Printer Friendly

N-Terminal pro-B-type natriuretic peptide concentrations predict the risk of cardiovascular adverse events from antiinflammatory drugs: a pilot trial.

The use of selective cyclooxygenase-2 inhibitors (coxibs) [5] and traditional nonsteroidal antiinflammatory drugs (tNSAIDs) has been documented to increase the incidence of myocardial infarction and stroke (1-3). The coxib rofecoxib has been withdrawn from the market because of its cardiovascular risk (4). Several observational studies have indicated a comparable risk for tNSAIDs (5-8). The prevailing interpretation is that drugs that block the production of cardiovascularly protective eicosanoids (e.g., prostacyclin) are associated with increased blood pressure, higher rates of thromboembolic events, and accelerated atherosclerosis. This information prompted the European Medicines Agency and the US Food and Drug Administration to send out alerts indicating that the prescription of coxibs was contraindicated for patients with symptomatic heart failure or a history of cardiovascular disease. Warnings were issued to exert caution before prescribing coxibs or tNSAIDs to individuals with risk factors for heart disease, such as hypertension, hyper-lipidemia, diabetes mellitus, smoking, or peripheral arterial disease. Physicians were advised to prescribe the lowest dose for the shortest feasible treatment period. These recommendations were recently reiterated by the American Heart Association (9). Because coxibs are extremely effective in providing excellent pain relief while causing fewer gastrointestinal side effects than tNSAIDs (6), many patients and physicians prefer to continue these agents despite the potential associated cardiovascular risk The availability of a simple screening test would be helpful for identifying the patients at risk.

Brain natriuretic peptide (BNP) and its stable signal peptide, N-terminal proBNP (NT-proBNP), are released in equimolar quantities from the myocardium after proteolytic cleavage of the precursor molecule, proBNP. Both peptides can be used as markers to detect or exclude heart failure in symptomatic and asymptomatic patients (10-14). Higher NT-proBNP concentrations have been correlated with functional impairment of the heart. NT proBNP has potent prognostic significance as a marker and may permit the monitoring of heart failure therapy (15). Increased NT-proBNP concentrations also predict cardiovascular events across the entire spectrum of ischemic heart diseases (16-19). We thus hypothesized that higher NT-proBNP concentrations might indicate greater cardiovascular risk when coxibs were being used and, conversely, that lower concentrations would indicate lesser risk. If so, this strategy could be used to identify patients in whom these potentially important agents could be used more safely. Accordingly, as an initial test we evaluated this strategy with samples from a small study of coxib use that prospectively identified cardiovascular adverse events (CV-AEs).

Materials and Methods


The study cohort was originally enrolled in a clinical trial designed and conducted by Hoffmann-La Roche (internal data based on Roche Pharma Clinical Study Protocol NI 15713 F + C), which was designed to investigate the effects of a novel matrix metalloproteinase inhibitor, Ro 113-0830, on the progression of primary osteoarthritis (OA). The study was of a double-blind, randomized, 5-arm, placebo-controlled, parallel-group, multicenter design and was a dose-ranging trial. The study was de signed for a period of 24 weeks (168 days, 8 visits) followed by a final safety visit after 200 days. Adverse events were recorded during these periods.

Patients. We enrolled 433 patients in the study. All patients gave informed consent for additional scientific analyses of the collected blood samples. Inclusion criteria were primary OA of the knees with or without OA of the hands, as diagnosed according to clinical and radiographic criteria. Exclusion criteria were a history of significant active gastrointestinal disease (e.g., erosions, ulcers, bleeding), major abnormalities of a hematologic, cardiac, pulmonary, metabolic, renal, or hepatic system, and/or other comorbidities that could compromise the patient's safety or participation in the study. The primary endpoint was the scoring of pain for knee OA according to the Western Ontario and McMaster Universities (WOMAC) score. The use of analgesics, steroids, or NSAADs was permitted as rescue medication. The study protocol stipulated that any analgesic/NSAID should be taken for 1 week before switching to another analgesic/NSAID. Rescue therapy was to start with acetaminophen at up to 650 mg 4 times a day. Then, low-dose NSAADs, full dose NSAADs, full-dose NSAADs plus acetaminophen, and finally full-dose NSAADs plus acetaminophen plus narcotic drugs could be prescribed in a stepped approach. Low-dose aspirin (e.g., [less than or equal to 325 mg/day) was permitted.

Patient groups. The study population consisted of groups of patients who received rescue analgesic/ NSAID/steroid medication and those who did not. Patients who did not receive rescue medication or who took only analgesics but no NSAADs or steroids were assigned to the comparison group (n = 180). Patients on coxibs, tNSAIDs, or glucocorticoids formed the any-inhibitor group (n = 253). Patient assignment was unrelated to the original study medication.

The any-inhibitor group included patients who received 1, 2, or even 3 types of antiinflammatory drugs. Table 1 displays the composition and numbers of patients in these subsets. Patients who used coxibs alone (stratum 2) or together with other inhibitors (strata 5, 6, and 8) formed the coxibs group (n = 55). Patients who received tNSAIDs alone (stratum 3) or together with other inhibitors (strata 5, 7, and 8) formed the tNSAIDs group (n = 177). Patients who received glucocorticoids alone (stratum 4) or together with other inhibitors (strata 6-8) formed the glucocorticoids group (n = 99). Patients who received only 1 inhibitor constituted the one-inhibitor group (n = 184), whereas those who received 2 or more inhibitors constituted the multiple-inhibitor group (n = 69).

