Cardiac myosin-binding protein C in the diagnosis and risk stratification of acute heart failure

: AIMS Cardiac myosin-binding protein C (cMyC) seems to be even more sensitive in the quantification of cardiomyocyte injury vs. high-sensitivity cardiac troponin, and may therefore have diagnostic and prognostic utility. METHODS AND RESULTS In a prospective multicentre diagnostic study, cMyC, high-sensitivity cardiac troponin T (hs-cTnT), and N-terminal pro-B-type natriuretic peptide (NT-proBNP) plasma concentrations were measured in blinded fashion in patients presenting to the emergency department with acute dyspnoea. Two independent cardiologists centrally adjudicated the final diagnosis. Diagnostic accuracy for acute heart failure (AHF) was quantified by the area under the receiver operating characteristic curve (AUC). All-cause mortality within 360 days was the prognostic endpoint. Among 1083 patients eligible for diagnostic analysis, 51% had AHF. cMyC concentrations at presentation were higher among AHF patients vs. patients with other final diagnoses [72 (interquartile range, IQR 39-156) vs. 22 ng/L (IQR 12-42), P < 0.001)]. cMyC’s AUC was high [0.81, 95% confidence interval (CI) 0.78-0.83], higher than hs-cTnT’s (0.79, 95% CI 0.76-0.82, P = 0.081) and lower than NT-proBNP’s (0.91, 95% CI 0.89-0.93, P < 0.001). Among 794 AHF patients eligible for prognostic analysis, 28% died within 360 days; cMyC plasma concentrations above the median indicated increased risk of death (hazard ratio 2.19, 95% CI 1.66-2.89; P < 0.001). cMyC’s prognostic accuracy was comparable with NT-proBNP’s and hs-cTnT’s. cMyC did not independently predict all-cause mortality when used in validated multivariable regression models. In novel multivariable regression models including medication, age, left ventricular ejection fraction, and discharge creatinine, cMyC remained an independent predictor of death and had no interactions with medical therapies at discharge. CONCLUSION Cardiac myosin-binding protein C may aid physicians in the rapid triage of patients with suspected AHF. Aims Cardiac myosin-binding protein C (cMyC) seems to be even more sensitive in the quantification of cardiomyocyte injury vs. high-sensitivity cardiac troponin, and may therefore have diagnostic and prognostic utility. ..................................................................................................................................................................... T and plasma in in Two final Diagnostic accuracy for acute failure was quantified by the area under the receiver characteristic curve All-cause mortality within 360days was the prognostic endpoint. Among 1 083 diagnostic 5 1 AHF. cMyC at higher among vs. final range, vs. 2–42), , AHF P < 0.00 1 ). cMyC’s prognostic accuracy was comparable with NT-proBNP’s and hs-cTnT’s. cMyC did not independently predict all-cause mortality when used in validated multivariable regression models. In novel multivariable regression models including medication, age, left ventricular ejection fraction, and discharge creatinine, cMyC remained an independent predictor of death and had no interactions with medical therapies at discharge. ..................................................................................................................................................................... which is used for detection by the diagnostic assay. 1 3, 1 5, 1 6 The fast release kinetics of cMyC quantifying cardiomyocyte injury and its abundance in cardiomyocytes led to the hypothesis that cMyC concentrations may have diagnostic, prognostic, and therapy guidance potential in patients with AHF. 11 , 1 2, 1 7 therefore, this of patients presenting to with suspected AHF. concentration for the rapid rule out of AHF was derived from the receiver operating characteristic (ROC) curve. The requirements for this cut-off concentration were imposed by clinical considerations: a minimum sensitivity was set as high as 95% because of the large potential adverse outcome of missing a patient with AHF. The relevant 95% confidence intervals (CIs) of the predictive values were defined by using the Wilson score method without continuity cor-rection. The utility of cMyC to predict all-cause mortality in AHF patients was assessed by comparing the time-dependent ROC curves of cMyC, hs-cTnT, and NT-proBNP. 27 All-cause mortality and AHF rehospitalization as well as the combination of both endpoints during follow-up were plotted in Kaplan–Meier curves, and the log-rank test was used to assess differences between groups. Spearman’s rho was used to analyse correlations between cMyC and selected variables. cMyC and pre-defined variables from validated risk models to predict all-cause mortality, or the combination of all-cause mortality or hospitalizations due to AHF were entered in multivariable Cox proportional hazard models. 28 In contrast to the validated risk model for the prediction of the combination of all-cause mortality or hospitalizations due to AHF, high-density lipoprotein plasma concentrations at baseline were not available. Interactions between cMyC plasma concentrations and treatment with diuretics, ACE-I or ARB, beta-blockers, and aldosterone antagonists at hospital discharge were explored. The interaction of P -values between biomarker plasma concentrations and the pre-defined AHF subgroups according to their medication on discharge was calculated in bivariable Cox proportional hazards models. These bivariable models were then further adjusted for age, left ventricular ejection fraction (LVEF) and creatinine plasma concentrations at discharge. Hazard ratios are presented with 95% CI. Subgroup analyses performed in renal cMyC, hs-cTnT, and NT-proBNP. This was an exploratory analysis within a prospective study, and sample size of the overall cohort was not specifically determined for this analysis. No imputation was performed for missing values. Patients without complete 1 -year clinical follow-up were censored at the time of the last known contact. Statistical analyses were performed using SPSS/PC Software Package (version 25.0) and R Statistical Software (version 3.5. 1 ), including packages ‘pROC’, ‘timeROC’, ‘cmprsk’, ‘sur-vival’, ‘ggplot’, ‘tableone’, ‘haven’, ‘tibble’, and ‘predictABLE’.


