Role of novel cardiac biomarkers for the diagnosis, risk stratification, and prognostication among patients with heart failure

Heart failure (HF) is a leading cause of hospitalizations among adults and the elderly in the US, with estimated medical costs expected to increase from $20.9 billion in 2012 to $53.1 billion by 2030 [1]. The mean per-patient cost of an HF-related hospitalization in the US was $10,775 in 2011, and the prevalence of HF is estimated to be 5.7 million with an incidence of 870,000 new cases diagnosed every year. Approximately ifty percent of patients with HF are expected to die within ive years of diagnosis [2]. HF thus confers not only a substantial burden on patients, but the hospitalization rates are at epidemic proportions and strategies are needed that can direct prevention, early detection and successful, tailored therapy. The current ACC/AHA guidelines for the management of HF de ine the condition as a “complex clinical syndrome that results from any structural or functional impairment of ventricular illing or ejection of blood” and that practically there is a lack of a “single diagnostic test for HF because it is largely a clinical diagnosis based on a careful history and physical examination” [3]. Furthermore, studies have identi ied the limited reliability and decreased speci icity Abstract


Introduction
Heart failure (HF) is a leading cause of hospitalizations among adults and the elderly in the US, with estimated medical costs expected to increase from $20.9 billion in 2012 to $53.1 billion by 2030 [1]. The mean per-patient cost of an HF-related hospitalization in the US was $10,775 in 2011, and the prevalence of HF is estimated to be 5.7 million with an incidence of 870,000 new cases diagnosed every year. Approximately ifty percent of patients with HF are expected to die within ive years of diagnosis [2]. HF thus confers not only a substantial burden on patients, but the hospitalization rates are at epidemic proportions and strategies are needed that can direct prevention, early detection and successful, tailored therapy. The current ACC/AHA guidelines for the management of HF de ine the condition as a "complex clinical syndrome that results from any structural or functional impairment of ventricular illing or ejection of blood" and that practically there is a lack of a "single diagnostic test for HF because it is largely a clinical diagnosis based on a careful history and physical examination" [3]. Furthermore, studies have identi ied the limited reliability and decreased speci icity

Biomarker assays
The research-use-only Erenna Immunoassay System, (Singulex, Inc., CA), powered by high-sensitivity singlemolecule counting (SMC) technology, was used to measure the biomarkers cTnl, TNF-α, IL-6, ET-1, and pMMP-9 in EDTA plasma samples. Plasma samples, controls, capture and detection reagents, and standards were added to a 96well plate and incubated. The aforementioned biomarkers of interest were bound to speci ic capture antibodies biotinylated to microparticles and luorescently-conjugated detection antibodies. The unbound luorescent antibodies were removed by a wash procedure, which was then followed by the addition of elution buffer to release luorescent detection antibody into the eluent. The remaining contents of the 96-well plate were transferred to a 384-well plate and onto the Erenna system for SMC by counting photons passing through an interrogation space, which was directly proportional to the amount of biomarker within the sample. NTproBNP was measured on Roche cobas 6000 System using electrochemilluminescence immunoassay.

Statistical analysis
Descriptive statistics of baseline demographic and clinical characteristics were reported as continuous variables with median (interquartile range or IQR) and percentages for categorical variables. The cross-sectional associations of continuous biomarker measurements for NTproBNP, cTnI, ET-1, IL-6, TNF-α, and pMMP-9 across the four ACC/AHA HF stages were reported by expressing the median (IQR) for each biomarker at each HF stage. Post-hoc comparisons were then used to assess for statistically signi icant differences between HF stages and biomarker concentrations using Bonferroni family-wise error rate of α=0.05. Kaplan-Meier curves were generated to assess transplant-and advanced therapy-free survival with differences determined using the log-rank test dichotomizing cTnI and IL-6 at their median values. Additional comparisons were made between Cox Hazard Ratios that were 1) unadjusted, 2) adjusted by age and sex, 3) adjusted for ACC/AHA HF classi ication stages, and 4) adjusted for age, sex, race, estimated glomerular iltration rate, blood urea nitrogen, body mass index, systolic blood pressure, and ejection fraction to predict irst incidence of death, heart transplant, or LVAD implantation.

