Steady-State Levels of Troponin and Brain Natriuretic Peptide for Prediction of Long-Term Outcome after Acute Heart Failure with or without Stage 3 to 4 Chronic Kidney Disease

Aim: To determine whether assessment of a combination of steady-state discharge levels of biomarkers improves risk stratification after acute decompensate HF. Study Design: Retrospective cohort study. Place and Duration of Study: Keio University Hospital, between January 2006 and September 2011. We analyzed 244 patients with acute HF due to ischemic or dilated cardiomyopathy who Research Article British Journal of Medicine & Medical Research, 2(4): 490-500, 2012 491 were enrolled in a prospective, single institution-based registry between January 2006 and September 2011. Patients were stratified by discharge values of BNP and/or TnT. The primary endpoint was a composite of HF readmission or death during the 2-year period after discharge. Results: The population was predominantly male (69.3%), and the mean age was 66.6±15.3 years. Patients with higher BNP levels or detectable TnT had a worse prognosis (BNP45.0% vs. 18.8%, p<0.001; TnT 43.8% vs. 25.1%, p=0.002, respectively). The primary event rate was additively worse among patients with both increased BNP levels and detectable TnT compared to those with increased levels of BNP or detectable TnT alone (log-rank p<0.001). A similar trend was observed in the subgroup of patients with CKD stage III–V (n=172). Conclusion: Assessment of both BNP and TnT values may have a significant predictive value for HF prognosis, even among patients with CKD, a condition affecting biomarker levels.


INTRODUCTION
Heart failure (HF) is a disorder that has major clinical and public health impacts worldwide (Young, 2004). In Japan, the number of HF admissions reached 582,000 in 2008 and HF consumes a significant proportion of the funds allocated for cardiovascular care, which totaled $713 billion in 2009 (Report from Japanese Health and Welfare, 2009). HF is estimated to affect over 5 million people in the US, and it is the most common Medicare diagnosis-related group. Furthermore, as the population ages, HF prevalence increases; currently, the HF incidence approaches 10 per 1000 among those older than 65 years, and the estimated cost of HF treatment in the United States was $37 billion in 2009 (Lloyd-Jones et al., 2009).
Several major advances in the management of HF have been achieved in the past decade, and some data suggest that these new advances are beginning to impact the prognosis of HF in the community (Polanczyk et al., 2000). One such advance is the reorganization of high-risk patients stratified by biomarkers. In particular, biomarkers representing a patient's fluid status (e.g., brain natriuretic peptide [BNP]) or degree of myocardial injury (e.g., highly sensitive troponins [TnT]) are valuable tools for predicting the long-term prognosis of HF (Peacock et al., 2008;Braunwald, 2008;Brugger-Andersen et al., 2008;Felker et al., 2000).
However, the impact of the combination of BNP and TnT values has not been thoroughly investigated, particularly among patients with chronic kidney disease (CKD). Kidney disease affects HF through various mechanisms, including those described by the low-flow-state hypothesis, intraabdominal and central venous pressure elevation, sympathetic overactivity, rennin-angiotensin-aldosterone system overactivity, and oxidative injury. HF affects biomarker levels, but its impact on the long-term outcome is unclear. Data on biomarker levels are also particularly sparse in the Japanese population where a non-ischemic etiology is the dominant characteristic of HF. Therefore, we sought to investigate whether evaluation of a combination of biomarkers could improve risk stratification of patients with acute decompensated HF, particularly those with CKD.

Study Subjects
Between January 2006 and September 2011, we prospectively registered 244 consecutive patients who were admitted to Keio University Hospital for the treatment of decompensate HF defined according to the Framingham criteria. To avoid the influence of acute coronary events and renal dysfunction, patients with clinical or electrocardiographic evidence of acute coronary syndrome in the previous 3 months, those with coronary revascularization in the previous 3 months, those with renal failure (serum creatinine concentration, 2.5 mg/dl), and those undergoing hemodialysis were excluded (Srisawasdi et al., 2010). Patients with terminal cancer, infections and inflammatory diseases were also excluded. The primary endpoint was a composite of all-cause mortality plus HF requiring hospitalization. The mean length of follow-up was 730 days (interquartile range 397 to 730). Informed consent was obtained from each patient upon enrollment. Dilated cardiomyopathy is characterized by dilatation and impaired contraction of 1 or both ventricles and an EF of <45% without the presence of obstructive coronary disease. Ischemic cardiomyopathy was defined by a-number-of-diseased-vessels classification which is the presence of obstruction (more than 75%) in one or more coronary (Felker et al., 2002). The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki, and the study was approved by the institution's human research committee.

