

From the aDepartment of Emergency Medicine, Carolinas Medical Center, Charlotte, NC, and bDepartment of Carolinas Heart and Vascular Institute, Carolinas Medical Center, Charlotte, NC. The Biosite Corporation performed assays for this study in kind and had no role in the design or conduct of this study or in writing this manuscript. This study is also supported by National Institutes of Health/National Heart, Lung, and Blood Institute (Bethesda, MD) grant no. RO1HL074384 and Carolinas Medical Center Health Services (Charlotte, NC) Foundation grant (Dr J A Kline). Submitted December 10, 2007; accepted March 25, 2008. Reprint requests: Jeffrey A. Kline, MD, Emergency Medicine Research, Department of Emergency Medicine, Carolinas Medical Center, PO Box 32861, Charlotte, NC 283232861. E-mail: [email protected] 0002-8703/$ - see front matter © 2008, Mosby, Inc. All rights reserved. doi:10.1016/j.ahj.2008.03.026

of the troponins and brain natriuretic proteins (BNP) to risk-stratify patients with PE.2,3 The RV probably releases these proteins as a result of subendocardial ischemia and intramural mechanical strain caused by impedance to RV ejection with PE. However, other biomarkers may rapidly increase in parallel with secondary inflammatory and repair mechanisms initiated by PE. Watts and et al4 demonstrated that experimental PE produces mechanical shear and ischemic injury to RV myocytes, which provokes a rapid and robust neutrophil-mediated inflammatory response with cardiac necrosis and subsequent remodeling.4-6 No study has yet determined if proteins associated with these secondary mechanisms predict RV damage from PE. In this report, we test 8 candidate protein biomarkers for their ability to prognose the 6-month outcome of RV function in humans with submassive PE. We chose these 8 biomarkers because of demonstrated or hypothesized pathophysiologic connection to the mechanisms of RV damage caused by PE (briefly described in parenthesis): troponin I (myocardial ischemia), BNP (cardiomyocyte stretch and shear stress), D-dimer (inflammatory response and thrombus mass), monocyte chemoattractant protein-

American Heart Journal Volume 156, Number 2

1 (MCP-1) (myocardial inflammation and chemotaxis), matrix metalloproteinase-9 (MMP-9) (neutrophilmediated myocardial remodeling), myeloperoxidase (MPO) (neutrophil activation), C-reactive protein (CRP) (systemic and myocardial inflammation), and caspase 3 (cardiomyocyte apoptosis). The specific aims of this project were first to measure and compare the prognostic accuracy of 8 biomarkers for the prediction of persistent RV hypokinesis observed on echocardiography in patients with acute, submassive PE and, second, to test if the biomarkers that predict RV hypokinesis also accurately predict the adverse outcomes of death, dyspnea, and exercise intolerance 6 months later.

Methods This study was approved by the institutional review board and privacy board at Carolinas Medical Center (Charlotte, NC), a large, urban, academic hospital with >110,000 ED visits each year. We prospectively enrolled ED patients and hospital inpatients and with confirmed PE from January 2002 to July 2004.

Patient selection We studied consecutive patients with diagnosed PE. Six days per week, at 7:00 AM and 2:00 PM each day, our research group received via email, a computerized standard query language output that catalogued all orders for computerized tomographic (C.T.) chest angiography. Each day of enrollment, one of the authors (either J.H. or J.A.K) immediately tracked each scan for the initial “wet read” positive for PE as interpreted by a boardcertified radiologist. We then investigated for any of the following exclusion criteria: (1) >12 hours since start of heparin therapy, (2) systolic hypotension (b100 mm Hg for 2 consecutive measurements obtained >15 minutes apart),7 (3) treatment with fibrinolytic therapy, catheter fragmentation, or surgical embolectomy within the previous 48 hours, (4) illnesses with a predicted 6-month mortality >50% (eg, metastatic cancer, end-stage acquired immunodeficiency syndrome, end-stage heart or renal failure with no plan for organ transplantation or hemodialysis therapy), (5) a do-not-resuscitate order with clinical plan to not treat the patient for PE, (6) anatomy or clinical condition that precluded the ability to measure RV function via echocardiography, (7) permanent inability to ambulate, (8) prisoners with expected incarceration time >6 months, (9) personal physicians' refusal, and (10) inability to obtain blood samples or echocardiography within 12 hours after heparin administration.