Cardiovascular endpoints. Cardiovascular events were tabulated from the investigator-documented records of adverse events, which medical professionals classified according to regulatory standards into different categories by means of the Medical Dictionary for Regulatory Activities (MedDRA). Adverse events assigned to a cardiac or vascular category formed the target composite and were termed CV-AEs. The recording of CV-AEs was part of the standard safety documentation of the study drug, and it was conducted completely independently of the use of rescue medications.

All CV-AEs were reconciled by 2 cardiologists who were not part of this trial and who were blinded to the NT-proBNP results. CV-AEs of special clinical focus included acute myocardial infarction, new diagnostic Q-waves or bundle branch block, stroke, the onset or worsening of heart failure as suggested by edema or worsening of preexisting edema of the lower extremities, rates on auscultation or pulmonary congestion documented by fluoroscopy, new onset of arterial hypertension or worsening of preexisting arterial hypertension, and confirmed venous thrombosis. Other electrocardiographic changes, unilateral edema, and isolated edema of the upper extremities did not qualify as CV-AEs. We used the entire set of CV-AEs for the statistical analysis.

NT-proBNP measurements. As part of the study de sign, 2 10-mL samples of venous blood had been collected into separate plain glass tubes, centrifuged, and stored at -70 [degrees]C at the Central Sample Office in poly styrene storage racks. NT-proBNP was measured with a highly sensitive and specific electrochemiluminescence immunoassay (Elecsys proBNP; Roche Diagnostics, Basel, Switzerland). The measurement range is 5 35 000 ng/L. The minimal detectable concentration is 5 ng/L, and the CV is 5.7% at 64 ng/L (20).


Assessment of the comparability of the rescue-medication groups. The rescue-medication groups were described and compared on the basis of available baseline data, including demographics, history, laboratory values, vital signs, NT-proBNP values, and study status.

NT-proBNP cutoff value. We used established NT-proBNP cutoff values of 125 ng/L (450 ng/L for patients older than 75 years) and 300 ng/L in the analysis for excluding patients with chronic and acute heart failure, respectively (13, 21 ). We did not use cutoff values based on an age older than 75 years or on renal function because the proportion of patients older than 75 years was only 6.93 (30 patients) and patients with an impaired renal function had already been excluded from the study. Unlike heart failure, NT-proBNP has not previously been studied in an OA population. Therefore potential cutoff values had to be determined to stratify cardiovascular risks. We used Cox regression analysis to plot hazard ratios (high vs low NT-proBNP concentration) against all possible cutoff values for each of the rescue-medication groups. We compared the cutoff values with the lowest P values and confirmed the optimal cutoff value with ROC curve analysis (see Fig. 1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem. org/content/vol54/issue7). Because this cutoff value (100 ng/L) was close to 125 ng/L, we compared sensitivities and negative predictive values in a secondary analysis.

Core analyses, interactions, and covariate adjustments. We used survival analysis to evaluate CV-AE risk, because the censoring and event times could be variable. CV-AE rates over time are displayed as Kaplan-Meier graphs. We used Cox regression (i.e., the Cox proportional hazards model) to investigate the time to first CV-AE to assess the predictive value of baseline NT proBNP values and their interaction with the rescue medication groups.

We used different regression models (M1-M3) with the factors detailed in Table 1. All involved the comparison group plus M1 (any-inhibitor factor), M2 (the 3 overlapping factors coxibs, tNSAIDs, and glucocorticoids), and M3 (the one-inhibitor and multiple-inhibitor factors). All models contained interaction terms (interaction of the respective medication factor with NT-proBNP status) that were formally tested. By the inclusion of interaction terms, the regression models also produced hazard ratios vs the comparison group that were based on NT-proBNP status, as well as hazard ratios for comparisons of NT-proBNP groups for a cutoff value of [greater than or equal to] 100 ng/L based on the specific medication factors).

To evaluate the independent predictive value of NT-proBNP relative to covariates and potential confounders, regression analysis with the factors described above without additional covariates (set A) was supplemented with analyses with adjusted regression models that used 4 additional different sets of covariates: the effect of the randomized study drug Ro 113-0830 at incremental doses (set B); relevant clinical covariates, including age, systolic blood pressure, diabetes mellitus (set C); previous and concomitant use of antithrombotic agents, antiplatelet drugs, antihypertensive drugs, and statins/ fibrates (set D); and a panel of baseline laboratory results for variables (leukocytes, platelets, cholesterol, sodium, potassium, and serum glutamic-pyruvic transaminase) with potential relevance to patients with cardiovascular disease (set E).

Metric covariates were dichotomized with the best cutoff value for predicting CV-AEs (found by Cox regression, similar to that of the search for the NT-proBNP cut-off value). The choice of this data-adapted dichotomization allowed the covariates to present an optimized competitor to challenge the predictive power of NT proBNP concentration. A test result was considered statistically significant when the P value was <0.05. Because the study was hypothesis-generating, we deemed that no adjustment for multiple testing was necessary.