Study subjects
A total of 200 patients within the Heart Failure cohort were enrolled and the demographics are listed in Supplemental  Table 1. The plasma biomarker concentrations for all patients are detailed in Table 1 and highlighted in Figure 1. Baseline demographic information (median age 54 years) for patients within the Heart Failure cohort are presented in Supplemental Table 1. The majority of patients enrolled had of physical exam indings (jugular venous distention, rales, edema) for the diagnosis of HF [4,5]. Natriuretic peptides (i.e. NT-proBNP) have been identi ied as biomarkers that assess the presence and severity of HF in the context of the aforementioned diagnostic limitations. The challenge is that confounders remain, such as obesity, increased age, renal insuf iciency, and anemia that may impact the diagnostic sensitivity and speci icity of these traditional cardiac biomarkers [6,7]. There are several promising cardiovascularbased biomarkers including: high-sensitivity cardiac troponin I (cTn1), a myo ibrillar protein released in the setting of supply/ demand mismatch; interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α), in lammatory markers associated with myocyte apoptosis and remodeling; endothelin-1 (ET-1), an acute-phase reactant that stimulates myocardial hypertrophy and cell injury; and pro-matrix metalloproteinase-9 (pMMP-9), a gelatinase subtype that contributes to left ventricular remodeling [7][8][9][10][11][12][13]. Whether these biomarkers can be incorporated routinely into standard clinical practice or if they are accurate prognosticators remains to be evaluated [14]. The objective of this cross-sectional study was to investigate the individual and combined role of biomarkers NT-proBNP, cTn1, IL-6, ET-1, pMMP-9, and TNF-α for risk strati ication and prediction of HF severity and disease progression in patients with ACC/AHA stages A-D ischemic and non-ischemic cardiomyopathy (CM).

Study subjects
Two hundred patients with ACC/AHA Stages A-D were retrospectively identi ied from the Vanderbilt Heart and Vascular Institute (VHVI) Main Heart Registry and Biorepository. Written consent and plasma samples were obtained from each patient at time of enrollment following approval of the Vanderbilt Institutional Review Board. Past medical history, cardiovascular risk factors, demographic information, smoking history, ACC/AHA HF stage at time of enrollment, and medication history were collected. All patients who consented were included based on the following criteria: diagnosis of HF with varying severity, secondary to either ischemic or non-ischemic CM, with or without the presence of malignancy. Patients identi ied were followed from date of enrollment-considered to be baseline outpatient visit-and censored at the last date of follow-up. Information on dates of death, transplants, or implantation of left ventricular assist device (LVAD) were collected through electronic medical records or published obituaries. Baseline demographic data were collected using Vanderbilt University Medical Center electronic health records. Patients were excluded if they had received inotropic therapy during time of blood draw. Approximately 400 μL of EDTA plasma was obtained from frozen samples stored at -70°C. De-identi ied samples were then sent for analysis to Singulex Clinical Laboratory (Alameda, CA).