Study protocol After obtaining written, informed consent, a study physician (J.A.K or J.H.) immediately completed a detailed data collection form. Blood specimens used for biomarker measurements and echocardiography examinations were obtained either before heparin administration or as soon as possible after heparin administration (but not >12 hours after heparin). Patients were followed for 6 months for mortality, and survivors returned at 6 months for another echocardiography, administration of questionnaires, and a 6-minute walk distance (6MWD) test.

Kline et al 309

Blood specimens were obtained by phlebotomy into Vacutainer tubes containing EDTA (1.8 g/L K3EDTA [Becton Dickinson, Franklin Lakes, NJ]). Samples were transported on ice and immediately centrifuged for 10 minutes at 2,000g, fractionated, and frozen at −70°F. At approximately 3-month intervals, frozen plasma samples were shipped on dry ice to Biosite (San Diego, CA) for biomarker quantification. Transthoracic echocardiography was performed using a prearranged stat protocol designed specifically to examine the right ventricle. At the time of diagnosis, echocardiography was performed either in the hospital's Intersocietal Commission for the Accreditation of Echocardiography Laboratories-approved laboratory or at the bedside using a small footprint 5 to 7.5 MHz probe and Vivid 5 or 7 machines (General Electric Healthcare, Piscataway, NJ). Images were obtained by a registered diagnostic cardiac ultrasonographer with echocardiography specialization. We collected data using a standard form on each day of hospitalization. Six months after discharge, surviving patients returned for another echocardiography and other testing as previously described.8 At 6-month follow-up, patients were administered a dyspnea questionnaire followed by echocardiography and a 6MWD, performed in accordance with published guidelines.9 All echocardiograms were performed in the echocardiography laboratory. Study end points were death within 6 months or 6-month follow-up.

Measurements Proteins were analyzed on thawed plasma that had been frozen once only. Duplicate measurements were performed on each sample, and the average of the 2 measurements is reported. For BNP and D-dimer, we measured the coefficient of variability (100*SD/mean) for 3 measurements. The method of protein quantification was based upon a 2-site “sandwich” immunoassay and chemiluminescence detection technique for all markers. A cardiologist with subspecialty certification in echocardiography, blinded to the results of any data collected for this study, interpreted digitalized videos of echocardiograms. The cardiologist used explicit criteria to grade the RV contraction and size, to assess for interventricular septal flattening, and to measure the tricuspid regurgitant jet velocity. The RV hypokinesis required the qualitative interpretation of unequivocally abnormal depression in global RV contraction or the observation of hypokinesis in the infundibular RV region (McConnell sign), observed in the apical 2-chamber or 4-chamber views.10 The RV dilation required an RV diameter >0.9 of the left ventricular diameter, observed in the apical 4-chamber view. Septal flattening was defined as the observation of definite thinning of the interventricular septal thinning during systole. The RV systolic pressure (RVSP) was estimated using the Bernoulli equation (RVSP = 10 + 4V2). To estimate the percentage of the pulmonary vasculature that was obstructed by PE from CT angiography, we used the method of Mastora et al.11

Classification of RV function Patients were classified into 1 of 3 groups as follows based upon the cardiologist's interpretation of RV geometry and contraction observed on either the echocardiography at diagnosis or at 6-month follow-up: (1) RV normal required absence of RV dilation and absence of RV hypokinesis on both echocardiograms; (2) RV dilation required RV dilation on either

American Heart Journal August 2008

310 Kline et al

Table I. Parametric clinical data measured at the time of PE diagnosis Echocardiographic classification

RV normal ⁎ (n = 72) Clinical feature Age SBP (mm Hg) SaO2 (%) Pulse rate (beats/min) Body mass index (kg/m2) Occlusion on CT (%)

RV dilated (n = 36)

RV hypokinesis (n = 37)

Mean

SD

Mean

SD

Mean

SD

52 129 95 95 ‡ 31 45

17 22 3 18 7 34

54 127 95 88 31 45

14 24 3 19 10 35

56 123 92 † 100 § 29 68 ‖

17 23 5 22 6 30

⁎Excluding 7 patients who died before 6-month follow-up. †P = .0019 versus RV normal and P = .012 versus RV dilated. ‡P = .020 versus RV dilated. §P = .018 versus RV dilated. ║P = .002 versus RV normal and P = .012 versus RV dilated.

echocardiogram but with normal RV contraction on both studies; (3) RV hypokinesis required RV hypokinesis on either echocardiogram regardless of other findings. We also recorded the frequency of interventricular septal flattening (IVSF), elevated RSVP (>40 mm Hg), and a depressed left ventricular ejection fraction (LVEF b45%).