The study was not powered to detect differences in CV-AE rates (no control for (3 error). Therefore, the lack of statistical significance does not disprove the presence of significant effects or interactions. Results


Table 2 summarizes the baseline characteristics of the 433 study patients. The median NT-proBNP value at baseline for the entire study cohort was 64 ng/L (quartile 1, 31 ng/L; quartile 3, 125 ng/L). The distributions of baseline NT-proBNP values for the comparison group and the rescue-medication groups did not differ markedly. The proportions of patients who received rescue medication consisting of coxibs, tNSAIDs, or glucocorticoids were similar across all incremental doses of the study drug Ro 113-0830 and placebo controls (data not shown). We observed 82 mild to serious CV-AEs during an observational period of 200 days. CV-AE incidence was not related to the dosage levels of the study drug (data not shown).

NT-proBNP and CV-AEs. Of the 433 study patients, 144 patients had increased NT-proBNP values (above the 100-ng/L cutoff value). Sixty-two of these patients were in the comparison group, and there were 7 CV-AEs in this group (11.3%). Twenty-two (26.8%) of the 82 patients with NT-proBNP values > 100 ng/L in the any-inhibitor group experienced a CV-AE overall (see Table 1 in the online Data Supplement), with 19 of these CV-AEs occurring within the first 200 days.

Table 3 also presents hazard ratios obtained from the Cox regression models for the occurrence of CV-AEs in the rescue-medication groups vs the comparison group and in the patients with NT-proBNP concentrations [greater than or equal to] 100 ng/L vs patients with NT-proBNP concentrations < 100 ng/L. The risk associated with the use of any anti-inflammatory drug (any-inhibitor group) was 2.4-fold higher in patients with NT-proBNP values [greater than or equal to] 100 ng/L than in the comparison group (difference not statistically significant). The coxibs group, however, had a 3.65-fold higher risk for CV-AEs than the comparison group (P < 0.01). The tNSAID group and the glucocorticoids group had a 1.38-fold higher risk (not significant) and a 2.36-fold higher risk (P < 0.05), respectively.

The risk (i.e., hazard ratios) for CV-AEs in patients with NT-proBNP values [greater than or equal to] 100 ng/L was 1.95-fold higher in the any-inhibitor group (P < 0.05), 7.41-fold higher in the coxibs group (P < 0.01), 1.89-fold higher in the tNSAID group (not significant), and 2.35-fold higher in the glucocorticoid group (not significant), compared with the patients in the corresponding groups with NT-proBNP values < 100 ng/L. The corresponding risk was 3.74-fold higher (P < 0.05) in the multiple-inhibitor group and 1.37-fold higher in the one-inhibitor group (not significant).

Conversely, very few events occurred in patients if their NT-proBNP values were < 100 ng/L, regardless of whether they were treated with antiinflammatory medication. The sensitivities, specificities, positive predictive values, and high negative predictive values for NT proBNP concentrations < 100 ng/L across all medication groups presented in Table 4 emphasize the usefulness of the NT-proBNP concentration for risk stratification.

The effects of the interactions between the comparison group and all rescue-medication groups and NT-proBNP status ([greater than or equal to] 100 ng/L or < 100 ng/L) were significant (overall P values for interaction in models M2 and M3 were 0.006 and 0.008, respectively). The estimated interaction effect in model M2 was strongest for the coxibs group despite the fact that this group was the smallest. All regression results were controlled for the effects of incremental doses of the study drug Ro 113-0830, concomitant cardiovascular and anti thrombotic medications, age, presence of hypertension, and history of diabetes. After adjustment for 4 different sets of potentially influential covariates, the effects on the CV-AE risk remained stable and did not indicate that confounding influences were responsible for the observed effects. The independent predictive value of NT-proBNP was confirmed in all runs (Fig. l ). Rates of CV-AEs over time (Kaplan-Meier graphs) for the rescue-medication groups were plotted separately for NT-proBNP concentrations [greater than or equal to] 100 ng/L and < 100 ng/L (Fig. 2). CV-AE rates were greater for patients with NT-proBNP concentrations [greater than or equal to] 100 ng/L and occurred earlier in the patients in the rescue-medication groups than in the comparison group.

An NT-proBNP cutoff value of 100 ng/L vs 125 ng/L. A clinically more conservative cutoff value of 100 ng/L was slightly superior to 125 ng/L for stratifying CV-AE risk in these patients. Compared with the 125-ng/L cut off value, the 100-ng/L cutoff exhibited higher sensitivities (75.0% vs 58.3%) and higher negative predictive values (90.9% vs 86.1%) in the coxib group. Given its higher negative predictive value, we considered the 100-ng/L cutoff to be better suited for identifying those individuals who might be treated safely with coxibs.


The results of our pilot study suggest that the NT-proBNP concentration may be useful in determining the extent of the cardiovascular risk in OA patients who are treated with antiinflammatory agents. NT-proBNP values [greater than or equal to] 100 ng/L at baseline identified patients at increased cardiovascular risk when treated with these agents. Specifically, patients who had received coxibs (alone or in combination with tNSAIDs or glucocorticoids) demonstrated the highest risk increase for CV-AEs (7.41 fold higher) when the NT-proBNP concentration was [greater than or equal to] 100 ng/L. Patients treated with 2 or more antiinflammatory drugs (i.e., the multiple-inhibitor group) were exposed to considerable risk (3.6-fold higher) when the NT-proBNP concentration was increased.


