Biomarkers and prognostication in heart failure with reduced and preserved ejection fraction: similar but different?

This article refers to ‘The prognostic value of highly sensitive cardiac troponin assays for adverse events in men and women with stable heart failure and a preserved vs. reduced ejection fraction’ by A. Gohar et al., published in this issue on pages 1638–1647. I first developed an interest in heart failure in the early 1990's; at that time, management of the condition looked not at all like it does now. Our patients received diuretics, and perhaps digoxin. Evidence of the potential for vasodilator therapy was just emerging, and the concept of therapeutic inhibition of the renin–angiotensin–aldosterone system (RAAS) in patients with heart failure was novel. Well, we are all very aware of the advances which have been made in the intervening 25 years, and some of us have been around long enough to have seen all of them emerge and take their place as standard of care. We are now able to manage the heart failure syndrome much more effectively than was the case in the past; RAAS inhibition remains central to the management of the patient, with even the latest addition to our armamentarium, the novel angiotensin receptor blocker/neutral endopeptidase inhibitor (ARNi; sacubitril/valsartan)1 being, in part, a RAAS inhibitor. Add our understanding of the roles of heart rate control and implantable devices to this, and the treatment panorama presents a very different vista to that I observed as a young researcher. Of course all of these developments are relevant to patients with heart failure with reduced ejection fraction (HFrEF) and over the same period during which we have seen paradigm shifts in the management of this condition, we have seen two other major changes in the heart failure landscape. The first is that we have recognised that heart failure is more than a single syndrome, and that a significant proportion of our patients with the heart failure syndrome have apparently preserved left ventricular contractile function. For these patients, little has changed, at least in terms of development of effective pharmacological therapies. The second important development is our identification of elevation of circulating levels of numerous biomarkers in heart failure, some of which have utility in aiding diagnosis and management of the condition.2 The most established of these is the natriuretic peptide family, the use of which is now part of international guidelines;3 indeed, we would do well to remember that the identification of B-type natriuretic peptide (BNP), and related peptides, has led us to understand the relevance of activation of the natriuretic peptide system to heart failure and thus contributed to the development of novel therapy in the form of sacubitril/valsartan. On this background, reports of the clinical utility of biomarkers in populations including patients with both HFrEF and heart failure with preserved ejection fraction (HFpEF) are of particular interest. In the current issue of the European Journal of Heart Failure, Gohar and colleagues report the prognostic utility of cardiac troponin T (TnT) and I (TnI) measured using high sensitivity assays.4 This report contains several interesting observations. First, both troponin assays were associated with the occurrence of the composite endpoint of all-cause mortality and first rehospitalisation with heart failure. Second, the strength of these associations appeared greater in HFpEF compared to HFrEF. Third, the utility of the two high-sensitivity troponin assays appeared to differ, the greatest additional prognostic value in HFpEF was seen with TnT rather than TnI, and was also superior to N-terminal-proBNP (NT-proBNP). Finally, there appeared to be a stronger association between TnI and the risk of the composite endpoint in men compared to women with HFpEF. The study population of 1096 patients was drawn from a multi-ethnic population attending six hospitals in Singapore as part of a national cohort study.5 The diagnosis of heart failure was made according to European Society of Cardiology criteria, and the majority of patients (67.3%) were recruited during an inpatient episode for heart failure, with biomarker sampling taking place ‘after in-hospital stabilisation just prior to discharge’, with the remainder recruited in the outpatient setting within 6 months of an acute decompensation. The population demographics were as might be expected: approximately 75% of patients had HFrEF and a similar proportion were male, and patients with HFpEF were older and more likely to be female compared to those with HFrEF (52% vs. 16%). Circulating levels of TnT, TnI and NT-proBNP were each higher in HFrEF than in HFpEF; in spite of this, the association with adverse prognosis was stronger for both TnT and TnI in the latter group. A recent report observed an association between admission levels of TnT or TnI and adverse prognosis in patients hospitalised with HFpEF,6 while a report from the VERITAS study7 did not see this association. Differences in study populations and the assays in use may explain these differing observations. In this context, the current report from Singapore makes interesting reading. It is the first to compare the utility of circulating troponin levels in HFrEF and HFpEF, and measured these biomarkers in both inpatients and outpatients. The authors claim that it is reasonable to consider these two groups together, as biomarker sampling in the inpatients was conducted after stabilisation and predischarge. To their credit the authors conducted sensitivity analysis and report that there was no interaction between either troponin assay and the location of recruitment in the HFrEF population; this analysis was not conducted in the HFpEF population in the context of small numbers. We might consider that the patient about to be discharged from an acute heart failure episode is very different to the patient some months later and it would be interesting to see the influence of recruitment location on the association between troponin and prognosis in HFpEF. However, this was made difficult by small numbers and indeed this is a potential criticism of the report as a whole, which included a relatively small population. Moreover, the influence of prescribed medication, and changes between discharge and clinic follow-up, on biomarker concentrations may be relevant, in particular to the apparent lack of prognostic value of NT-proBNP. However, to counterbalance the criticism of small numbers in the study, the event rate was high, with 460 (of 1096) patients experiencing the endpoint in a median follow-up period of just over 1 year. These were clearly high-risk patients and the observations should be considered in this context. What of the observed associations themselves, and can we consider the observed differences between HFrEF and HFpEF, and between men and women, to be real? The authors note that age was the only variable to associate with TnI in all groups studied, and with TnT only in HFpEF. However, it is clear from Table 2 that the absolute strength of association of each troponin with age is very similar across all ejection fraction groups, with numbers likely to influence statistical strength. Further to this, and in keeping with previous reports,6 it appears from Table 2 that renal function shows consistent strong association with both TnT and TnI; indeed, the authors present renal function in terms of creatinine rather than glomerular filtration rate, which may have influenced the observed associations between troponin and outcomes. Notwithstanding these points, the observations reported are of potential clinical relevance. While we are all aware of the phenomenon of TnT or TnI elevation in patients admitted with decompensated heart failure, the relevance of this has hitherto been unclear. We are now presented with evidence that this is relevant beyond the acute admission, and indeed the authors' report of the superiority of troponin to NT-proBNP is also worthy of note. The strength of association with outcome for TnT [hazard ratio (HR) 1.48, 95% confidence interval (CI) 1.33–1.64] and TnI (HR 1.36, 95% CI 1.24–1.45) was very similar in the whole population but much stronger in HFpEF for both TnT and TnI. The observation that, for TnI, the 95% CIs for prediction of outcome in HFpEF (HR 2.32, 95% CI 1.60–3.36) do not overlap with those in HFrEF (HR 1.29, 95% CI 1.16–1.42), with similar observations for TnT, perhaps lends credence to the apparent difference. However, for both TnT and TnI, the CIs around the HR are much wider for HFpEF compared to HFrEF. This raises questions as to the reproducibility of the observed associations, and their clinical utility. As to the apparent difference between men and women, with the association with outcome for TnI being evident only in men, the HR in women is very close to statistical significance (1.35, 95% CI 0.94–1.93, P = 0.10) and this observation should, for the moment, be viewed with caution. As the authors point out, the mechanism by which decompensated heart failure leads to troponin elevation is unknown. However, the phenomenon is real, and as indicated in this, and other6 reports, may be of clinical relevance. A biomarker with greater apparent clinical utility in HFpEF compared to HFrEF is of particular interest. Investigation of the association between biomarkers and outcomes in HFpEF is likely to shed light on the nature of this condition and may lead to development of therapeutic interventions. It should not be difficult to examine the observations in the current report in additional cohorts. Conflict of interest: none declared.