Reference standard We used RV hypokinesis (defined above) as the reference standard for receiver operating characteristic curve analysis. Rationale for choosing RV hypokinesis includes methodological clarity and its widespread recognition among echocardiographers and strength of association with mortality after submassive PE.12

Adverse outcomes Functional cardiopulmonary limitations at 6 months were either the presence of dyspnea at rest on more than one half of days or exercise intolerance based upon a 6MWD b330 m at 6-month follow-up. We followed all patients until July 2007 for the end point of death, based upon family or physician report or positive identification in the US social security death index; we obtained death certificates on all patients and performed chart review on all patients who died.

Cardiam Statistical analysis Descriptive data are shown as means ± SD and medians with first to third interquartile ranges. Means were compared with a one-way analysis of variance (ANOVA) followed by the TukeyKramer test. Proportions are shown with 95% CIs from the Clopper-Pearson exact formula. Medians were compared with the Kruskal-Wallis ANOVA, followed by the Dwass-SteelChritchlow-Fligner test for significance. We then tested the diagnostic accuracy of each biomarker for predicting 2 outcomes: (1) the criterion standard positive was RV hypokinesis on either echocardiogram, and all other findings were criterion standard negative and (2) the criterion standard positive was either RV hypokinesis or dilation, and the criterion standard negative was a completely normal RV at diagnosis and 6-month follow-up. We examined the diagnostic

performance of each biomarker using receiver operating characteristic curve analysis with area under the curve (Wilcoxon method) as the primary efficacy measurement. For any biomarker with a lower limit 95% CI >0.5 (indicating possible use as a diagnostic test) in the primary analysis, we determined the threshold that produced the highest likelihood ratio positive (sensitivity/[1 − specificity]) and used this as the cutoff as the threshold where for test positive. We assessed the influence of significant biomarkers status on survival curves from the day of enrollment until July 31, 2007, using a Cox proportional hazards test that included age and echocardiographic RV functional status as covariates. We compared frequencies of adverse outcomes based upon dichotomized biomarker results (positive or negative) using a 2-sided Fisher exact test. We estimated a sample size of 150 patients would narrow the 95% CI for the under the receiver operating characteristic curve to b±0.1, assuming that at least one quarter would have the disease-positive outcome of RV hypokinesis. Statistical analyses were performed using StatsDirect v. 2.6.2 (StatsDirect, Cheshire, England).

Results We enrolled 152 patients with PE diagnosed with CT angiography and none of the 10 exclusion criteria. All were treated initially with unfractionated heparin. Eighty-seven patients were enrolled from the ED, and 65 were enrolled from the inpatient setting. We classified these 152 patients into 3 groups based upon our predefined echocardiographic definitions of RV function as follows: 1. RV normal (n = 72): seventy-nine patients had normal RV size and function on echocardiography at diagnosis. Of 79, 7 (9%, 95% CI 4%-14%) died before follow-up with a mean survival of 82 ± 59 days. The causes of death were cancer (n = 4), end-stage renal failure (n = 2), and recurrent PE (n = 1). Thus, 72 patients had a normal echocardiogram at diagnosis, survived to 6 months, and

American Heart Journal Volume 156, Number 2

Kline et al 311

Table II. Comorbid conditions recorded at the time of PE diagnosis Echocardiographic classification RV hypokinesis (n = 37)

RV normal (n = 72)

RV dilated (n = 36)

Clinical feature

n

%

n

%

n

%

Prior MI CAD without MI Prior PE COPD Inactive cancer Active cancer CHF Idiopathic PE ⁎

1 8 2 3 3 9 4 16

1 11 3 4 4 13 6 22

2 4 5 3 1 2 2 11

5 11 14 8 3 5 5 30

1 2 5 1 2 2 0 11

3 6 14 3 6 6 0 31

MI, Myocardial infarction; CAD, coronary artery disease; COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure. ⁎PE with no identified risk factor for thrombosis.