Conversely, CV-AEs rates did not appear to be increased by the use of antiinflammatory drugs in patients with NT-proBNP values < 100 ng/L. Even the administration of several types of these drugs (together or consecutively) in a patient with NT-proBNP values < 100 ng/L was not associated with an increased incidence of CV-AEs, as the negative predictive values of >85% convincingly indicate. Our study lacks the power to say that these results definitively exclude an effect, especially in "harder" endpoints such as mortality and myocardial infarction, but it does provide a template for evaluating NT-proBNP in larger data sets. If these results are confirmed, a strategy that uses NT proBNP concentration as a marker could improve the safety of treatment with antiinflammatory agents.

The use of drugs that interfere with the production of cyclooxygenase-2-derived eicosanoids (i.e., coxibs, tNSAIDs, and glucocorticoids) appears to affect the production of prostacyclin (22-25) and thus shifts the balance between cardioprotective and prothrombotic prostanoids (26, 27). Our finding that increased CV-AEs occurred only in patients with NT-proBNP values [greater than or equal to] 100 ng/L is most likely related to this balance. It is well known that the concentrations of these regulators of vascular and kidney function (28) are often abnormal in the presence of cardiovascular disease and that evaluating NT-proBNP is an excellent way to unmask subclinical cardiovascular impairment and thus predict cardiovascular events (15-19). Coxibs and tNSAIDs both cause water and salt retention (29, 30), increase ventricular wall tension, and impair cardiac function. Individuals with increased NT-proBNP values constitute a group much more likely to have sub-clinical cardiovascular impairment and therefore more likely to anticipate specific or nonspecific CV-AEs than those without increased NT-proBNP concentrations. The CV-AEs observed in the present study included the development of edema, worsening of arterial hypertension, heart failure, angina pectoris, and a number of less severe CV-AEs. We did not have a large number of highly morbid events. The small number of major CV AEs is probably due to exclusion of patients with overt cardiac or vascular diseases, the liberal use of aspirin, and the relatively short observation period of only 200 days. In the APPROVe trial, the excess of CV-AEs in patients treated with rofecoxib did not occur before 18 months (4 ).


The need for an effective risk-stratification paradigm is highlighted by a recent nested case-control study of a cohort of 486 378 patients. Even patients who had no history or symptoms of coronary artery disease, hypertension, or diabetes mellitus and were receiving coxibs or tNSAIDs had a higher risk of acute myocardial infarction (31 ). Thus, screening patients for traditional risk factors may not be an effective strategy (32 ). Monitoring the NT proBNP concentration, which can reveal both overt and subclinical cardiovascular impairment, may be more suitable (33, 34 ).


Antiinflammatory drugs were not allocated randomly in this study; however, the fact that drug administration was driven by bone pain and not by cardiovascular status allowed a study of the relationship of drug therapy to CV-AEs. Despite this limitation, it is notable that the negative predictive values of NT-proBNP concentrations were still robust.

The observational structure of our analysis is an other limitation. We attempted to control aggressively for possible confounders, and the effects remained stable, even with these adjustments. At the time of this study, the use of antiinflammatory drugs appeared to be safe with respect to CV-AEs; hence, these drugs may have been used with less caution than they would be used to day. Therefore, the total number of CV-AEs, particularly for nonserious AEs (e.g., silent ischemia, worsening of heart failure, or hypertension) as well as other subclinical events, such as deep venous thrombosis or pulmonary embolism, might have been underestimated.

Finally, we used an NT-proBNP cutoff value derived from our OA study population, because prior data in this area were lacking. The cutoff value chosen (100 ng/L) is close to the 125-ng/L cutoff value established for the exclusion of heart failure. Given the high degree of biological variation in NT-proBNP measurements, we acknowledge that this value should be considered not as an absolute but rather as a guide to clinicians (34). Finally, we acknowledge that our investigation was a hypothesis-generating study and that there is a need to examine additional, larger data sets so that harder endpoint events can be evaluated more definitively.

Grant/Funding Support: This study was funded by a grant from Roche Diagnostics, Germany.

Financial Disclosures: K. Brunes is Doerenkamp Professor for Innovations in Animal and Consumer Protection. He has been and is a consultant to MSD, Novartis, and Pfizer. He received research support from the same companies. E. Giannitsis has received financial support for clinical trials from Roche Diagnostics and MSD Ger many. He is a consultant to Bristol-Myers and receives honoraria for lectures from Takeda, MSD, Roche, Lilly, Novartis, BMS, Astra, and Sanofi-Aventis. H. Katus has developed the cardiac-specific troponin T assay and holds a patent jointly with Roche Diagnostics. He has received grants and research support from several companies and has received honoraria for lectures from Roche Diagnostics, MSD, Roche, Lilly, Novartis, BMS, Astra, and Sanofi-Aventis. J. Moecks is a former employee of Roche Diagnostics and presently is a statistical consultant for Roche Pharmaceuticals. E. Spanuth is a former employee of Roche Diagnostics. A. Jaffe is a consultant to and receives research support from Dade Behring, Beckman Cooper, and Ortho Diagnostics. He is or has been a consultant to most of the major diagnostic companies, including Roche.

Acknowledgments: We are indebted to Roche Diagnostics for making the results of the study (Roche Pharma Clinical Study Protocol NI 15713 F + C) available to us and for assessing NT-proBNP in the frozen serum samples.