Introduction
Acute heart failure (AHF) is the most common diagnosis in the emergency department (ED) leading to hospitalization. 1,2 In contrast to the enormous improvements achieved in the management of patients with chronic heart failure, morbidity and mortality remain unacceptably high in patients with AHF. Specifically, the combination of all-cause mortality or recurrent hospitalization at 6 months approaches 50%. [1][2][3] The dismal outcome of patients with AHF may be at least in part related to diagnostic and prognostic uncertainty in the ED and the associated delay in the identification of patients for early and effective treatments and an incomplete understanding of the pathophysiological mechanisms involved in the AHF syndromes, which may require tailored therapeutic strategies. 4 Cardiovascular biomarkers incorporated in clinica practice address some of these challenges. First, as quantitative markers of haemodynamic stress and heart failure, natriuretic peptides have substantially improved the rapid detection of AHF among patients presenting with acute dyspnoea. 1,2,5-8 Second, high-sensitivity cardiac troponin (hs-cTn) concentrations allow quantification of cardiomyocyte injury and enhance risk stratification of AHF patients. 6,9,10 Cardiac myosin-binding protein C (cMyC) (Figure 1)i sa novel biomarker that is unique to cardiomyocytes and can possibly quantify cardiomyocyte injury even more accurately than hs-cTn. [11][12][13] In addition, it has shown promising diagnostic and prognostic utility in the management of acute coronary syndromes. 11 cMyC may also have a role in the pathogenesis of AHF as various animal models for hypertrophic cardiomyopathy have shown that homozygous cMyC knock-out mice develop heart failure. 14 Definite pathogenic mutations within cMyC are the most common genetic cause of hypertrophic cardiomyopathy, responsible for 40-50% of new diagnoses. 15 These mutations, and known post-translational modifications of the cMyC protein, occur outside the N-terminal C0 region of the protein which is used for detection by the diagnostic assay. 13,15,16 The fast release kinetics of cMyC quantifying cardiomyocyte injury and its abundance in cardiomyocytes led to the hypothesis that cMyC concentrations may have diagnostic, prognostic, and therapy guidance potential in patients with AHF. 11,12,17 We, therefore, aimed to test this hypothesis in a large multicentre cohort of patients presenting to the ED with suspected AHF.

Study population and design
Basics in Acute Shortness of Breath EvaLuation (BASEL V) was a prospective, multicentre diagnostic study aimed at contributing to advancing the early detection and management of patients with AHF (ClinicalTrials.gov, NCT01831115). 18,19 Adult patients presenting with acute dyspnoea to the ED of two university hospitals and three further tertiary care centres in Switzerland were enrolled (Basel, Zurich, Lucerne, St. Gallen, and Aarau). While enrolment was independent of renal function, patients with terminal kidney failure on chronic dialysis were excluded. For this analysis, patients were also excluded if they did not have cMyC plasma concentrations measured from study blood samples at ED presentation, if the final diagnosis remained unclear even after central adjudication, or if the patients were adjudicated as having cardiac dyspnoea due to an acute coronary syndrome or arrhythmia without any other evidence of AHF. For the prognostic analyses, patients with an adjudicated final diagnosis of AHF enrolled in an AHF therapy study were also eligible if study blood was available for the measurement of cMyC. 20 The study was carried out according to the principles of the Declaration of Helsinki and was approved by the local ethics committees. All patients provided written informed consent. The authors designed the study, gathered, and analysed the data according to the STARD guidelines for studies of diagnostic accuracy and the TRIPOD statement for studies reporting multivariable prediction models for individual prognosis (online supplementary Table S1), 21,22 vouched for the data and analysis, wrote the paper, and decided to submit it for publication.