Table III. Plasma concentrations of each biomarker stratified by echocardiographic classification of the right ventricle Echocardiographic classification RV normal (n = 72) Biomarker Troponin (ng/mL) BNP (pg/mL) D-dimer (μg/mL) MCP-1 (ng/mL) MMP-9 (ng/mL) MPO (ng/mL) CRP (mg/dL) Caspase 3 (ng/mL)

RV dilated (n = 36)

RV hypokinesis (n = 37)

Median

1st-3rd IQR

Median

1st-3rd IQR

Median

1st-3rd IQR

0.0 12 12.4 129 76 81 54 1.7

0.0-0.0 0-63 7.3-15.4 81-199 43-162 43-123 23-130 1.0-3.3

0.0 22 10.3 120 49 75 42 1.8

0.0-0.3 10-78 7.5-18.6 60-168 35-128 47-105 60-105 0.9-3.9

0.3 ⁎ 129 † 13.6 145 83 80 49 1.5

0.0-1.1 41-448 8.8-22.9 99-222 34-152 57-118 22-80 1.1-3.3

IQR, Interquartile range. ⁎P b .0001 versus RV normal and P = .0023 versus RV dilated. †P = .0001 versus RV normal and P = .0048 versus RV dilated.

had a normal echocardiogram at 6-month followup. Associated echocardiographic findings included none with IVSF, RVSP >40 mm Hg (n = 10), and an LVEFb45% (n = 4). 2. RV dilation (n = 36): RV dilation without RV hypokinesis on the initial echocardiogram (n = 18), on the successive echocardiogram (n = 7), or on both echocardiograms (n = 11). One patient (3%, 95% CI 0%-15%) died of septic shock 82 days after enrollment. Associated echocardiographic findings included IVSF (n = 2), RVSP >40 mm Hg (n = 8), and LVEF b45% (n = 4). 3. RV hypokinesis (n = 37): RV hypokinesis on the initial echocardiogram (n = 27), on the successive echocardiogram (n = 1), or on both echocardiograms (n = 9). Five patients (14%, 95% CI 5%-29%) with RV hypokinesis died before follow-up with a mean survival of 53 ± 40 days. The causes of death were circulatory shock from PE (n = 3), septic shock (n = 1), and cardiac arrest, etiology uncertain

(n = 1). Associated echocardiographic findings included RV dilation (n = 37), IVSF (n = 11), RVSP>40 mm Hg (n = 24), and LVEF b45% (n = 5). Table I presents parametric clinical data for each of the 3 groupings based upon echocardiographic classification of the RV. This table excludes the 7 patients who had an RV normal at diagnosis but who died before 6-month follow-up. Analysis of variance testing demonstrated that the RV hypokinesis group had a significantly lower mean pulse oximetry reading and a higher proportion of their pulmonary vasculature obstructed by pulmonary embolism compared with the other 2 groups. In addition, the RV hypokinesis group had a higher pulse rate than the RV dilation group. Table II presents the frequency of comorbid conditions within each group based upon echocardiographic classification of RV function. The data suggest that patients with RV normal were less likely to have a history of PE compared with RV hypokinesis or RV

American Heart Journal August 2008

312 Kline et al

Table IV. Areas under the receiver operating characteristic curve for each biomarker RV hypokinesis vs all other

RV normal vs all other

Biomarker

AUC ROC

LL 95% CI

UL 95% CI

AUC ROC

LL 95% CI

UL 95% CI

Troponin I BNP D-dimer MCP-1 MMP-9 MPO CRP Caspase 3

0.71 0.71 0.62 0.55 0.50 0.54 0.55 0.51

0.62 0.61 0.5 0.45 0.39 0.43 0.44 0.41

0.81 0.80 0.73 0.66 0.61 0.64 0.65 0.61

0.51 0.54 0.54 0.54 0.58 0.54 0.52 0.50

0.42 0.44 0.43 0.43 0.46 0.44 0.41 0.39

0.60 0.64 0.65 0.64 0.69 0.65 0.62 0.61

AUC ROC, Area under the receiver operating characteristic curve for the ability to discriminate between patients based on echocardiographic results. LL, lower limit; UL, upper limit.

dilated; otherwise, the 3 groups had no salient differences in comorbidities.