(1.) Hippisley-Cox J, Coupland C. Risk of myocardial infarction in patients taking cyclo-oxygenase-2 inhibitors or conventional non-steroidal anti-in-flammatory drugs: population based nested case control analysis. BMJ 2005;330:1366.

(2.) Solomon SD, McMurray JJ, Pfeffer MA, Wittes J, Fowler R, Finn P, et al., for the Adenoma Prevention with Celecoxib (APC) Study Investigators. Cardiovascular risk associated with celecoxib in a clinical trial for colorectal adenoma prevention. N Engl J Med 2005;352:1071-80.

(3.) American College of Rheumatology. Hotline Web page. 0305NSAIDs.asp (Accessed April 22, 2008).

(4.) Bresalier RS, Sandier RS, Quan H, Bolognese JA, Oxenius B, Horgan K, et al., for the Adenomatous Polyp Prevention on Vioxx (APPROVe) Trial Investigators. Cardiovascular events associated with rofecoxib in a colorectal adenoma chemoprevention trial. N Engl J Med 2005;352:1092-102.

(5.) Nussmeier NA, Whelton AA, Brown MT, Langford RM, Hoeft A, Parlow JL, et al. Complications of the COX-2 inhibitors parecoxib and valdecoxib after cardiac surgery. N Engl J Med 2005;352: 1081-91.

(6.) Bombardier C, Laine L, Reicin A, Shapiro D, Burgos-Vargas R, Davis B, et al., for the VIGOR Study Group. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. N Engl J Med 2000;343: 1520-8.

(7.) Singh G, Mithal A, Triadafilopoulos G. Both in hibitors and non-selective NSAIDs increase the risk of acute myocardial infarction in patients with arthritis: selectivity is with the patient, not with the drug class [abstract]. In: European Congress of Rheumatology annual meeting program; 2005 June 8-11; Vienna, Austria. Abstract OP0091.

(8.) Schnitzer TJ, Burmester GR, Mysler E, Hochberg MC, Doherty M, Ehrsam E, et al., on behalf of the TARGET Study Group. Comparison of lumiracoxib with naproxen and ibuprofen in the Therapeutic Arthritis Research and Gastrointestinal Event Trial (TARGET), reduction in ulcer complications: randomised controlled trial. Lancet 2004;364:665-74.

(9.) Antman EM, Bennett JS, Daugherty A, Furberg C, Roberts H, Taubert KA. Use of nonsteroidal anti inflammatory drugs: an update for clinicians: a scientific statement from the American Heart Association. Circulation 2007;115:1634-42.

(10.) Hunt PJ, Richards AM, Nicholls MG, Yandle TG, Doughty RN, Espiner EA. Immunoreactive amino-terminal pro-brain natriuretic peptide (NT-proBNP): a new marker of cardiac impairment. Clin Endocrinol 1997;47:287-96.

(11.) McDonagh TA, Holmer S, Raymond I, Luchner A, Hildebrandt P, Dargie HJ. NT-proBNP and the diagnosis of heart failure: a pooled analysis of three European epidemiological studies. Eur J Heart Fail 2004;6:269-73.

(12.) Seino Y, Ogawa A, Yamashita T, Fuhushima M, Ogata K, Fukumoto H, et al. Application of NT-proBNP and BNP measurements in cardiac care: a more discerning marker for the detection and evaluation of heart failure. Eur J Heart Fail 2004; 6:295-300.

(13.) Svendstrup-Nielsen LS, Svanegaard J, Klitgaard NA, Egeblad H. N-terminal pro-brain natriuretic peptide for discriminating between cardiac and non-cardiac dyspnoea. Eur J Heart Fail 2004;6:63-70.

(14.) Mueller T, Gegenhuber A, Poelz W, Haltmayer M. Head-to-head comparison of the diagnostic utility of BNP and NT-proBNP in symptomatic and asymptomatic structural heart disease. Clin Chim Acta 2004;341:41-8.

(15.) Hartmann F, Packer M, Coats AJ, Fowler MB, Krum H, Mohacsi P, et al. Prognostic impact of plasma N-terminal pro-brain natriuretic peptide in severe chronic heart failure: a substudy of the Carvedilol Prospective Randomised Cumulative Survival (COPERNICUS) trial. Circulation 2004; 110:1780-6.

(16.) Galvani M, Ottani F, Oltrona L, Ardissino D, Gensini GF, Maggioni AP, et al. N-terminal pro-brain natriuretic peptide on admission has prognostic value across the whole spectrum of acute coronary syndromes. Circulation 2004;110:128-34.

(17.) Wang TJ, Larson MG, Levy D, Benjamin EJ, Leip EP, Omland T, et al. Plasma natriuretic peptide levels and the risk of cardiovascular events and death. N Engl J Med 2004;350:655-63.

(18.) Kragelund C, Gronning B, Kober L, Hildebrandt P, Steffensen R. N-terminal pro-B-type natriuretic peptide and long-term mortality in stable coronary heart disease. N Engl J Med 2005;352: 666-75.

(19.) De Suffer J, De Bacquer D, Cuypers S, Delanghe J, De Buyzere M, Kornitzer M, De Backer M. Plasma N-terminal pro-brain natriuretic peptide concentration predicts coronary events in men at work: a report from the BELSTRESS study. Eur Heart J 2005;26:2644-9.