Adjudication of the final diagnosis
The final underlying cause of shortness of breath was centrally adjudicated by two independent cardiologists who had access to all patients' medical records (clinical history, physical examination, 12-lead electrocardiogram, laboratory findings including natriuretic peptide measurements, 1,2 estimated glomerular filtration rate, chest X-ray, echocardiography, lung function testing, computed tomography, response to therapy, and 360-day follow-up, according to current guidelines). 1,2 cMyC measurements were not available for the adjudication of the final diagnosis. In situations of disagreement related to the diagnosis, cases were reviewed and adjudicated in conjunction with a third cardiologist.

Patient follow-up
Patients were contacted after 90 days and at 1 year by telephone calls or in written form by trained researchers, unaware of the patients' cMyC plasma concentrations during the index hospitalization. In case of a possible relevant medical event such as heart failure rehospitalization or all-cause death during the follow-up period, further information was obtained from the hospital records, general physician records, or the national death registry.

Biochemical measurements
At ED presentation, blood samples were collected in tubes containing heparin for the determination of cMyC and high-sensitivity cardiac troponin T (hs-cTnT) concentrations, and potassium ethylenediaminetetraacetic acid for analysis of N-terminal pro-B-type natriuretic peptide (NT-proBNP), respectively. After centrifugation, samples were frozen at −80 ∘ C until assayed in a blinded fashion in a dedicated core laboratory. cMyC was measured using an established high-sensitivity assay on the Erenna platform performed by Millipore Sigma. 13 The assay has a limit of detection of 0.4 ng/L and a lower limit of quantification of 1.2 ng/L with a ≤20% coefficient of variation at limits of quantification, and ≤10% coefficient of variation at the 99th centile. Assay precision is not affected by freeze/thaw cycles. 23 The 99th percentile concentration determined previously in patients without obstructive coronary artery disease on invasive angiography is 87 ng/L. 11,13 In this study, the 99th centiles calculated for hs-cTnI/T matched the established thresholds. The hs-cTnT assay (Elecsys 2010 high-sensitivity troponin T, Roche Diagnostics, Rotkreuz, Switzerland) has a 99th percentile concentration of 14ng/L. 24 NT-proBNP concentrations were measured using the Elecsys proBNP assay (Elecsys proBNP, Roche Diagnostics AG, Rotkreuz, Switzerland). The intra-assay coefficients of variation ranged from 1.8% to 2.7% and from 2.4% to 3.2% for within-run and total imprecision, respectively. 25 The laboratory technicians who measured cMyC, hs-cTnT, and NT-proBNP were blinded to the clinical data of the patients.

Outcome measures
The diagnostic accuracy for AHF quantified by the area under the receiver operating characteristic curve (AUC) was the primary diagnostic outcome measure. Only patients with confirmed final diagnosis of AHF were eligible for the prognostic analyses. The primary prognostic outcome measure was the utility of cMyC to predict all-cause mortality in AHF patients within 360 days. Secondary outcome measures were the prediction of AHF rehospitalizations; the combination of all-cause mortality or AHF rehospitalizations; the identification of AHF phenotypes according to cMyC plasma concentrations with disproportional benefit or harm in terms of all-cause mortality during the 360-day follow-up in pre-defined subgroups according to their treatment with diuretics, angiotensin-converting enzyme inhibitors (ACE-I) or angiotensin receptor blockers (ARB), beta-blockers, aldosterone antagonists, calcium channel blockers, and digoxin at hospital discharge; the diagnostic and prognostic performance in the subgroup of patients with renal dysfunction.

Assessment of renal function
For this analysis, renal dysfunction was defined as estimated glomerular filtration rate <60 mL/min/1.73 m 2 at presentation to the ED and was calculated by using the Chronic Kidney Disease Epidemiology Collaboration formula.