Biomarker results Table III shows the median concentrations of each biomarker based upon echocardiographic classification of the RV. Comparison of medians values with the Kruskal-Wallis ANOVA revealed that patients with RV hypokinesis had significantly lower median concentrations of BNP and troponin I compared with patients with RV normal or RV dilated. No other biomarker was significantly different. The mean coefficients of variability were 4% ± 6% for troponin and 17% ± 20% for BNP. Table IV shows the areas under the receiver operating characteristic curves with associated 95% CIs for the 8 biomarkers examined. When RV hypokinesis was used as disease-positive, 2 biomarkers had an area under the curve with a lower limit of the 95% CI >0.50: BNP and troponin I. All other biomarkers had had a lower limit 95% CI that overlapped 0.50 and were therefore eliminated as candidate biomarkers for the detection of RV hypokinesis. Table IV presents the area under the receiver operating characteristic curves for each biomarker when the RV normal patients were considered disease-negative, and all other patients were disease-positive. In this analysis, all 8 biomarkers had an area under the curve with a lower limit 95% CI that overlapped 0.5 and receiver operating characteristic curves that overlapped the diagonal midline (plots not shown), and no cutoff for any of the 8 biomarkers yielded a likelihood ratio negative b0.30. For BNP, the cutoff that yielded the highest likelihood ratio positive (LR+) for the detection of RV hypokinesis was 100 pg/mL, yielding a sensitivity = 57% (95% CI 40%73%), a specificity = 78% (95% CI 70%-86%), and an LR+ = 2.6. The threshold for troponin I was 0.1 ng/mL, yielding a sensitivity = 68% (95% CI 50%-82%), specificity = 73%

(95% CI 64%-81%), and LR+ = 2.5. Table V shows the percentage of patients who died, had dyspnea at rest, or exercise intolerance at 6-month follow-up based upon classification using these cutoffs. The mortality rate at 6 months was significantly higher in patients with a BNP >100 and a greater proportion of survivors who had a BNP >100 pg/mL had a 6MWD b330 m. The Cox proportional regression analysis presented in Table VI demonstrates that only BNP maintained significant influence on the time to death over long-term follow-up (hazards ratio 2.74, 95% CI 1.07-6.96).

Discussion This study simultaneously and prospectively examined the ability of BNP, troponin I, D-dimer, MCP-1, MPO, MMP-9, CRP, and caspase 3 to distinguish the subset of patients with normotensive PE who manifested RV hypokinesis on echocardiography performed at either diagnosis or on 6-month follow-up. The main finding was that only 2 of the 8 markers, BNP and troponin I, demonstrated significant, albeit modest prognostic accuracy. Both BNP and troponin I had an area under the receiver operating characteristic curve of 0.71 with lower limit 95% CIs >0.60. Because the areas under receiver operating characteristic curves exhibited a lower limit of the 95% CI that overlapped 0.50, we excluded the other 6 proteins as candidate biomarkers. We then found the optimal cutoffs for prediction of RV hypokinesis were 100 pg/mL for BNP and 0.1 ng/mL for troponin I—values that are very similar cutoffs chosen by other investigators.2,3 When BNP and troponin I were treated as dichotomous predictors at these thresholds, BNP demonstrated an advantage over troponin I, inasmuch as only BNP was associated with a statistically significant increase in frequency of exercise intolerance 6 months after diagnosis and the rate of mortality on long-term followup. Previous reports found that low concentrations of BNP and NT-proBNP had high negative predictive value for inhospital complications.3,13,14 However, in the present study, none of the 8 biomarkers demonstrated significant accuracy to discriminate between patients with submassive PE who survived with persistently normal RV function from patients who had either an abnormal RV or died within 6 months. This report adds significantly to existing literature because of its relatively large sample size, rigorous methodology of patient selection, attention to procedural details regarding the collection and processing of blood and echocardiograms, and inclusion of 6-month follow-up in the reference standard. Moreover, we unambiguously confirm that none of the 152 patients in this study had any missing data. We enrolled patients consecutively, 6 d/ wk, and we obtained written informed consent from a well-defined population of both inpatients and outpatients with submassive PE. This was not a convenience

American Heart Journal Volume 156, Number 2

Kline et al 313

Table V. Outcomes of patients at 6 months stratified by BNP or troponin I concentrations Outcome

BNP (≤100 pg/mL)

BNP (>100 pg/mL)

P⁎

Troponin I (≤0.1 ng/mL)

Troponin I (>0.1 ng/mL)

