Improving Risk Stratification in Heart Failure: A Role for New Biomarkers?

This editorial refers to ‘The predictive value of stable precursor fragments of vasoactive peptides in patients with chronic heart failure: data from the GISSI-heart failure (GISSI-HF) trial’, by S. Masson et al., on page 338. Even with modern therapy and improved diagnosis, heart failure remains a lethal condition, characterized by high mortality and hospitalization rates.1,2 Early identification of individuals at higher risk for adverse events could lead to earlier intervention that could potentially improve outcomes. Although consideration of demographic and clinical status in combination with indices of cardiac and renal function can identify a large proportion of at-risk individuals, risk stratification based on these factors alone remains imprecise, with some at-risk individuals not recognized and others inappropriately marked as high risk. Measurement of molecular biomarkers has improved our understanding of heart failure pathophysiology and can help to refine risk stratification.3 An ideal molecular biomarker should be highly sensitive and specific for the condition in question.3,4 From an analytic point of view, there should be excellent analytical sensitivity and assays should be standardized and reproducible, with clearly defined normal ranges based on age and gender.5 To be clinically useful, measurement of a biomarker should influence clinical management in a cost effective manner—either by improving diagnosis or by refining risk stratification or guiding therapy.5 A large number of molecular biomarkers have been evaluated in heart failure and to date the B-type natriuretic peptides (BNP) have become established as the most ideal markers available.6 This group of peptides includes the 126 amino acid prohormone (proBNP), which is cleaved into two major derivatives—a bioactive c-terminal peptide (BNP) and an inert amino-terminal peptide (NT-proBNP). All three are found in the circulation and commercial assays with good analytical sensitivity, reproducibility, and established normal ranges are widely available.6 The rationale for their use as biomarkers is clear based on their role in cardiorenal homeostasis. They are synthesized and secreted predominantly by cardiomyocytes and the major stimulus for secretion is left ventricular wall stress related to volume or pressure overload. Levels of all three peptides are increased in heart failure and have been shown to reflect the severity of cardiac dysfunction and symptomatic status.6 More importantly, there is consistent evidence across multiple studies that these peptides improve the diagnosis of acute heart failure6,7 and they are amongst the most powerful markers of risk across the spectrum of heart failure stages.8 These peptides may even have a role in guiding heart failure therapy.9,10 Despite their value as diagnostic and prognostic biomarkers, risk stratification with models that include BNP/NT-proBNP is still imperfect.8 The desire for improved risk stratification in heart failure is one factor that has driven the search for new biomarkers. In a study published in the current issue of this journal, Masson et al.11 assessed the prognostic value of four new biomarkers in stable chronic heart failure. Using recently developed assays they measured plasma levels of the precursor peptides for four vasoactive peptides with important roles in cardiovascular and renal homeostasis. Adrenomedullin (ADM) is a powerful vasodilator peptide that is secreted predominantly from vascular endothelial cells. Earlier studies have suggested that plasma ADM levels are increased in heart failure and after myocardial infarction and may be associated with increased risk.12 Endothelin-1 (ET-1) is a potent vasoconstrictor with elevated plasma levels in heart failure. Arginine vasopressin (AVP) levels are also known to be elevated in heart failure and its actions in promoting salt and water retention and as a vasoconstrictor are well documented. Atrial natriuretic peptide (ANP) is a sister peptide to the BNP, but to date has performed less strongly as a diagnostic and prognostic biomarker. Previous studies evaluated plasma levels and prognostic value of circulating vasoactive ADM, ET-1, AVP, and ANP peptides. In contrast, the immunoassays tested in the current study; Mid Regional (MR) proANP, MR-proADM, Carboxy Terminal (CT) pro-ET-1, and CT-proAVP—also known as copeptin, measure inactive peptides that enter the circulation after excision of the respective bioactive peptides from their larger pro-peptide precursors (Figure Figure 1). All four assays have been developed by the company BRAHMS. The MR-proANP assay essentially measures NT-proANP—the ANP analogue of the more familiar NT-proBNP. Both ANP and BNP are initially synthesized as the larger precursor peptides proANP and proBNP. After cleavage of ANP and BNP from the carboxy-terminal (c-terminal) end of these precursors, the aminoterminal fragments NT-proANP and NT-proBNP enter the circulation where they accumulate, most likely due to lower clearance rates. The other precursors are cleaved in a similar manner, but give rise to differing numbers of active and inactive peptide products. The ET-1 precursor, proET-1, undergoes multiple cleavage giving rise to big ET-1, plus three additional peptides including CT-ET-1 from its c-terminal end, although proAVP produces two functional peptides (AVP and Neurophysin II) plus the inactive c-terminal fragment CT-proAVP (copeptin). Cleavage of proADM yields the two active peptides, ADM and pro-adrenomedullin n-terminal 20 peptide (PAMP), plus MR-proADM, and another peptide. All of these peptides enter the circulation but why measure them when analytical methods already exist for the biologically