Discussion
This study examined the diagnostic and prognostic utility of several biomarkers among individuals with heart failure and left ventricular systolic dysfunction. We begin by demonstrating that NTproBNP, cTnI, IL-6, ET-1, and pMMP-9 are elevated in early, asymptomatic stages of HF. From this, we identi ied signi icant differences in plasma concentrations of NTproBNP, pMMP-9, and IL-6 between stage D and earlier HF stages, suggesting these might correlate with disease severity. With regards to prognostication, we observed that the combined effects of higher levels of NTproBNP, cTnI, and IL-6 were associated with worse survival compared to lower concentrations of these biomarkers. Additionally, when adjusted for ACC/AHA HF stages, there was a signi icant difference between IL-6 and cTnI concentrations with regards to time to death, LVAD, or heart transplantation. * * * Figure 1: Plasma biomarker measurements across all ACC/AHA HF stages. Boxand whisker plots of plasma biomarker concentrations with box plot boundaries representing 25 th and 75 th percentile, horizontal line denoting the median, whisker extending to 1.5x interquartile range; Statistical signifi cance determined by Bonferroni post-hoc analyses to compare concentrations across HF groups; *p<0.05, denoting signifi cant differences in plasma biomarker concentrations in stage D HF compared to stages A, B, and C for NTproBNP, pMMP-9, and IL-6; ACC/ AHA, American College of Cardiology/American Heart Association; NTproBNP, N-terminal pro-Brain Natriuretic Peptide; cTnI, high sensitivity Troponin I; ET-1, Endothelin-1; pMMP-9, plasma Matrix Metalloproteinase-9; TNF, Tumor Necrosis Factor-alpha; IL-6, Interleukin-6. These indings contribute to the current understanding in which pleiotropic cytokine IL-6 regulates host-immune response, speci ically in the setting of myocardial injury. While it has been shown to exert cardioprotective effects on myocytes during the acute response by reducing oxidative stress and enhancing anti-apoptotic pathways, chronically increased expression and signaling of IL-6 eventually depresses contractility through activation of calcium-independent iNOS proteins and leads to pathological hypertrophy. Regarding its role in HF mortality, epidemiological studies have shown that elevated IL-6 concentrations portend worse prognosis for patients with HF and correlate with HF severity [11,[15][16][17][18][19][20][21][22][23][24][25][26][27][28]. More notably, cytokine database analysis from the Vesnarinone trial (VEST) demonstrated that among the 1200 individuals enrolled in the multicenter trial receiving placebo, increased concentrations of circulating IL-6 were associated with increased mortality, with cytokine levels signi icantly higher in patients with ischemic heart disease compared to non-ischemic [29]. Plasma IL-6 is also strongly associated with cardio-renal dysfunction and neurohormonal activation in patients with HF on high-dose loop diuretics [15]. Our study has demonstrated that IL-6 is elevated during asymptomatic stages of HF, with high plasma levels independently predicting time to LVAD, heart transplantation, or death from all causes after adjusting for ACC/AHA HF stages.
In addition, we have shown that elevated IL-6 concentrations in combination with NTproBNP and cTnI predicted worse survival for patients with HF after a 1500-day followup period. This study con irms previously established indings regarding the diagnostic and prognostic utility of NT-proBNP and cTnI. The BNP gene is activated in the setting of myocardial wall stress that ultimately leads to the processing of precursor proBNP into NTproBNP and BNP [30]. Increased levels of NTproBNP predict worse prognosis for HF and LVSD, and correspond with increased HF severity [30]. Although conventional assay systems have been less sensitive compared to BNP and NTproBNP kits, cTnI remains an important marker of survival and decompensation as it is also up-regulated during increased wall or oxidative stress [31]. We have demonstrated that rising levels of NTproBNP and cTnI were associated with increased HF severity, and that together with IL-6, predict worse survival in patients with HF and LVSD. This supports the role of IL-6 as an additive marker to aid in risk strati ication and monitoring response to therapy. To our knowledge, this study is among the largest that con irms prior observations regarding the potential maladaptive role IL-6 plays in HF progression and mortality.
Regarding ET-1, we have shown that it is elevated during asymptomatic stages of HF. On a molecular level, ET-1 is an important mediator of cardiomyocyte hypertrophy through up-regulation of COX-2 resulting in an increase in cell surface area of rat neonatal cardiomyocytes and BNP mRNA synthesis, an effect that is augmented by its inter-dependency with angiotensin and norepinephrine [32,33]. In terms of its role in HF, ET-1 causes both afferent and efferent arteriole constriction, reducing renal blood low and GFR [33]. ET-1 has also been shown to be an independent, short-term predictor of mortality in patients with heart failure with reduced ejection fraction (HFrEF) and acute heart failure [34,35]. These indings support the role of ET-1 as a potential diagnostic marker for HF during earlier stages, but the data for its role in risk strati ication and prognostication is less compelling compared to the other biomarkers in this study ( Table 3).
The diagnostic utility of pMMP-9 in the setting of heart failure and LVSD was also examined, with results suggesting that pMMP-9 is detectable in earlier stages of heart failure and correlates with HF severity. While pMMP-9 values independently predicted time to event in the fully adjusted model (HR 1.8; CI 1.1-2.9; p=0.02), we did not observe the Patients with elevated levels of all three biomarkers had signifi cantly higher risk of adverse events compared to those with any one of the biomarkers elevated and those with biomarker concentrations below the median cutoff.  Continuous biomarker concentrations depicted as median (interquartile range Q1, Q3); ACC/AHA, American College of Cardiology/American Heart Association; NTproBNP, N-terminal pro-Brain Natriuretic Peptide; cTnI, high sensitivity Troponin I; ET-1, Endothelin-1; pMMP-9, plasma Matrix Metalloproteinase-9; TNF-, Tumor Necrosis Factor-alpha; IL-6, Interleukin-6.
same effect in the model adjusted for ACC/AHA HF stages. Upregulation of MMP-9 contributes to ventricular remodeling through increased gelatinolytic activity that results in matrix degradation and myocardial collagen turnover [36,37]. However, epidemiologic studies have produced mixed results on the predictive utility of markers of collagen turnover such as MMP-9 in the setting of HF. Previous indings have demonstrated the former two observations showing that levels of pMMP-9 increased with increasing severity of NYHA class, and that higher pMMP-9 values were associated with higher probability of HF events alongside LV remodeling [38]. Yet, other studies have not observed such correlations between pMMP-9 levels, HF severity, and adverse outcomes, with another identifying it as a useful risk strati ication marker in systolic HF only [39][40][41]. Additionally, when compared to BNP, pMMP-9 remained a poor clinical biomarker for ventricular remodeling and HF events [39]. Nonetheless, while its prognostic utility remains unclear, pMMP-9 may be a useful biomarker to measure alongside natriuretic peptides in order to evaluate for disease severity.
This study has several limitations. Several patients were lost to follow-up, and further adjustments could have been made during analysis to account for various causes of HF (i.e. ischemic vs. non-ischemic). However, other parameters were adjusted during survival analysis, including ACC/AHA HF stages, creatinine values, and LVEF. Furthermore, we have been able to retrospectively quantify on a relatively large scale a panel of diverse biomarkers in patients with HF and LVSD. As no interventions were performed, this limits the presence of confounding variables, which provided an assessment of the endogenous effects of these circulating biomarkers. Consequently, this has allowed for the investigation of both their diagnostic utility during early stages of HF, as well as prognosticating potential for patients with HF events despite optimal medication therapy.