P⁎

3% 10% 24%

23% 9% 50%

.003 .821 .006

6% 11% 30%

13% 6% 31%

.205 .358 .936

Death Dyspnea at rest 6MWD b330 m ⁎From Fisher exact test.

sample. We explicitly required echocardiography and blood samples to be obtained either before or within 12 hours after heparin administration in all patients. Echocardiography was performed using a predefined image acquisition protocol, designed by a study author and echocardiography expert (GAR) to assess RV function in the setting of acute PE. We clearly specify that only 8 survivors were lost to follow-up. We performed preplanned 6-month follow-up on all patients who included questionnaires and a 6-minute walk test to assess functional outcomes, and we continued to follow the cohort for up to 5 years. Within the context of rigorous methodology, we performed a head-to-head comparison of the candidate biomarkers. In choosing these 8 biomarkers, we drew from published evidence linking each protein at least one mechanism of RV damage caused by PE. Troponin probably leaks into the plasma as a result of from the combined stresses of ischemia and mechanical shear and stretch on the RV muscle.4,15 In response to mechanical stretching, RV cardiomyocytes likely accelerate the transcription, translation, and secretion of natriuretic peptides, including BNP. The magnitude of the D-dimer concentration varies with thrombus burden, and we hypothesize that a larger thrombus will cause more pulmonary vascular occlusion, with higher RVafterload and more RV damage.16,17 Regarding MCP-1, our group has previously demonstrated that the RV content of MCP-1 protein increased 25-fold in a rat model of PE, and MCP-1 concentrations are increased in the blood of patients with pulmonary hypertension.4,18 Moreover, autopsy studies have demonstrated accumulation of monocytes and macrophages in the RVof humans who died of PE.19 In the setting of experimental PE, the RV content of MPO increased 100-fold in association with neutrophilic infiltration of the RV, and pharmacological blockade of neutrophil chemotaxis reduced RV damage in parallel with MPO concentrations in the RV.5 Souza-Costa et al20 have reported that treatments associated with decreased MMP-9 activity produced hemodynamic improvements in rats with experimental PE, and MMP-9 has been implicated in RV remodeling with chronic pulmonary hypertension in humans.21 Taken together, these data supported the inclusion of MPO and MMP-9 in this report. Regarding CRP, Chung et al22 demonstrated a positive statistical correlation between RV size on echocardiography and CRP concentration in humans with PE. Lastly, caspase 3 is a protease required for apoptosis. In view of electron microscopy that

Table VI. Cox proportional regression survival analysis Covariate BNP (>100 pg/mL) Troponin I (>0.1 ng/mL) RV hypokinesis on echocardiogram Age (per year)

Hazards ratio

95% CI

2.74 1.41 1.22 1.02

(1.07-6.96) (0.54-3.61) (0.34-1.98) (0.99-1.05)

One hundred fifty-one subjects with 27 deaths observed for a median follow-up period of 1,106 days. Deviance (likelihood ratio) χ2 = 8.82, df = 4, and P = .066.

demonstrated ultrastructural processes often associated with apoptosis, we hypothesized that caspase 3 would signal the presence of RV damage. This study has several potential limitations. The main limitation is that we did not find that any biomarker to produce diagnostic performance to use as an exclusionary criterion for significant adverse outcomes. Although BNP and troponin I advantaged the other 6 biomarkers, neither protein demonstrated good prognostic performance, if we assume that good performance requires an area under the receiver operating characteristic curve >0.80. In addition, we did not study all potential biomarkers of RV damage, including N-terminal pro-brain natriuretic peptide, P-selectin, or fatty acid binding protein, each of which may have had better prognostic accuracy. In conclusion, we compared of 8 plasma proteins as potential biomarkers of clinically significant RV damage after submassive PE. Both BNP and troponin I demonstrated significant prognostic accuracy to predict RV hypokinesis. Patients with BNP concentrations >100 pg/mL had increased frequency of mortality and exercise intolerance 6 months after diagnosis and increased risk of dying on long-term follow-up. We thank Courtney Peterson for her assistance.

References 1. Stevinson BG, Hernandez-Nino J, Rose G, et al. Echocardiographic and functional cardiopulmonary problems 6 months after first-time pulmonary embolism in previously healthy patients. Eur Heart J 2007; 28:2517-24. 2. Becattini C, Vedovati MC, Agnelli G. Prognostic value of troponins in acute pulmonary embolism: a meta-analysis. Circulation 2007;116: 427-33.