(20.) Barnes SC, Collinson PO, Galasko G, Senior R. Evaluation of N-terminal pro-B type natriuretic peptide analysis an the Elecsys 1010 and 2010 analyzers. Ann Clin Biochem 2004;41: 459-63.

(21.) Januzzi JL Jr, Camargo CA, Anwaruddin S, Baggish AL, Chen AA, Krauser DG, et al. The N-terminal Pro-BNP Investigation of Dyspnea in the Emergency Department (PRIDE) study. Am J Cardiol 2005;95:948-54.

(22.) Rabausch K, Bretschneider E, Sarbia M, Meyer Kirchrath J, Censarek P, Pape R, et al. Regulation of thrombomodulin expression in human vascular smooth muscle cells by COX-2-derived prosta-glandins. Circ Res 2005;96:e1-6.

(23.) Cheng Y, Austin SC, Rocca B, Koller BH, Coffman TM, Grosser T, et al. Role of prostacyclin in the cardiovascular response to thromboxane A2. Science 2002;296:539-41.

(24.) Kearney PM, Baigent C, Godwin J, Halls H, Emberson JR, Patrono C. Do selective cyclo-oxygenase-2 inhibitors and traditional non-steroidal anti-inflammatory drugs increase the risk of atherothrombosis? Meta-analysis of randomised trials. BMJ 2006;332:1302-8.

(25.) Cipollone F, Fazia M, Mezzetti A. Novel determinants of plaque instability. J Thromb Haemost 2005;3:1962-75.

(26.) McAdam BF, Catella-Lawson F, Mardini IA, Kapoor S, Lawson JA, FitzGerald GA. Systemic biosynthesis of prostacyclin by cyclooxygenase (COX)-2: the human pharmacology of a selective inhibitor of COX-2. Proc Natl Acad Sci U S A 1999;96:272-7.

(27.) Baigent C, Patrono C. Selective cyclooxygenase 2 inhibitors, aspirin, and cardiovascular disease: a reappraisal. Arthritis Rheum 2003;48:12-20.

(28.) Harris RC, Breyer MD. Physiological regulation of cyclooxygenase-2 in the kidney. Am J Physiol Renal Physiol 2001;281:F1-11.

(29.) Brater DC. Anti-inflammatory agents and renal function. Semin Arthritis Rheum 2002;32:33-42.

(30.) Schwartz JI, Vandormael K, Malice MP, Kalyani RN, Lasseter KC, Holmes GB, et al. Comparison of rofecoxib, celecoxib, and naproxen on renal function in elderly subjects receiving a normal-salt diet. Clin Pharmacol Ther 2002;72:50-61.

(31.) Anderson F, Suissa S, Garbe E. Use of first- and second-generation cyclooxygenase-2-selective nonsteroidal anti-inflammatory drugs and risk of acute myocardial infarction. Circulation 2006; 113:1950-7.

(32.) Giannitsis E. Rationale for testing the cardiovascular risk for patients with COX-2 inhibitors on the basis of biomarker NT-proBNP. Clin Lab 2005; 51:63-72.

(33.) Blankfield RP. Can natriuretic peptide levels predict the cardiovascular complications of COX-2 inhibitors and nonsteroidal anti-inflammatory drugs? J Am Board Fam Med 2006;19:178-82.

(34.) Wu AH. Serial testing of B-type natriuretic peptide and NTproBNP for monitoring therapy of heart failure: the role of biologic variation in the interpretation of results. Am Heart J 2006;152: 828-34.

Kay Brune, [1] Hugo A. Katus, [2] Joachim Moecks, [3] Eberhard Spanuth, [3] Allan 5. Jaffe, [4] * and Evangelos Giannitsis [2]

[1] Department of Experimental and Clinical Pharmacology and Toxicology, Friedrich-Alexander University of Erlangen-Nuremberg, Erlangen, Germany; [2] Department of Internal Medicine, University Hospital of Heidelberg, Heidelberg, Germany; [3] DIAneering GmbH, Diagnostics Engineering, Research and Know-How Services, Heidelberg, Germany; [4] Cardiovascular Division, Department of Internal Medicine and Department of Laboratory Medicine and Pathology, Mayo Clinic and Mayo Medical School, Rochester, MN.

[5] Nonstandard abbreviations: coxib, cyclooxygenase-2 inhibitor; tNSAID, traditional nonsteroidal antiinflammatory drug; BNP, B-type natriuretic peptide; NT-proBNP, N-terminal proBNP; N-AE, cardiovascular adverse event; OA, osteoarthritis.

* Address correspondence to this author at: Mayo Clinic, Gonda Bldg., 5th Floor, 200 First St., Rochester, MN 55905. E-mail

Previously published online at D01: 10.1373/clinchem.2007.097428

Received September 5, 2007; accepted April 4, 2008.
Table 1. Composition of the patient groups and factors
of regression models. (a)

 Regression models

 M1 factor M2 factors

 Any-inhibitor Coxlbs tNSAIDs Glucocorticoids
 Strata n group group group group

 1 180 0 0 0 0
 2 25 1 1 0 0
 3 118 1 0 1 0
 4 41 1 0 0 1
 5 11 1 1 1 0
 6 10 1 1 0 1
 7 39 1 0 1 1
 8 9 1 1 1 1
[SIGMA], n 433 253 55 177 99

 Regression models

 M3 factors

 One-inhibitor Multiple-inhibitors
 Strata group group

 1 0 0
 2 1 0
 3 1 0
 4 1 0
 5 0 1
 6 0 1
 7 0 1
 8 0 1
[SIGMA], n 184 69

(a) Stratum 1, comparison group; M1: strata 2-8 (any inhibitor);
M2: strata 2, 5, 6, 8 (coxibs), strata 3, 5, 7, 8 (tNSAIDs), strata
4, 6, 7, 8 (glucocorticoids); M3: strata 2-4 (one inhibitor), 5-8
(multiple inhibitors).