Statistical methods
The Kolmogorov-Smirnov test and visual inspection of the shape of the distribution of the variables were used to assess their normality. Continuous variables are presented as medians with interquartile range (IQR) as appropriate. Categorical variables are presented as numbers and percentages. Comparisons between groups were made using the Chi-square test, the Mann-Whitney U test, or the Kruskal-Wallis test as appropriate. Comparison of the AUCs of cMyC, hs-cTnT, and NT-proBNP was performed as recommended by DeLong et al. 26 The diagnostic accuracy of the combination of cMyC and NT-proBNP was assessed by logistic regression analysis. The optimal cMyC cut-off concentration for the rapid rule out of AHF was derived from the receiver operating characteristic (ROC) curve. The requirements for this cut-off concentration were imposed by clinical considerations: a minimum sensitivity was set as high as 95% because of the large potential adverse outcome of missing a patient with AHF. The relevant 95% confidence intervals (CIs) of the predictive values were defined by using the Wilson score method without continuity correction. The utility of cMyC to predict all-cause mortality in AHF patients was assessed by comparing the time-dependent ROC curves of cMyC, hs-cTnT, and NT-proBNP. 27 All-cause mortality and AHF rehospitalization as well as the combination of both endpoints during follow-up were plotted in Kaplan-Meier curves, and the log-rank test was used to assess differences between groups. Spearman's rho was used to analyse correlations between cMyC and selected variables. cMyC and pre-defined variables from validated risk models to predict all-cause mortality, or the combination of all-cause mortality or hospitalizations due to AHF were entered in multivariable Cox proportional hazard models. 28 In contrast to the validated risk model for the prediction of the combination of all-cause mortality or hospitalizations due to AHF, high-density lipoprotein plasma concentrations at baseline were not available. Interactions between cMyC plasma concentrations and treatment with diuretics, ACE-I or ARB, beta-blockers, and aldosterone antagonists at hospital discharge were explored. The interaction of P-values between biomarker plasma concentrations and the pre-defined AHF subgroups according to their medication on discharge was calculated in bivariable Cox proportional hazards models. These bivariable models were then further adjusted for age, left ventricular ejection fraction (LVEF) and creatinine plasma concentrations at discharge. Hazard ratios are presented with 95% CI. Subgroup analyses were performed in patients with renal dysfunction to compare the diagnostic and prognostic accuracy of cMyC, hs-cTnT, and NT-proBNP. This was an exploratory analysis within a prospective study, and sample size of the overall cohort was not specifically determined for this analysis. No imputation was performed for missing values. Patients without complete 1-year clinical follow-up were censored at the time of the last known contact. Statistical analyses were performed using SPSS/PC Software Package (version 25.0) and R Statistical Software (version 3.5.1), including packages 'pROC', 'timeROC', 'cmprsk', 'survival', 'ggplot', 'tableone', 'haven', 'tibble', and 'predictABLE'.

Patient demographics and characteristics
A total of 1330 patients were enrolled between November 2007 and August 2013. Of these, 1083 were eligible for the diagnostic analysis (online supplementary Figure S1). In the overall cohort, the median age was 75 years, and 42% of all patients were women ( Table 1). Roughly one third (35%) of the patients had a history of AHF, and 41% had known coronary artery disease.

Acute heart failure and cardiac myosin-binding protein C plasma concentrations
Acute heart failure was the adjudicated final diagnosis in 548 patients (51%) included in the diagnostic analysis and in 794 patients (60%) in the overall cohort. cMyC concentrations at .   supplementary Table S4).
Beyond the presence or absence of AHF, when comparing different cMyC quartiles, patients with higher plasma concentrations were more likely to be older and to have prior history of coronary artery disease. Patients in the top cMyC quartile also had a lower LVEF, and higher NT-proBNP as well as higher creatinine concentrations (online supplementary Table S5).

Interaction with treatment at hospital discharge
In bivariable Cox proportional hazards models, a significant interaction was found between cMyC plasma concentrations at presentation and treatment with ACE-I at discharge (online supplementary Table S7A). After multivariable adjustment including age, LVEF, and creatinine plasma concentrations at discharge, the interactions regarding cMyC plasma concentrations lost their statistical significance (online supplementary Table S7B).