314 Kline et al

3. Kucher N, Printzen G, Goldhaber SZ. Prognostic role of brain natriuretic peptide in acute pulmonary embolism. Circulation 2003; 107:2545-7. 4. Watts JA, Zagorski J, Gellar MA, et al. Cardiac inflammation contributes to right ventricular dysfunction following experimental pulmonary embolism in rats. J Mol Cell Cardiol 2006;41:296-307 Ref Type: Abstract. 5. Zagorski J, Gellar MA, Obraztsova M, et al. Inhibition of CINC-1 decreases right ventricular damage caused by experimental pulmonary embolism in rats. J Immunol 2007;179:7820-6. 6. Kaczynska A, Szulc M, Styczynski G, et al. Right ventricle injury during acute pulmonary embolism leads to its remodeling. Int J Cardiol 2008;125:120-1. 7. Jones AE, Tayal VS, Sullivan DM, et al. Randomized controlled trial of immediate versus delayed ultrasound to identify the cause of nontraumatic hypotension in emergency department patients. Crit Care Med 2004;32:1703-8. 8. Kline JA, Hernandez J, Rose G, et al. Surrogate markers for adverse outcomes in normotensive patients with pulmonary embolism. Crit Care Med 2006;34:2180-773. 9. Crapo RO, Enright PL, Zeballos RJ. ATS statement: Guidelines for the six-minute walk test. Am J Respir Crit Care Med 2002;166:111-7. 10. McConnell MV, Solomon SD, Rayan ME, et al. Regional right ventricular dysfunction detected by echocardiography in acute pulmonary embolism. Am J Cardiol 1996;78:469-73. 11. Mastora I, Remy-Jardin M, Masson P, et al. Severity of acute pulmonary embolism: evaluation of a new spiral CT angiographic score in correlation with echocardiographic data. Eur Radiol 2003; 13:29-35. 12. ten Wolde M, Sohne M, Quak E, et al. Prognostic value of echocardiographically assessed right ventricular dysfunction in patients with pulmonary embolism [Review] [31 refs]. Arch Intern Med 2004;164:1685-9.

American Heart Journal August 2008

13. Kucher N, Printzen G, Doernhoefer T, et al. Low pro-brain natriuretic peptide levels predict benign clinical outcome in acute pulmonary embolism. Circulation 2003;107:1576-8. 14. Binder L, Pieske B, Olschewski M, et al. N-terminal pro-brain natriuretic peptide or troponin testing followed by echocardiography for risk stratification of acute pulmonary embolism. Circulation 2005; 112:1573-9. 15. Gold FL, Bache RJ. Transmural right ventricular blood flow during acute pulmonary artery hypertension in the sedated dog. Circ Res 1982;51:196-204. 16. Galle C, Papazyan J, Miron M, et al. Prediction of pulmonary embolism extent by clinical findings, D-dimer level and deep vein thrombosis shown by ultrasound. Thromb Haemost 2001;86: 1156. 17. Ghanima W, Abdelnoor M, Holmen LO, et al. D-dimer level is associated with the extent of pulmonary embolism. Thromb Res 2007; 120:281-8. 18. Kimura H, Okada O, Tanabe N, et al. Plasma monocyte chemoattractant protein-1 and pulmonary vascular resistance in chronic thromboembolic pulmonary hypertension. Am J Respir Crit Care Med 2001;164:319-24. 19. Iwadate K, Doi M, Tanno K, et al. Right ventricular damage due to pulmonary embolism: examination of the number of infiltrating macrophages. Forensic Sci Int 2003;134:147-53. 20. Souza-Costa DC, Figueiredo-Lopes L, ves-Filho JC, et al. Protective effects of atorvastatin in rat models of acute pulmonary embolism: involvement of matrix metalloproteinase-9. Crit Care Med 2007;35: 239-45. 21. Cantini-Salignac C, Lartaud I, Schrijen F, et al. Metalloproteinase-9 in circulating monocytes in pulmonary hypertension. Fundam Clin Pharmacol 2006;20:405-10. 22. Chung T, Connor D, Joseph J, et al. Platelet activation in acute pulmonary embolism. J Thromb Haemost 2007;5:918-24.