Table 2. Baseline characteristics of the patient groups. (a)


 Comparison Any-inhibitor Coxibs
 group group group
 (n = 180) (n = 253) (b) (n = 55)

Age, years 62 (9) 60 (8) 62 (9)
Male, % 32 24 22
Caucasian, % 88 91 96
BMI (c) 30 (5) 30 (5) 31 (5)
Previous CVD, % 29 24 22
Diabetes mellitus, % 9 5 5
Hypertension, % 35 28 31
Stroke, % 3 1 0
Systolic BP, mmHg 133 (17) 132 (16) 130 (14)
Diastolic BP, mmHg 80 (9) 79 (8) 79 (9)
Heart rate, /min 72 (9) 73 (9) 74 (8)
Serum creatinine,
 [micro]mol/L 68 (15) 68 (15) 71 (15)
Hemoglobin, g/L 150 (10) 150 (10) 150 (10)
 Patients <100 ng/L, % 66 68 60
 Median (Q1, Q3)
 concentration, ng/L 65 (30, 38) 73 (38, 125) 83 (38, 187)

 Any-inhibitor subgroups

 tNSAIDs Glutotortitoids
 group group
 (n = 177) (n = 99)

Age, years 59 (8) 60 (8)
Male, % 24 20
Caucasian, % 89 92
BMI (c) 30 (5) 30 (5)
Previous CVD, % 23 32
Diabetes mellitus, % 4 4
Hypertension, % 27 27
Stroke, % 1 2
Systolic BP, mmHg 131 (16) 31 (16)
Diastolic BP, mmHg 80 (8) 79 (8)
Heart rate, /min 73 (9) 73 (9)
Serum creatinine,
 [micro]mol/L 66 (14) 69 (17)
Hemoglobin, g/L 150 (10) 150 (10)
 Patients <100 ng/L, % 72 62
 Median (Q1, Q3)
 concentration, ng/L 58 (30, 102) 73 (35, 151)

(a) Data are presented as the mean (SD) unless otherwise indicated.

(b) A total number less than the sum of any-inhibitor subgroups
(coxibs, tNSAIDs, and glucocorticoids group) is due to the use
of multiple inhibitors in some patients. See Table 1 for details.
There are no significant differences between groups with respell
to the above baseline characteristics.

(c) BMI, body mass index; CVD, cardiovascular disease; BP, blood
pressure; Q1, first quartile; Q3, third quartile.

Table 3. Cox proportional hazards regression for time
to first CV-AE during 200-day follow-up. (a)

 NT-proBNP <100 ng/L

 HR vs
 Events (b) groups (c)

Comparison group 20/118 (17%) 1.0
Any-inhibitor group 22/171 (13%) 0.72 (0.40-1.33)
Any-inhibitor subgroups
 Coxibs group 3/33 (9%) 0.53 (0.16-1.72)
 (M2; n = 55)
 tNSAIDs group 15/128 (12%) 0.67 (0.35-1.25)
 (M2; n = 177)
 Glucocorticoids group 9/61 (15%) 1.09 (0.52-2.30)
 (M2; n = 99)
 One-inhibitor group 17/126 (13%) 0.77 (0.40-1.47)
 (M3; n = 174)
 Multiple-inhibitors 5/45 (11%) 0.60 (0.23-1.60)
 (M3; n = 69)

 NT-proBNP [greater than or equal to] 100 ng/L

 HR vs
 Events (b) groups

Comparison group 6/62 (10%) 1.0
Any-inhibitor group 19/82 (23%) 2.40 (0.96-6.00)
Any-inhibitor subgroups
 Coxibs group 9/22 (41 %) 3.65 ** (1.58-8.46)
 (M2; n = 55)
 tNSAIDs group 10/49 (20%) 1.38 (0.60-3.17)
 (M2; n = 177)
 Glucocorticoids group 12/38 (32%) 2.36 * (1.06-5.22)
 (M2; n = 99)
 One-inhibitor group 10/58 (17%) 1.80 (0.65-4.94)
 (M3; n = 174)
 Multiple-inhibitors 9/24 (38%) 3.81 * (1.36-10.7)
 (M3; n = 69)


 HR, [greater than or
 Interaction equal to] 100 ng/L
 HR (c) <100 ng/L (c)

Comparison group 1.0 0.59 (0.24-1.46)
Any-inhibitor group 3.31 * (1.10-9.95) 1.95 * (1.05-3.60)
Any-inhibitor subgroups
 Coxibs group 6.92 ** (1.62-29.6) 7.41 ** (1.90-28.9)
 (M2; n = 55)
 tNSAIDs group 2.07 (0.73-5.90) 1.89 (0.84-4.25)
 (M2; n = 177)
 Glucocorticoids group 2.16 (0.73-6.42) 2.35 (0.98-5.63)
 (M2; n = 99)
 One-inhibitor group 2.33 (0.70-7.75) 1.37 (0.63-2.99)
 (M3; n = 174)
 Multiple-inhibitors 6.35 * (1.53-26.4) 3.74 * (1.25-11.1)
 (M3; n = 69)

(a) Displayed are results for 3 different analysis models (M1-M3;
see Table 1), including pairwise interaction terms of medication
groups with NT-proBNP status. Conditional hazard ratios (HRs) are
as estimated from interaction models. M1 model includes one factor
for medication with any inhibitor. M2 model includes 3 factors of
medications with coxibs, tNSAIDs, and glucocorticoids. M3 model
includes one factor for medication with one inhibitor and
medication with multiple inhibitors. * P < 0.05; ** P < 0.01.