Conclusion
These indings support the clinical applicability of NTproBNP, cTnI and IL-6 among patients with HF and LVSD for diagnostic and prognostic purposes, particularly when measured together. These biomarkers may also aid in risk strati ication of asymptomatic individuals before symptoms of wall stress and volume overload become manifest, and these data could provide insight regarding preferred treatments to potentially prevent adverse outcomes. Future studies involving precision medicine are needed to examine and compare the individual and combined effects of multiple biomarkers on early detection and response to HF treatment based on individual pro iles and development of targeted therapies in order to optimize HF management.

Con licts of interest
DJL is a consultant for P izer, Roche, and Prothena, none of which are relevant to this study. Singulex employees were employed by Singulex Clinical Laboratory at the time of this study. All other authors report no con licts of interest. Hazard ratios adjusted for 1) ACC/AHA classifi cation and 2) age, sex, race, baseline estimated glomerular fi ltration rate, blood urea nitrogen level, body mass index, baseline left ventricular ejection fraction, and baseline systolic blood pressure; ACC/AHA, American College of Cardiology/American Heart Association; NTproBNP, N-terminal pro-Brain Natriuretic Peptide; cTnI, high sensitivity Troponin I; ET-1, Endothelin-1; pMMP-9, plasma Matrix Metalloproteinase-9; TNF-, Tumor Necrosis Factor-alpha; IL-6, Interleukin-6.