Discussion
This secondary analysis within a large prospective diagnostic study aimed to evaluate the clinical utility of cMyC in the diagnosis, risk stratification and therapy guidance of patients with AHF. 1,2 We report six major findings.
First, in line with our hypothesis, cMyC concentrations were higher in patients with an adjudicated final diagnosis of AHF vs. patients with other causes of acute dyspnoea. In the context of the ability of cMyC to precisely quantify cardiac injury, this finding is in full agreement with the previously reported high incidence of elevated hs-cTn concentrations in AHF. 2,9,10 Second, the diagnostic accuracy of cMyC for AHF was high, being slightly higher than hs-cTnT, but lower than the current laboratory gold standard for NT-proBNP. Third, cMyC might help in better AHF phenotyping, as the concentrations were higher in more severe forms such as pulmonary oedema and AHF combined with acute coronary syndrome. Fourth, all-cause mortality was higher in patients with above median cMyC concentrations. Overall, the prognostic accuracy was moderate-to-high and comparable to that of NT-proBNP and hs-cTnT, with some incremental prognostic value on top of NT-proBNP or hs-cTnT. However, in the fully adjusted multivariable prediction model, cMyC was no longer an independent predictor of death. 28 Fifth, in the bivariable Cox .............. proportional hazards models, a significant interaction between cMyC plasma concentrations at presentation and treatment with ACE-I at discharge was found, indicating that patients with higher cMyC concentrations might derive particular long-term benefits from these drugs. As the interaction was no longer significant  .................................................................................................................................... Values are given as median (interquartile range), or n (%). ACE, angiotensin-converting enzyme; AHF, acute heart failure; ARB, angiotensin receptor blocker; BMI, body mass index; CAD, coronary artery disease; cMyC, cardiac myosin-binding protein C; CCB, calcium channel blocker; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; ED, emergency department; hs-cTnT, high-sensitivity cardiac troponin T; LVEF, left ventricular ejection fraction; NT-proBNP, N-terminal pro-B-type natriuretic peptide; PAD, peripheral arterial disease; SBP, systolic blood pressure. after multivariate adjustment including LVEF, the incremental value of cMyC in the identification of AHF patients deriving long-term benefits from ACE-I remains uncertain. Future studies assessing dynamic changes in cMyC seem warranted, since the change in concentrations may be more relevant for therapy guidance. 29 The dynamic nature of cMyC, with its fast release kinetics and abundance in cardiomyocytes, may provide a particular advantage in this setting. 11,12 To our knowledge, this is the first large prospective clinical trial to explore the diagnostic and prognostic utility of cMyC for AHF in adults and has demonstrated incremental value to its potential use in the ED. These findings extend and corroborate . previous work on the clinical use of cMyC for the diagnosis of AHF in children. In a prospective case-control study that involved 50 children with AHF and 25 healthy children matched for age and sex as a control group, the high diagnostic and prognostic value of cMyC was demonstrated. With the use of a cut-off concentration of 45 ng/mL, the sensitivity reached 100% and its specificity for diagnosing AHF was 96%, while at the cut-off concentration of 152 ng/mL, its sensitivity and specificity for predicting adverse outcomes were as high as 90% and 93%, respectively. The AUC was 0.999 (95% CI 0.997-1.002) for the diagnosis and 0.915 (95% CI 0.796-1.034) for the prognosis of AHF. 30 However, comparisons between the results of the two studies should be assessed with caution, mainly since the aetiology of AHF in children with a mean age < 2 years in this study substantially differed when compared to the elderly adult AHF population as in our study. BASEL V has important methodological strengths including its large sample size, highly representative elderly patient population for acute dyspnoea and AHF, 18,19 and adjudicated final diagnosis by two independent cardiologists/internists according to current guidelines. 1,2 This study also has several limitations. First, its findings are specific to patients presenting with acute dyspnoea to the ED, and may not apply to the use of cMyC for screening in asymptomatic patients or in patients presenting with very mild symptoms to a general practitioner. 1,2 Further studies are necessary to address uncertainties regarding cut-off concentrations in these settings. Second, despite using a very strong methodology for the central adjudication of the final diagnosis, a very small number of patients may still have been misclassified. This would have led to an underestimation of the diagnostic accuracy of cMyC. Third, in order to maximize the accuracy in the adjudication of the final diagnosis leading to ED presentation with dyspnoea in BASEL V, central adjudication included BNP or NT-proBNP measurements, putting the blinded cMyC measurement at a disadvantage for direct comparison of diagnostic accuracy. Accordingly, the real difference in diagnostic accuracy between cMyC and NT-proBNP may be smaller than found in this study. Fourth, further studies are needed to prospectively validate the cut-offs for the optimal clinical use of cMyC for the diagnosis and prognosis of AHF in the ED, taking into account clinical characteristics that could potentially confound the performance of cMyC. Accordingly, as a biomarker in the evaluation of patients presenting with possible AHF, cMyC should always be used in conjunction with all other clinical information. Fifth, this study required written informed consent. Therefore, as for all studies requiring written informed consent, selection bias towards the enrolment of patients eligible to provide consent was unavoidable. Sixth, while enrolment was independent from renal function, and a substantial number of patients with renal dysfunction were included in this analysis, this study did not include patients with terminal kidney failure on chronic haemodialysis. Accordingly, we cannot comment on the performance of cMyC in this vulnerable patient population.
In conclusion, this large multicentre diagnostic study using central adjudication demonstrated that cMyC plasma concentrations may aid physicians in the rapid triage of patients presenting to the ED with suspected AHF.

Supplementary Information
Additional supporting information may be found online in the Supporting Information section at the end of the article.