(b) Number of events/number of patients (percent).

(c) HRs are expressed as HR (95% confidence interval).

Table 4. Sensitivities, specificities, and negative and positive
predictive values for predicting any CV-AE with an NT-proBNP cutoff
value of 100 ng/L. (a)

 All patients Coxibs group

Patients at 433 55
 baseline, n
Patients with 67 60
 <100 ng/L,
Sensitivity, % 37.3 (25.8; 50.0) 75.0 (42.8; 94.5)
Specificity, % 67.5 (62.4; 72.3) 69.8 (53.9; 82.8)
ROC AUC 0.58 * (0.51; 0.66) 0.72 * (0.56; 0.89)
PPV, % 17.4 (11.6; 24.6) 40.9 (20.7; 63.6)
NPV, % 85.5 (80.9; 89.3) 90.9 (75.7; 98.1)

 tNSAIDs Glutotortitoids
 group group

Patients at 177 99
 baseline, n
Patients with 72 62
 <100 ng/L,
Sensitivity, % 40.0 (21.1; 61.3) 57.1 (34.0; 78.2)
Specificity, % 74.3 (66.6; 81.1) 66.7 (55.1; 76.9)
ROC AUC 0.66 * (0.55; 0.77) 0.71 * (0.58; 0.84)
PPV, % 20.4 (10.2; 34.3) 31.6 (17.5; 48.7)
NPV, % 88.3 (81.4; 93.3) 85.2 (73.8; 93.0)

 Any-inhibitor Multiple-inhibitors
 group group

Patients at 253 69
 baseline, n
Patients with 68 65
 <100 ng/L,
Sensitivity, % 46.3 (30.7; 62.6) 64.3 (35.1; 87.2)
Specificity, % 70.3 (63.6; 76.3) 72.7 (59.0; 83.9)
ROC AUC 0.66 * (0.57; 0.75) 0.76 ** (0.61; 0.92)
PPV, % 23.2 (14.6; 33.8) 37.5 (18.8; 59.4)
NPV, % 87.1 (81.2; 91.8) 88.9 (75.9; 96.3)


Patients at 180
 baseline, n
Patients with 66
 <100 ng/L,
Sensitivity, % 23.1 (9.0; 43.6)
Specificity, % 63.6 (55.5; 71.2)
ROC AUC 0.47 (0.34; 0.60)
PPV, % 9.7 (3.6; 19.9)
NPV, % 83.1 (75.0; 89.3)

(a) Sensitivity, specificity, ROC curve, positive predictive value
(PPV), and negative predictive value (NPV) data are expressed with
the 95% confidence limits in parentheses. * P < 0.05, and
** P < 0.01, for differences in ROC areas under the curve (AUCs)
with respell to the comparison group.
COPYRIGHT 2008 American Association for Clinical Chemistry, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2008 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Proteomics and Protein Markers
Author:Brune, Kay; Katus, Hugo A.; Moecks, Joachim; Spanuth, Eberhard; Jaffe, Allan S.; Giannitsis, Evangel
Publication:Clinical Chemistry
Article Type:Clinical report
Date:Jul 1, 2008
Previous Article:Utility of oligonucleotide array--based comparative genomic hybridization for detection of target gene deletions.
Next Article:Measuring carbohydrate-deficient transferrin by direct immunoassay: factors affecting diagnostic sensitivity for excessive alcohol intake.

Related Articles
National academy of clinical biochemistry laboratory medicine practice guidelines: use of cardiac troponin and B-type natriuretic peptide or...
N-terminal pro-B-type natriuretic peptide (NT-proBNP) concentrations in hemodialysis patients: prognostic value of baseline and follow-up...
N-terminal pro-B-type natriuretic peptide predicts total and cardiovascular mortality in individuals with or without stable coronary artery disease:...
B-type natriuretic peptide concentrations predict the progression of nondiabetic chronic kidney disease: The mild-to-moderate kidney disease study.
Role of N-terminal pro-B-type natriuretic peptide in risk stratification in patients presenting in the emergency room.
Rapid brain natriuretic peptide test and Doppler echocardiography for early diagnosis of mild heart failure.
Diagnostic accuracy and prognostic relevance of the measurement of cardiac natriuretic peptides: a review.
Analytical and clinical evaluation of the Bayer ADVIA centaur automated B-Type natriuretic peptide assay in patients with heart failure: a multisite...
Strategies for developing biomarkers of heart failure.
Utility of N-terminal pro-B-type natriuretic peptide to differentiate cardiac diseases from noncardiac diseases in young pediatric patients.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters