Original Article doi: 10.1111/joim.12316
Cardiac biomarkers and clinical scores for risk stratification in elderly patients with non-high-risk pulmonary embolism N. Vuilleumier1, A. Limacher2, M. Mean3, J. Choffat1, P. Lescuyer1, H. Bounameaux4, D. Aujesky3 & M. Righini4 From the 1Faculty of Medicine, Division of Laboratory Medicine, Department of Genetics and Laboratory Medicine, Geneva University Hospital, Geneva; 2Clinical Trials Unit (CTU) Bern, Department of Clinical Research and Institute of Social and Preventive Medicine (ISPM), University of Bern; 3Division of General Internal Medicine, Bern University Hospital and University of Bern, Bern; and 4Faculty of Medicine, Division of Angiology and Haemostasis, Geneva University Hospital, Geneva, Switzerland
Abstract. Vuilleumier N, Limacher A, M ean M, Choffat J, Lescuyer P, Bounameaux H, Aujesky D, Righini M (Geneva University Hospital, Geneva; University of Bern; Bern University Hospital and University of Bern, Bern; and Geneva University Hospital, Geneva, Switzerland). Cardiac biomarkers and clinical scores for risk stratification in elderly patients with non-highrisk pulmonary embolism. J Intern Med 2014; doi:10.1111/joim.12316. Objective. To determine the prognostic accuracy of cardiac biomarkers alone and in combination with clinical scores in elderly patients with non-highrisk pulmonary embolism (PE). Design. Ancillary analysis of a Swiss multicentre prospective cohort study. Subjects. A total of 230 patients aged ≥65 years with non-high-risk PE. Main outcome measures. The study end-point was a composite of PE-related complications, defined as PE-related death, recurrent venous thromboembolism or major bleeding during a follow-up of 30 days. The prognostic accuracy of the Pulmonary Embolism Severity Index (PESI), the Geneva Prognostic Score (GPS), the precursor of brain
Introduction The vast majority of patients (>90%) with pulmonary embolism (PE) have non-high-risk PE, defined as PE with normal blood pressure [1]. Although ambulatory treatment of patients with non-highrisk PE was shown to be effective and safe in both nonrandomized and randomized prospective trials [2–4], management of this condition is still predominantly inpatient based because of a highly heterogeneous 30-day mortality rate (between II for the PESI and a cut-off value of >2 points for the GPS to define high- versus low-risk PE patients, according to commonly accepted definitions [4, 9]. For the GPS, we used arterial oxygen saturation 2 points (25.2% vs. 8.0% for ≤2, P = 0.04). No difference in complication rates was observed between patients with PESI risk class I/II and those with class III/V (P = 0.58). ROC analyses ROC curve analyses demonstrated that hs-cTnT had the highest prognostic accuracy for PE-related complications, with an AUC of 0.75 (95% CI: 0.63– 0.86, P < 0.001), followed by NT-proBNP (0.69, 95% CI: 0.58–0.79, P = 0.003) and the GPS (0.64, 95% CI: 0.51–0.78, P = 0.01) (Table 3). The PESI score was not a significant predictor of a complicated outcome. According to the method of Delong et al., the only significant difference in the AUC
Cumulative incidence of 30−days PE−related complications (%)
At the pre-specified cut-off value, NT-proBNP had a sensitivity of 95%, a specificity of 34.3%, a positive predictive value of 12.1% and a negative predictive value of 98.6% (Table 3). For hs-cTnT, the sensitivity, specificity and positive and negative predictive values were 80.0%, 44.8%, 12.1% and 95.9%, respectively, and the corresponding values for the GPS were 20%, 92.9%, 21.1% and 92.4% (Table 3). When 30-day mortality was considered as the endpoint in this study, all the potential prognostic indicators presented in this study had significant discriminatory abilities, with AUC values of 0.69 for NT-proBNP and the GPS, 0.73 for hs-cTnT and 0.76 for the PESI. Reclassifications statistics When biomarkers were compared to clinical scores, the IDI was 2.5% for NT-proBNP and the PESI (P = 0.03), 0.7% for NT-proBNP and the GPS (P = 0.73), 6.4% for hs-cTnT and the PESI (P = 0.03) and 3.2% for hs-cTnT and the PESI (P = 0.31) (Table 4). Furthermore, the IDI was 3.9% when hs-cTnT was compared to NT-proBNP (P = 0.20). Overall, these results indicate that both NT-proBNP and hs-cTnT provide a modest but significant incremental prognostic value over the PESI, but not over the GPS. AUC comparisons and reclassification statistics
15
hs−cTnT > 14 pg mL–1 hs−cTnT ≤ 14 pg mL–1 NT−proBNP > 300 pg mL–1 NT−proBNP ≤ 300 pg mL–1
10
5
0 0
10
20
30
Follow−up (days)
Fig. 1 PE-related complications according to hs-cTnT and NT-proBNP values during follow-up. hs-cTnT, high-sensitivity cardiac troponin T; NT-proBNP, N-terminal pro-brain natriuretic peptide. PE, pulmonary embolism. 6
was observed between hs-cTnT and the PESI (P = 0.01; Table 4).