active products that should, intuitively, better reflect biological function and outcomes? There are good analytical reasons for choosing these alternative fragments, including low, difficult to measure, circulating concentrations of the bioactive peptides (AVP, ET-1), technical problems such as low and variable recovery of the active peptides from plasma (ADM) and lack of stability in blood following collection. In contrast, many of the ‘inactive’ products circulate at concentrations several fold higher than the ‘active’ peptides—due in part to lower clearance rates—and present fewer analytical problems. The BRAHMS assays above have been successful from this analytical point of view. The direct two-site format using relatively low sample volumes represents a major improvement on analytical methods, particularly those for ET-1 that has very low circulating levels, and for ADM that adheres strongly to surfaces. CT-proET-1 (44 pmol/L) and MR-proADM (330 pmol/L) concentrations in healthy subjects are much higher than those of ET-1 (about 1 pmol/L) and ADM (about 5 pmol/L) allowing the development of more robust two-site assays. In contrast, the similar circulating concentrations of CT-proAVP (copeptin) and AVP (about 4 pmol/L), combined with the use of small sample volumes, have resulted in an assay that cannot detect CT-proAVP in the bottom 2.5% of normal samples from healthy subjects. The MR-proANP, like other NT-proANP assays, measures circulating levels that are some three- to five-fold higher than those reported for ANP. Although this approach has generated some potentially useful assays, it is less certain that they will result in better diagnostic tests. It cannot be assumed that these ‘inactive’ peptides can be used as surrogate measures of their corresponding bioactive peptides because of their origin from the same gene precursor. Differences in secretion and clearance rates of the peptides and their variation in these in different pathologies may result in better (or worse) test performances. Further, measurement of higher circulating levels with better assays and a more stable analyte do not necessarily result in better test performance. A current example is the use of NT-proBNP as an alternative test for the diagnosis of heart failure in subjects with dyspnoea. When we first identified this peptide we hoped that the higher circulating levels, some 10-fold greater than BNP, would result in better assay sensitivity and a better diagnostic test than BNP.13 However, despite NT-proBNP's higher circulating levels, ease of measurement, and stability in plasma, both BNP and NT-proBNP have similar performance in the diagnosis of heart failure.7 In the current study, Masson et al.11 have addressed the question of whether there is any value in measuring these precursor peptides to assess prognosis in heart failure. In a well characterized population of heart failure subjects they demonstrated that plasma levels for each of these precursor peptides are higher in association with increasing age, increasing symptomatic limitation and with increasing degrees of renal dysfunction. In addition, they have demonstrated that there is a gradient of increasing risk for mortality or cardiovascular hospitalization with increasing levels of each precursor peptide. The univariate gradient of risk was similar for all four peptides and did not differ from the established cardiac marker, NT-proBNP. In multivariate analysis, the highest tertile of each precursor peptide was associated with increased risk independently of clinical and echocardiographic markers. More particularly, the authors have then demonstrated that availability of MR-ANP improved the accuracy of risk stratification for adverse events, whereas MR-ADM and the established marker NT-proBNP had a more marginal effect on risk stratification. When MR-ANP was considered in addition to clinical factors, this resulted in a net reclassification of 12% of subjects, reflecting more accurate identification of subjects at higher risk who subsequently died as well as more appropriate downgrading of risk in subjects who survived. Improved risk stratification by MR-proANP proved to be incremental to any benefit from considering levels of the established biomarker NT-proBNP. The authors then confirmed the findings from several recent studies by demonstrating that a repeat biomarker measurement during follow-up further improves risk stratification. Subjects with greater increases in either MR-proANP or NT-proBNP during follow-up had a significantly higher risk of mortality or hospitalization. The findings from this study suggest that measurement of these precursor peptides improves risk stratification over models that consider clinical variables alone. The prognostic value of these peptides was similar to the established marker NT-proBNP. In the case of MR-proANP, this precursor peptide provided risk stratification that was incremental to NT-proBNP. Why should this be, given the similar pathophysiological patterns for atrial and BNP? Analytical differences in measurement or assay performance are unlikely to explain this. Instead, it is possible that biological differences related to the site and stimulus for secretion and the pattern of synthesis and release may be relevant.6,14 The findings from this study suggest potential clinical benefit from risk stratification using these new assays. Should we therefore be measuring these precursor peptides routinely in heart failure subjects? The scene is set for further studies to validate the findings in this cohort and to confirm whether there is any incremental benefit to measurement of these peptides in addition to established biomarkers. Conflict of interest: R.W.T. has received honoraria (modest) from Roche Diagnostics.