Measurement of Cardiac Biomarkers
Plasma TnT and BNP levels were measured before discharge. Commercially available assay kits were used for the measurement of TnT (4th generation high sensitivity assay,Roche Diagnostics, Tokyo, Japan) and BNP (Shionogi, Tokyo, Japan). The lower limit of detection for TnT was 0.01 ng/mL, and BNP was subclassified by level according to receiver operating characteristic (ROC) analysis (Cut-off value; 239.5 pg/mL). Serum creatinine and hemoglobin levels were determined by standard laboratory methods. Glomerular filtration rate (GFR) was estimated using the equation from the Modification of Diet in Renal Disease Study: eGFR (mL•min-1•1.73 m-2) = 0.741 × 175 × Age-0.203 × SCr-1.154 (0.724 for females). CKD was defined as an eGFR of <60 mL/min at the time of discharge. Diabetes mellitus was defined according to the criteria of the American Diabetes Association. Before discharge, experienced technicians who had no knowledge of the biochemical data performed two-dimensional echocardiography in a standard manner using a Hewlett Packard 5500. Clinical data were obtained by interviewing patients and from hospital medical records. Physicians were blinded to the data on biochemical markers and treatment was selected based on the patient's symptoms and physical findings.

Statistical Analysis
Categorical variables were expressed as numbers (percentages) and continuous variables were expressed as the mean  standard deviation. An unpaired t-test and chi-square test were used for between-group comparisons of continuous and categorical variables, respectively. If the data were skewed, the nonparametric Mann-Whitney test was used to compare continuous variables. ROC analysis was performed to determine the cut-off values for the conversion of continuous variables into categorical variables when analyzing BNP. Overall survival and survival without hospitalization for HF were analyzed by the Kaplan-Meier method, and the curves were compared by the log-rank test. We also performed an analysis based on the presence/absence of CKD.
Univariate and multivariate Cox regression analyses were performed to determine the associations between TnT and other variables. For each covariate, categorical variables were allowed to enter in a stepwise forward multivariate Cox model with the use of a probability value ≤0.10 for inclusion or ≤0.05 for deletion. The multivariate model included categorical variables that were statistically significant according to univariate analysis as well as clinically important. A P value of <0.05 was considered significant. Statistical analyses were performed with SPSS version 16.0 software (SPSS Inc., Chicago, Illinois).

Study Population
At baseline, the mean age of all patients was 66.6±15.3 years; 69.3% were men. About half of the patients had non-ischemic cardiomyopathy (47.5%). TnT was detectable (≥0.01 ng) in 73(29.9%) patients at discharge, with interquartile range 0.00 to 0.02. The median level of BNP was 206 pg/mL (interquartile range 108 to 490) at discharge. Notably, our population included a large group of patients with CKD (68.0%). Threshold for BNP was determined via ROC curve analysis (239.5 pg/mL). TnT threshold for detectable vs non detectable was 0.01 ng/mL.
The patients were subsequently divided into 3 groups based on biomarker levels at discharge: patients with both lower BNP levels and undetectable TnT (group 0, n=107); those with either higher BNP levels or detectable TnT (group 1, n=84); and those with both higher BNP level and detectable TnT (group 2, n=53). Patients with elevated biomarker levels also had a greater likelihood of diabetes mellitus (p=0.007), factors suggestive of impaired renal function (such as eGFR, p<0.001), elevated blood urea nitrogen (p<0.001), and a higher rate of CKD (p<0.001) and anemia (p<0.001) at discharge (Table 1).

Outcomes
During the 2-year follow-up period, 75(30.7%) events were recorded (28 deaths and 66 readmissions). Kaplan-Meier event curves comparing the prognosis of patients with higher vs. lower BNP levels, and patients with detectable vs. undetectable TnT levels are shown in Fig. 1A. During the follow-up period, primary event rates were higher among patients with elevated BNP levels (log-rank test; p<0.001) or detectable TnT (p=0.007) compared with those with lower BNP levels or undetectable TnT. When the patients were stratified into 3 groups based on biomarker values, the primary event rate was additively worse among patients with both increased BNP values and detectable TnT levels ( Fig. 1B;

Outcomes of the CKD Subgroup
The presence of CKD had a significant negative impact on patient survival; overall, CKD patients had significantly more primary events than did non-CKD patients (log-rank test; p=0.004). A separate analysis was performed to demonstrate the impact of biomarkers in CKD patients ( Fig. 2A) in comparison to non-CKD patients (Fig. 2B). Similar to our main results, primary events were more frequent among patients with higher BNP levels (log-rank test; p=0.007) or detectable TnT levels (p=0.082) compared with their counterparts. Event rates were also additively worse as the number of elevated biomarkers increased ( Fig. 2B; log-rank p=0.012).