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
Pairwise comparisons demonstrated that the addition of hs-cTnT to a clinical score significantly improved the AUC of the PESI by 19%, but did not significantly affect the AUC of the GPS (Table 4). On the other hand, the combination of NT-proBNP with either clinical score did not translate into a gain in prognostic accuracy, and the combination of hscTnT and NT-proBNP did not provide any significant increase in AUC compared to NT-proBNP alone (Table 4). Reclassification statistics indicated that adding hs-cTnT to either clinical score could improve the prognostic performance of the clinical score alone. The addition of hs-cTnT to the PESI and the GPS improved the risk reclassification by 6.4% (P = 0.03) and 7.7% (P = 0.03), respectively. NTproBNP added to the clinical scores provided a modest but significant increase compared to the PESI alone (+2.6%; P = 0.03), but not when com-
N. Vuilleumier et al.
Cardiac biomarkers and clinical scores
Table 3 Prognostic accuracy of cardiac biomarkers and clinical scores for detecting PE-related complications at 30 days AUCa (95% CI)
Sensitivity
Specificity
Positive predictive
Predictors alone
P-value
(95% CI)
(95% CI)
value (95% CI)
value (95% CI)
NT-proBNP
0.69 (0.58–0.79)
95.0 (76.4–99.1)
34.3 (28.2–40.9)
12.1 (7.9–18.1)
98.6 (92.6–99.8)
80 (58.4–91.9)
44.8 (38.2–51.5)
12.1 (7.6–18.8)
95.9 (90.0–98.4)
70.0 (48.1–85.5)
35.7 (29.5–42.4)
9.4 (5.7–15.2)
92.6 (84.8–96.6)
20.0 (8.1–42.6)
92.9 (88.6–95.6)
21.1 (8.5–43.3)
92.4 (88.0–95.3)
Sensitivity
Specificity
Positive predictive
Negative predictive
Cut-off 300 pg mL
1
Cut-off 14 pg mL
0.003 0.75 (0.63–0.86)
hs-cTnT 1
PESI Cut-off class II GPS Cut-off 2 points
Negative predictive
95% for PE-related complications. As such, both biomarkers are potentially useful for identifying patients with very low-risk PE who are likely to benefit from ambulatory treatment, as demonstrated for NT-proBNP in a prospective multicentre study of patients without age restriction [6]. Although differences in AUC and negative
predictive values were small between NT-proBNP and hs-cTnT, the proportion of patients identified as low risk was significantly lower for NT-proBNP (31.7% vs. 42.6%, P = 0.02). Thus, hs-cTnT may be more appropriate for identifying low-risk patients with nonmassive PE who could benefit from less costly outpatient treatment, without compromising safety. This possibility needs to be assessed in a prospective management study. Another finding of this study is that both NT-proBNP and hs-cTnT appeared to be superior to the PESI score to predict 30-day complications, but did not outperform the GPS. Furthermore, reclassification statistics indicated that the addition of hs-cTnT to both the GPS and the PESI significantly improved the 30-day risk prediction of both clinical scores by 7.7% and 6.4%, respectively. This suggests that the prognostic accuracy of cardiac biomarkers in the elderly is at least equivalent to that of clinical scores and that the addition of hs-cTnT to clinical scores improves the efficacy of currently used clinically based risk-stratification models. Nevertheless, this combination did not improve the positive predictive values of the models and did not improve the negative predictive value of NT-proBNP alone. Thus, ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
7
N. Vuilleumier et al.
Cardiac biomarkers and clinical scores
Table 4 Area under the curve comparisons and reclassification statistics Mean predicted risk Predictor
PE-related complications
NT-proBNP >300 pg mL hs-cTnT >14 pg mL
1
1
GPS >2 points PESI risk class >II Predictor alone
No complications
11.0%
8.5%
14.6%
8.1%
11.6%
8.4%
8.8%
8.7%
AUC comparison D AUC; P-value
Improvement in predicted risk IDI value (95% CI; P-value)
NT-ProBNP vs. PESI
+0.13;0.08
+2.5% (0.3 to 4.7; 0.03)
NT-proBNP vs. GPS
+0.04; 0.60
–0.7% (–4.3 to 3.0; 0.73)
hs-cTnT vs. PESI
+0.19;0.01
+6.4% (0.6 to 12.2; 0.03)
hs-cTnT vs. GPS
+0.11;0.21
+3.2% (-3.0 to 9.5; 0.31)
NT-proBNP vs. hs-cTnT Combined predictors
+0.06;0.13
+3.9 (–2.0 to 9.8; 0.20)
AUC comparison D AUC; P-value
(biomarker added to
Improvement in predicted risk IDI (95% CI; P-value)
clinical scores) GPS vs. GPS+NT-proBNP
+0.07; 0.18
PESI vs. PESI+NT-proBNP
+ 0.13; 0.13
+1.5% (–0.05 to 3.5; 0.15) + 2.6% (0.3 to 4.9; 0.03)
GPS vs. GPS+hs-cTnT
+0.09; 0.07
+7.7% (0.7 to 14.7; 0.03)
PESI vs. PESI+hs-cTnT
+0.19; 0.01
+6.4 (0.6 to 12.2; 0.03)
NT-proBNP vs. hs-cTnT+NT-proBNP
+0.06; 0.07
+5% (0.1 to 10.5; 0.04)
AUC, area under the curve; IDI, integrated discrimination improvement; CI, confidence interval; hs-cTnT, high-sensitivity cardiac troponin T; NT-proBNP, N-terminal pro-brain natriuretic peptide.
whether the combination of NT-proBNP with clinical scores could be useful for ‘rule-out’ purposes requires further clarification. The present study has several limitations. First, a considerable number of patients were excluded because blood samples were taken more than 1 day after diagnosis, which may have introduced a selection bias in this study and explains the relatively small number of patients enrolled in this ancillary study. Secondly, we found that the PESI was not a significant predictor of PE-related complications in this study. However, this is not surprising, given that this score was originally developed to predict 30-day overall mortality rather than the composite outcome used in this study of PE-related death, recurrence of venous thromboembolism or major bleeding within 30 days of PE diagnosis. Indeed, it was previously demonstrated that the PESI accurately predicts 30day overall mortality in elderly patients with PE, with an AUC of 0.77 [15]. Although major bleeding is a common component of the composite end-point definition in PE [6, 9, 10, 12], the high proportion of major bleeding events observed here as part of the 8
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
composite study end-point could have confounded our results. For this reason, we performed additional analyses in parallel considering 30-day mortality as the end-point, which confirmed that all the candidate markers/scores presented in this study had the expected significant discriminatory abilities, with AUC values of 0.69 for NT-proBNP and GPS, 0.73 for hs-cTnT and 0.76 for the PESI (data not shown). Another open question concerns the optimal NT-proBNP cut-off value to be chosen, varying among different studies between 300 and 1000 pg mL 1, according to: (i) whether it was used to identify intermediate or low-risk patients, and (ii) the way it has been defined (prospectively or retrospectively) [6, 9–11, 26]. As the aim of this study was to identify very low-risk patients focusing on the negative predictive value, we used the previously prospectively validated cut-off level of 300 pg mL 1 for similar use in non-high-risk PE patients, [10] rather than a higher value. Furthermore, this cut-off is currently used routinely in clinical settings to exclude heart failure in dyspnoeic patients [20], therefore, using the same value for PE risk stratification would markedly
N. Vuilleumier et al.
simplify the interpretation of NT-proBNP levels in daily clinical practice. Finally, we did not test other promising potential biomarkers, such as hearttype fatty acid-binding protein, copeptin, midregion pro-adrenomedullin and cystatin C, in elderly patients with nonmassive PE; therefore, further studies are warranted on this topic. In conclusion, in elderly patients with non-highrisk PE, both NT-proBNP and hs-cTnT are independently associated with a composite end-point of PE-related death, recurrence of venous thromboembolism or major bleeding within 30 days of PE diagnosis, with a negative predictive value of >95% for this end-point. Hs-cTnT appears to identify a higher proportion of low-risk patients with nonmassive PE than NT-proBNP and was slightly better able to discriminate between patients with and without 30-day complications than NT-proBNP and the clinical scores. Finally, adding hs-cTnT to the GPS or the PESI significantly improved the prognostic accuracy of both clinical scores. Whether hs-cTnT could be an adequate alternative to clinical scores for identifying low-risk patients with PE who could be treated in an outpatient setting needs to be prospectively studied. Conflict of interest statement Hoffmann-La Roche (Rotkreuz, Switzerland) provided support for some of the reagent costs, but was not involved in designing the study, data analysis or interpretation or writing the manuscript. The authors have no other conflict of interests to disclose. Acknowledgements This work was supported by a grant from the Swiss National Science Foundation (no. 33CSO-122659/ 139470).
References 1 Torbicki A, Perrier A, Konstantinides S et al.; ESC Committee for Practice Guidelines (CPG). Guidelines on the diagnosis and management of acute pulmonary embolism: the Task Force for the Diagnosis and Management of Acute Pulmonary Embolism of the European Society of Cardiology (ESC). Eur Heart J 2008; 29: 2276–315. 2 Kovacs MJ, Anderson D, Morrow B, Gray L, Touchie D, Wells PS. Outpatient treatment of pulmonary embolism with deltaparin. Thromb Haemost 2000; 83: 209–11.