Univariate and Multivariate Cox Hazard Models
The results of the univariate and multivariate models are shown in Table 2

Discussion
The present study revealed that discharge levels of both BNP and TnT were associated with the composite of HF readmission and all-cause mortality in patients with acute decompensate HF. In addition, the combination of elevated BNP and TnT levels further stratified the risk and had an additive effect on the patients' long-term prognosis. Measuring the combination of these biomarkers on discharge appeared to be a useful method for stratification of HF patients, including those with CKD.
Ample evidence has demonstrated the value of natriuretic peptides for predicting adverse outcomes, and the prognostic potential of BNP values was examined in the multicenter Rapid Emergency Department Heart Failure Outpatients Trail (REDHOT), which showed that BNP values were very strong predictors of 90-day outcomes (Maisel et al., 2004). The many mechanisms of BNP release in acute HF, including myocyte stretch and cardiac remodeling, show why these markers are so profoundly prognostic in HF patients.
Serum biomarkers of cardiac stress and malfunction as well as myocyte injury have grown in clinical importance for predicting the prognosis of HF patients. TnT is produced from cardiac myocytes as a consequence of myocardial ischemia. The level of TnT has a significant negative predictive value among HF patients with CKD (Koide et al., 2010). Furthermore, patients with preserved kidney condition may very well be able to tolerate ongoing myocardial injury, whereas patients with CKD are strongly affected by troponin leakage (Kociol et al., 2010).
The assessment of a combination of these conventional biomarkers could potentially improve the risk stratification of acute decompensated HF patients; moreover, kits for such assessments are both inexpensive and widely available. In previous studies, Ishii et al. reported that the combination of cardiac TnT and BNP levels on admission might be highly effective for the risk stratification of patients with chronic HF (Ishii et al., 2002), but little information is available on the association of this combination of biomarkers after acute decompensation of HF (Nishio et al., 2007;Taniguchi et al., 2006). In our study, CKD had a significant impact on the long-term outcome, as has previously been reported, and we demonstrated that patients with CKD can also be efficiently stratified by levels of BNP, TnT, and the combination of these 2 biomarkers. Patients with even mild chronic renal insufficiency have significantly increased cardiovascular morbidity and mortality, and chronic renal insufficiency also affects concentrations of biomarkers (Heywood et al., 2007;Ronco et al., 2008;McAlister et al., 2004). We showed, for the first time, that the prognostic value of biomarkers and their combination is not necessarily altered by the presence of renal impairment.
Our study has several limitations. Since we performed a retrospective analysis of registry data, we cannot establish cause and effect. However, the associations are consistent with prior analyses of troponin in patients with acute decompensate HF. Plenty of factors could have affected the results of the biomarker tests. First, we used the results of various biomarker assays for which we defined cutoff points rather than core laboratory results. Second, bias may have been introduced because we were unable to analyze patients with HF in whom troponin was not assessed, and we were unable to determine why physicians obtained, or did not obtain, biomarker measurements. Third, our cohort included mainly Japanese individuals, and there is some difference between Japan and western countries in terms of HF etiology and medication (Oshima et al., 2009).We also did not have information on BNP isoform. However, most of them showed equal predictability when compared to original BNP assay (van Kimmenade et al., 2009) and we believe that the absence of BNP isoform measurements will not alter our main results. Fourth, the 4 th generation cTnT assay was used in our study, which is no longer available in most parts of the world and BNP was measured with Shionogi assay which is not used outside of Japan. This BNP assay is known to provide lower values than most of the commercially available BNP assays. Thus, the study has limited generalizability, but we believe that they confirm the principle that discharge values of both markers are suitable for risk stratification. Finally, we excluded the patients undergoing haemodialysis (CKD stage 5) since effect of biomarkers was unclear in these patients. BNP is known to increase exponentially with the stage of renal disease and likely skew the main result of our analysis.

CONCLUSION
In conclusion, our study suggests that levels of both BNP and TnT have a significant predictive value for the prognosis of HF. Assessment of the combination of these biomarkers, which is both inexpensive and readily available, may provide additional information.