Cardiac biomarkers and clinical scores
3 Beer JH, Burger M, Gretener S, Bernard Bagattini S, Bounameaux H. Outpatients treatment of pulmonary embolism is feasible and safe in a substantial proportion of patients. J Thromb Haemost 2003; 1: 186–7. 4 Aujesky D, Roy PM, Verschuren F et al. Outpatient versus inpatient treatment for patients with acute pulmonary embolism: an international, open-label, randomised, non-inferiority trial. Lancet 2011; 378: 41–8. 5 Goldhaber SZ, Eliott CG. Acute pulmonary embolism, II: risk stratification, treatment, and prevention. Circulation 2003; 108: 2834–8. 6 Agterof MJ, Schutgens RE, Snijder RJ et al. Out of hospital treatment of acute pulmonary embolism in patients with a low NT-proBNP level. J Thromb Haemost 2010; 8: 1235–41. 7 Aujesky D, Smith KJ, Cornuz J, Roberts MS. Cost effectiveness of low molecular heparin for treatment of pulmonary embolism. Chest 2005; 128: 1601–10. 8 Nendaz MR, Bandelier P, Aujesky D et al. Validation of a risk score identifying patients with acute pulmonary embolism, who are at low risk of clinical adverse outcome. Thromb Haemost 2004; 91: 1232–6. 9 Coutance G, Le Page O, Lo T, Hamon M. Prognostic value of brain natriuretic peptide in acute pulmonary embolism. Crit Care 2008; 12: R109. 10 Vuilleumier N, Le Gal G, Verschuren F et al. Cardiac biomarkers for risk stratification in non-massive pulmonary embolism: a multicenter prospective study. J Thromb Haemost 2009; 7: 391–8. 11 Lankeit M, Friesen D, Aschoff J et al. Highly sensitive troponin T assay in normotensive patients with acute pulmonary embolism. Eur Heart J 2010; 31: 1836–44. 12 Lankeit M, Jim enez D, Kostrubiec M et al. Predictive value of the high-sensitivity troponin T assay and the simplified Pulmonary Embolism Severity Index in hemodynamically stable patients with acute pulmonary embolism: a prospective validation study. Circulation 2011; 124: 2716–24. 13 Loke I, Squire IB, Davies JE, Ng LL. Reference ranges for natriuretic peptides for diagnostic use are dependent on age, gender and heart rate. Eur J Heart Fail 2003; 5: 599–606. 14 Chenevier-Gobeaux C, Meune C, Freund Y et al. Influence of age and renal function on high-sensitivity cardiac troponin T diagnostic accuracy for the diagnosis of acute myocardial infarction. Am J Cardiol 2013; 111: 1701–7. 15 Zwierzina D, Limacher A, M ean M et al. Prospective comparison of clinical prognostic scores in elderly patients with pulmonary embolism. J Thromb Haemost 2012; 10: 2270–6. 16 M ean M, Righini M, Jaeger K et al. The Swiss cohort of elderly patients with venous thromboembolism (SWITCO65 + ): rationale and methodology. J Thromb Thrombolysis 2013; 36: 484–91. 17 Wicki J, Perrier A, Perneger TV, Bounameaux H, Junod AF. Predicting adverse outcome in patients with acute pulmonary embolism: a risk score. Thromb Haemost 2000; 84: 548–52. 18 Apple FS, Jaffe AS. Clinical implications of a recent adjustment to the high-sensitivity cardiac troponin T assay: user beware. Clin Chem 2012; 58: 1599–600. 19 Thygesen K, Alpert JS, Jaffe AS, Simoons ML, Chaitman BR, White HD; Joint ESC/ACCF/AHA/WHF Task Force for Universal Definition of Myocardial Infarction. Third universal
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
9
N. Vuilleumier et al.
20
21
22
23
10
definition of myocardial infarction. J Am Coll Cardiol 2012; 60: 1581–98. Januzzi JL Jr, Camargo CA, Anwaruddin S et al. The N-terminal Pro-BNP investigation of dyspnea in the emergency department (PRIDE) study. Am J Cardiol 2005; 15: 948–54. deFilippi C, Seliger SL, Kelley W et al. Interpreting cardiac troponin results from high-sensitivity assays in chronic kidney disease without acute coronary syndrome. Clin Chem 2012; 58: 1342–51. DeFilippi CR, Fink JC, Nass CM, Chen H, Christenson R. N-terminalpro-B-type natriuretic peptide for predicting coronary disease and leftventricular hypertrophy in asymptomatic CKD not requiring dialysis. Am J Kidney Dis 2005; 46: 35–44. Zoccali C, Mallamaci F, Benedetto FA et al. Cardiac natriuretic peptides are related to left ventricular mass andfunction and predict mortality in dialysis patients. J Am Soc Nephrol 2001; 12: 1508–15.
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
Cardiac biomarkers and clinical scores
24 DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: 837–45. 25 Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008; 27: 157–72. 26 Lankeit M, Jim enez D, Kostrubiec M et al. Validation of N-terminal pro-brain natriuretic peptide cut-off values for risk stratification of pulmonary embolism. Eur Respir J 2014; 43: 1669–77. Correspondence: Nicolas Vuilleumier, Department of Genetics and Laboratory Medicine, Geneva University Hospitals, 4 rue Gabrielle Perret-Gentil 1211 Geneva, Switzerland. (fax: +41223727399; e-mail: [email protected]).
Cardiac biomarkers and clinical scores for risk stratification in elderly patients with non-high-risk pulmonary embolism N. Vuilleumier1, A. Limacher2, M. Mean3, J. Choffat1, P. Lescuyer1, H. Bounameaux4, D. Aujesky3 & M. Righini4 From the 1Faculty of Medicine, Division of Laboratory Medicine, Department of Genetics and Laboratory Medicine, Geneva University Hospital, Geneva; 2Clinical Trials Unit (CTU) Bern, Department of Clinical Research and Institute of Social and Preventive Medicine (ISPM), University of Bern; 3Division of General Internal Medicine, Bern University Hospital and University of Bern, Bern; and 4Faculty of Medicine, Division of Angiology and Haemostasis, Geneva University Hospital, Geneva, Switzerland
Abstract. Vuilleumier N, Limacher A, M ean M, Choffat J, Lescuyer P, Bounameaux H, Aujesky D, Righini M (Geneva University Hospital, Geneva; University of Bern; Bern University Hospital and University of Bern, Bern; and Geneva University Hospital, Geneva, Switzerland). Cardiac biomarkers and clinical scores for risk stratification in elderly patients with non-highrisk pulmonary embolism. J Intern Med 2014; doi:10.1111/joim.12316. Objective. To determine the prognostic accuracy of cardiac biomarkers alone and in combination with clinical scores in elderly patients with non-highrisk pulmonary embolism (PE). Design. Ancillary analysis of a Swiss multicentre prospective cohort study. Subjects. A total of 230 patients aged ≥65 years with non-high-risk PE. Main outcome measures. The study end-point was a composite of PE-related complications, defined as PE-related death, recurrent venous thromboembolism or major bleeding during a follow-up of 30 days. The prognostic accuracy of the Pulmonary Embolism Severity Index (PESI), the Geneva Prognostic Score (GPS), the precursor of brain
Introduction The vast majority of patients (>90%) with pulmonary embolism (PE) have non-high-risk PE, defined as PE with normal blood pressure [1]. Although ambulatory treatment of patients with non-highrisk PE was shown to be effective and safe in both nonrandomized and randomized prospective trials [2–4], management of this condition is still predominantly inpatient based because of a highly heterogeneous 30-day mortality rate (between II for the PESI and a cut-off value of >2 points for the GPS to define high- versus low-risk PE patients, according to commonly accepted definitions [4, 9]. For the GPS, we used arterial oxygen saturation 2 points (25.2% vs. 8.0% for ≤2, P = 0.04). No difference in complication rates was observed between patients with PESI risk class I/II and those with class III/V (P = 0.58). ROC analyses ROC curve analyses demonstrated that hs-cTnT had the highest prognostic accuracy for PE-related complications, with an AUC of 0.75 (95% CI: 0.63– 0.86, P < 0.001), followed by NT-proBNP (0.69, 95% CI: 0.58–0.79, P = 0.003) and the GPS (0.64, 95% CI: 0.51–0.78, P = 0.01) (Table 3). The PESI score was not a significant predictor of a complicated outcome. According to the method of Delong et al., the only significant difference in the AUC
Cumulative incidence of 30−days PE−related complications (%)
At the pre-specified cut-off value, NT-proBNP had a sensitivity of 95%, a specificity of 34.3%, a positive predictive value of 12.1% and a negative predictive value of 98.6% (Table 3). For hs-cTnT, the sensitivity, specificity and positive and negative predictive values were 80.0%, 44.8%, 12.1% and 95.9%, respectively, and the corresponding values for the GPS were 20%, 92.9%, 21.1% and 92.4% (Table 3). When 30-day mortality was considered as the endpoint in this study, all the potential prognostic indicators presented in this study had significant discriminatory abilities, with AUC values of 0.69 for NT-proBNP and the GPS, 0.73 for hs-cTnT and 0.76 for the PESI. Reclassifications statistics When biomarkers were compared to clinical scores, the IDI was 2.5% for NT-proBNP and the PESI (P = 0.03), 0.7% for NT-proBNP and the GPS (P = 0.73), 6.4% for hs-cTnT and the PESI (P = 0.03) and 3.2% for hs-cTnT and the PESI (P = 0.31) (Table 4). Furthermore, the IDI was 3.9% when hs-cTnT was compared to NT-proBNP (P = 0.20). Overall, these results indicate that both NT-proBNP and hs-cTnT provide a modest but significant incremental prognostic value over the PESI, but not over the GPS. AUC comparisons and reclassification statistics
15
hs−cTnT > 14 pg mL–1 hs−cTnT ≤ 14 pg mL–1 NT−proBNP > 300 pg mL–1 NT−proBNP ≤ 300 pg mL–1
10
5
0 0
10
20
30
Follow−up (days)
Fig. 1 PE-related complications according to hs-cTnT and NT-proBNP values during follow-up. hs-cTnT, high-sensitivity cardiac troponin T; NT-proBNP, N-terminal pro-brain natriuretic peptide. PE, pulmonary embolism. 6
was observed between hs-cTnT and the PESI (P = 0.01; Table 4).
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
Pairwise comparisons demonstrated that the addition of hs-cTnT to a clinical score significantly improved the AUC of the PESI by 19%, but did not significantly affect the AUC of the GPS (Table 4). On the other hand, the combination of NT-proBNP with either clinical score did not translate into a gain in prognostic accuracy, and the combination of hscTnT and NT-proBNP did not provide any significant increase in AUC compared to NT-proBNP alone (Table 4). Reclassification statistics indicated that adding hs-cTnT to either clinical score could improve the prognostic performance of the clinical score alone. The addition of hs-cTnT to the PESI and the GPS improved the risk reclassification by 6.4% (P = 0.03) and 7.7% (P = 0.03), respectively. NTproBNP added to the clinical scores provided a modest but significant increase compared to the PESI alone (+2.6%; P = 0.03), but not when com-
N. Vuilleumier et al.
Cardiac biomarkers and clinical scores
Table 3 Prognostic accuracy of cardiac biomarkers and clinical scores for detecting PE-related complications at 30 days AUCa (95% CI)
Sensitivity
Specificity
Positive predictive
Predictors alone
P-value
(95% CI)
(95% CI)
value (95% CI)
value (95% CI)
NT-proBNP
0.69 (0.58–0.79)
95.0 (76.4–99.1)
34.3 (28.2–40.9)
12.1 (7.9–18.1)
98.6 (92.6–99.8)
80 (58.4–91.9)
44.8 (38.2–51.5)
12.1 (7.6–18.8)
95.9 (90.0–98.4)
70.0 (48.1–85.5)
35.7 (29.5–42.4)
9.4 (5.7–15.2)
92.6 (84.8–96.6)
20.0 (8.1–42.6)
92.9 (88.6–95.6)
21.1 (8.5–43.3)
92.4 (88.0–95.3)
Sensitivity
Specificity
Positive predictive
Negative predictive
Cut-off 300 pg mL
1
Cut-off 14 pg mL
0.003 0.75 (0.63–0.86)
hs-cTnT 1
PESI Cut-off class II GPS Cut-off 2 points
Negative predictive
95% for PE-related complications. As such, both biomarkers are potentially useful for identifying patients with very low-risk PE who are likely to benefit from ambulatory treatment, as demonstrated for NT-proBNP in a prospective multicentre study of patients without age restriction [6]. Although differences in AUC and negative
predictive values were small between NT-proBNP and hs-cTnT, the proportion of patients identified as low risk was significantly lower for NT-proBNP (31.7% vs. 42.6%, P = 0.02). Thus, hs-cTnT may be more appropriate for identifying low-risk patients with nonmassive PE who could benefit from less costly outpatient treatment, without compromising safety. This possibility needs to be assessed in a prospective management study. Another finding of this study is that both NT-proBNP and hs-cTnT appeared to be superior to the PESI score to predict 30-day complications, but did not outperform the GPS. Furthermore, reclassification statistics indicated that the addition of hs-cTnT to both the GPS and the PESI significantly improved the 30-day risk prediction of both clinical scores by 7.7% and 6.4%, respectively. This suggests that the prognostic accuracy of cardiac biomarkers in the elderly is at least equivalent to that of clinical scores and that the addition of hs-cTnT to clinical scores improves the efficacy of currently used clinically based risk-stratification models. Nevertheless, this combination did not improve the positive predictive values of the models and did not improve the negative predictive value of NT-proBNP alone. Thus, ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
7
N. Vuilleumier et al.
Cardiac biomarkers and clinical scores
Table 4 Area under the curve comparisons and reclassification statistics Mean predicted risk Predictor
PE-related complications
NT-proBNP >300 pg mL hs-cTnT >14 pg mL
1
1
GPS >2 points PESI risk class >II Predictor alone
No complications
11.0%
8.5%
14.6%
8.1%
11.6%
8.4%
8.8%
8.7%
AUC comparison D AUC; P-value
Improvement in predicted risk IDI value (95% CI; P-value)
NT-ProBNP vs. PESI
+0.13;0.08
+2.5% (0.3 to 4.7; 0.03)
NT-proBNP vs. GPS
+0.04; 0.60
–0.7% (–4.3 to 3.0; 0.73)
hs-cTnT vs. PESI
+0.19;0.01
+6.4% (0.6 to 12.2; 0.03)
hs-cTnT vs. GPS
+0.11;0.21
+3.2% (-3.0 to 9.5; 0.31)
NT-proBNP vs. hs-cTnT Combined predictors
+0.06;0.13
+3.9 (–2.0 to 9.8; 0.20)
AUC comparison D AUC; P-value
(biomarker added to
Improvement in predicted risk IDI (95% CI; P-value)
clinical scores) GPS vs. GPS+NT-proBNP
+0.07; 0.18
PESI vs. PESI+NT-proBNP
+ 0.13; 0.13
+1.5% (–0.05 to 3.5; 0.15) + 2.6% (0.3 to 4.9; 0.03)
GPS vs. GPS+hs-cTnT
+0.09; 0.07
+7.7% (0.7 to 14.7; 0.03)
PESI vs. PESI+hs-cTnT
+0.19; 0.01
+6.4 (0.6 to 12.2; 0.03)
NT-proBNP vs. hs-cTnT+NT-proBNP
+0.06; 0.07
+5% (0.1 to 10.5; 0.04)
AUC, area under the curve; IDI, integrated discrimination improvement; CI, confidence interval; hs-cTnT, high-sensitivity cardiac troponin T; NT-proBNP, N-terminal pro-brain natriuretic peptide.
whether the combination of NT-proBNP with clinical scores could be useful for ‘rule-out’ purposes requires further clarification. The present study has several limitations. First, a considerable number of patients were excluded because blood samples were taken more than 1 day after diagnosis, which may have introduced a selection bias in this study and explains the relatively small number of patients enrolled in this ancillary study. Secondly, we found that the PESI was not a significant predictor of PE-related complications in this study. However, this is not surprising, given that this score was originally developed to predict 30-day overall mortality rather than the composite outcome used in this study of PE-related death, recurrence of venous thromboembolism or major bleeding within 30 days of PE diagnosis. Indeed, it was previously demonstrated that the PESI accurately predicts 30day overall mortality in elderly patients with PE, with an AUC of 0.77 [15]. Although major bleeding is a common component of the composite end-point definition in PE [6, 9, 10, 12], the high proportion of major bleeding events observed here as part of the 8
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
composite study end-point could have confounded our results. For this reason, we performed additional analyses in parallel considering 30-day mortality as the end-point, which confirmed that all the candidate markers/scores presented in this study had the expected significant discriminatory abilities, with AUC values of 0.69 for NT-proBNP and GPS, 0.73 for hs-cTnT and 0.76 for the PESI (data not shown). Another open question concerns the optimal NT-proBNP cut-off value to be chosen, varying among different studies between 300 and 1000 pg mL 1, according to: (i) whether it was used to identify intermediate or low-risk patients, and (ii) the way it has been defined (prospectively or retrospectively) [6, 9–11, 26]. As the aim of this study was to identify very low-risk patients focusing on the negative predictive value, we used the previously prospectively validated cut-off level of 300 pg mL 1 for similar use in non-high-risk PE patients, [10] rather than a higher value. Furthermore, this cut-off is currently used routinely in clinical settings to exclude heart failure in dyspnoeic patients [20], therefore, using the same value for PE risk stratification would markedly
N. Vuilleumier et al.
simplify the interpretation of NT-proBNP levels in daily clinical practice. Finally, we did not test other promising potential biomarkers, such as hearttype fatty acid-binding protein, copeptin, midregion pro-adrenomedullin and cystatin C, in elderly patients with nonmassive PE; therefore, further studies are warranted on this topic. In conclusion, in elderly patients with non-highrisk PE, both NT-proBNP and hs-cTnT are independently associated with a composite end-point of PE-related death, recurrence of venous thromboembolism or major bleeding within 30 days of PE diagnosis, with a negative predictive value of >95% for this end-point. Hs-cTnT appears to identify a higher proportion of low-risk patients with nonmassive PE than NT-proBNP and was slightly better able to discriminate between patients with and without 30-day complications than NT-proBNP and the clinical scores. Finally, adding hs-cTnT to the GPS or the PESI significantly improved the prognostic accuracy of both clinical scores. Whether hs-cTnT could be an adequate alternative to clinical scores for identifying low-risk patients with PE who could be treated in an outpatient setting needs to be prospectively studied. Conflict of interest statement Hoffmann-La Roche (Rotkreuz, Switzerland) provided support for some of the reagent costs, but was not involved in designing the study, data analysis or interpretation or writing the manuscript. The authors have no other conflict of interests to disclose. Acknowledgements This work was supported by a grant from the Swiss National Science Foundation (no. 33CSO-122659/ 139470).
References 1 Torbicki A, Perrier A, Konstantinides S et al.; ESC Committee for Practice Guidelines (CPG). Guidelines on the diagnosis and management of acute pulmonary embolism: the Task Force for the Diagnosis and Management of Acute Pulmonary Embolism of the European Society of Cardiology (ESC). Eur Heart J 2008; 29: 2276–315. 2 Kovacs MJ, Anderson D, Morrow B, Gray L, Touchie D, Wells PS. Outpatient treatment of pulmonary embolism with deltaparin. Thromb Haemost 2000; 83: 209–11.
Cardiac biomarkers and clinical scores
3 Beer JH, Burger M, Gretener S, Bernard Bagattini S, Bounameaux H. Outpatients treatment of pulmonary embolism is feasible and safe in a substantial proportion of patients. J Thromb Haemost 2003; 1: 186–7. 4 Aujesky D, Roy PM, Verschuren F et al. Outpatient versus inpatient treatment for patients with acute pulmonary embolism: an international, open-label, randomised, non-inferiority trial. Lancet 2011; 378: 41–8. 5 Goldhaber SZ, Eliott CG. Acute pulmonary embolism, II: risk stratification, treatment, and prevention. Circulation 2003; 108: 2834–8. 6 Agterof MJ, Schutgens RE, Snijder RJ et al. Out of hospital treatment of acute pulmonary embolism in patients with a low NT-proBNP level. J Thromb Haemost 2010; 8: 1235–41. 7 Aujesky D, Smith KJ, Cornuz J, Roberts MS. Cost effectiveness of low molecular heparin for treatment of pulmonary embolism. Chest 2005; 128: 1601–10. 8 Nendaz MR, Bandelier P, Aujesky D et al. Validation of a risk score identifying patients with acute pulmonary embolism, who are at low risk of clinical adverse outcome. Thromb Haemost 2004; 91: 1232–6. 9 Coutance G, Le Page O, Lo T, Hamon M. Prognostic value of brain natriuretic peptide in acute pulmonary embolism. Crit Care 2008; 12: R109. 10 Vuilleumier N, Le Gal G, Verschuren F et al. Cardiac biomarkers for risk stratification in non-massive pulmonary embolism: a multicenter prospective study. J Thromb Haemost 2009; 7: 391–8. 11 Lankeit M, Friesen D, Aschoff J et al. Highly sensitive troponin T assay in normotensive patients with acute pulmonary embolism. Eur Heart J 2010; 31: 1836–44. 12 Lankeit M, Jim enez D, Kostrubiec M et al. Predictive value of the high-sensitivity troponin T assay and the simplified Pulmonary Embolism Severity Index in hemodynamically stable patients with acute pulmonary embolism: a prospective validation study. Circulation 2011; 124: 2716–24. 13 Loke I, Squire IB, Davies JE, Ng LL. Reference ranges for natriuretic peptides for diagnostic use are dependent on age, gender and heart rate. Eur J Heart Fail 2003; 5: 599–606. 14 Chenevier-Gobeaux C, Meune C, Freund Y et al. Influence of age and renal function on high-sensitivity cardiac troponin T diagnostic accuracy for the diagnosis of acute myocardial infarction. Am J Cardiol 2013; 111: 1701–7. 15 Zwierzina D, Limacher A, M ean M et al. Prospective comparison of clinical prognostic scores in elderly patients with pulmonary embolism. J Thromb Haemost 2012; 10: 2270–6. 16 M ean M, Righini M, Jaeger K et al. The Swiss cohort of elderly patients with venous thromboembolism (SWITCO65 + ): rationale and methodology. J Thromb Thrombolysis 2013; 36: 484–91. 17 Wicki J, Perrier A, Perneger TV, Bounameaux H, Junod AF. Predicting adverse outcome in patients with acute pulmonary embolism: a risk score. Thromb Haemost 2000; 84: 548–52. 18 Apple FS, Jaffe AS. Clinical implications of a recent adjustment to the high-sensitivity cardiac troponin T assay: user beware. Clin Chem 2012; 58: 1599–600. 19 Thygesen K, Alpert JS, Jaffe AS, Simoons ML, Chaitman BR, White HD; Joint ESC/ACCF/AHA/WHF Task Force for Universal Definition of Myocardial Infarction. Third universal
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
9
N. Vuilleumier et al.
20
21
22
23
10
definition of myocardial infarction. J Am Coll Cardiol 2012; 60: 1581–98. Januzzi JL Jr, Camargo CA, Anwaruddin S et al. The N-terminal Pro-BNP investigation of dyspnea in the emergency department (PRIDE) study. Am J Cardiol 2005; 15: 948–54. deFilippi C, Seliger SL, Kelley W et al. Interpreting cardiac troponin results from high-sensitivity assays in chronic kidney disease without acute coronary syndrome. Clin Chem 2012; 58: 1342–51. DeFilippi CR, Fink JC, Nass CM, Chen H, Christenson R. N-terminalpro-B-type natriuretic peptide for predicting coronary disease and leftventricular hypertrophy in asymptomatic CKD not requiring dialysis. Am J Kidney Dis 2005; 46: 35–44. Zoccali C, Mallamaci F, Benedetto FA et al. Cardiac natriuretic peptides are related to left ventricular mass andfunction and predict mortality in dialysis patients. J Am Soc Nephrol 2001; 12: 1508–15.
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
Cardiac biomarkers and clinical scores
24 DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: 837–45. 25 Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008; 27: 157–72. 26 Lankeit M, Jim enez D, Kostrubiec M et al. Validation of N-terminal pro-brain natriuretic peptide cut-off values for risk stratification of pulmonary embolism. Eur Respir J 2014; 43: 1669–77. Correspondence: Nicolas Vuilleumier, Department of Genetics and Laboratory Medicine, Geneva University Hospitals, 4 rue Gabrielle Perret-Gentil 1211 Geneva, Switzerland. (fax: +41223727399; e-mail: [email protected]).
