Age and Biomarkers in Heart Failure: Challenging the Current Model to Select Patients for Clinical Trials

This article refers to ‘Circulating levels and prognostic value of soluble ST2 in heart failure are less influenced by age than N-terminal pro-B-type natriuretic peptide and high-sensitivity troponin T’ by A. Aimo et al., published in this issue on pages 2078–2088.. The field of oncology is a great example where the use of biomarkers has revolutionized treatment approaches for decades. Studies using biomarkers are also likely to help us move forward in the care of patients with heart failure (HF), as they help improve our understanding of the mechanisms that determine prognosis in HF as well the mechanisms of action of HF therapies. However, the methodology must be standardized and collaboration within the research community is warranted, with the objective of increasing our knowledge on how to make the best use of biomarkers. B-type natriuretic peptide (BNP) and its amino-terminal fragment, NT-proBNP, are derived from a single polypeptide synthesized within the ventricles in response to elevation in filling pressures and wall stress.1 Amongst patients with HF and a reduced ejection fraction (HFrEF), natriuretic peptide (NP) levels are increased proportionally to disease severity, and modulation of NP levels with pharmacologic therapy over time correlates with ventricular remodelling and HF progression.2, 3 Elevated NP levels are powerful predictors of mortality and cardiovascular (CV) events,1 and are thus increasingly used as part of the entry criteria in HF clinical trials,4-7 as well as in clinical practice. Several factors have been invoked to explain the pathophysiology of cardiac troponin release in HF, including subendocardial ischaemia and myocyte necrosis, cardiomyocyte damage from inflammatory cytokines or oxidative stress, apoptosis, and leakage of troponin from the cytosolic pool due to increased membrane permeability.8 The degree of high-sensitivity (hs) troponin elevation is a powerful predictor of mortality and CV events in both ambulatory and acutely decompensated9 patients with chronic HFrEF,10 even after adjustment for traditional risk predictors including NPs. Recent studies have also demonstrated the prognostic role of hs-troponin I in HF with a preserved (HFpEF) or mid-range ejection fraction (HFmrEF).11, 12 Soluble ST2 (sST2) is a secreted decoy receptor that disrupts the binding of interleukin-33 with the full-length ST2 receptor, promoting cardiac hypertrophy, fibrosis, and ventricular dysfunction.13 In patients with HFrEF, serum levels of sST2 have been independently associated with mortality and disease progression and provide incremental prognostic value over NT-proBNP.14 Biomarkers are only one aspect of risk assessment and many important clinical covariables are used in daily practice, either in isolation or in composite risk scores. The Meta-Analysis Global Group in Chronic Heart Failure (MAGGIC) score15 has now become part of a standard of prognostic evaluation for patients with HF in many institutions. The score was derived using multivariable piecewise Poisson regression methods with stepwise variable selection, which identified 13 highly significant independent predictors of mortality. Age was the most powerful predictor of survival across all ejection fraction HF groups. Considering that patients with HF in the community are increasingly older and that with age the number of comorbidities increases, it is highly relevant to determine if HF therapies are as effective and as safe in elderly patients with HF as in the younger ones.16 Whether the additional prognostic value of biomarkers remains across the age spectrum of patients with HF is an important question, relevant to both clinical and research contexts. In this issue of the Journal, Aimo and colleagues17 assessed the influence of age on the circulating levels and the added prognostic value of the three most studied biomarkers in HF, utilizing data from an international cohort including stable chronic HF patients with a majority of HFrEF (83% of HFrEF while HFmrEF and HFpEF were each 8%). Patients with available data on age, left ventricular ejection fraction (LVEF), hs-troponin T (hs-TnT), NT-proBNP and sST2 were selected, hence 5301 patients were included in this report. The assays used to measure the biomarkers correspond to the current standards in clinical application as well as in research.4-7 The exact timing of sample collection is not reported and likely varied across the included studies but NT-proBNP and hs-TnT were assayed during each of the original studies, while sST2 was measured on EDTA plasma samples stored at −20°C. Samples were collected during an outpatient visit, in a ‘clinically stable’ condition, without changes in therapy for ≥1 month. The median biomarker levels were: NT-proBNP 1564 ng/L, hs-TnT 21 ng/L, and sST2 29 ng/mL; levels similar to those observed in recent HFrEF clinical trials.14 The authors selected four age categories: <60 years (n = 1332, 25%), 60–69 years (n = 1628, 31%), 70–79 years (n = 1662, 31%), and ≥ 80 years (n = 679, 13%). Patients had a median age of 66 years, 75% were men, median LVEF was 28%, and 64% had ischaemic aetiology of HF. Such patient characteristics are typical of the HFrEF trials from which the data are derived and the proportion of women included also reflects the strong representation of ischaemic aetiology. Women may be underrepresented in HFrEF clinical trials in comparison to their contribution to registries.18 This age and sex distribution likely does not represent the current portrait of patients attending HF clinics in most West European or North American countries. As the authors mention,17 registries may be best suited to test these findings in a closer to ‘real-life’ setting (with the caveats of registries). Still, 13% of patients were 80 years old or older. Of note, all studies included were published before the availability of sacubitril/valsartan and before the recently demonstrated outcome benefits of dapagliflozin in HFrEF.7 The use of mineralocorticoid receptor antagonists was also relatively low in comparison to recent clinical trials. The applicability of the study results for prediction of prognosis in todays' HF patients is therefore questionable. Age independently predicted NT-proBNP and hs-TnT (both P < 0.001), but not sST2 (P = 0.849). The authors identified NT-proBNP and hs-TnT cut-offs for prediction of 1-year and 5-year all-cause and CV mortality and 1- to 12-month HF hospitalization, which increased with age. However, the identified sST2 cut-offs did not increase with age.17 In the 2002 participants of the PARADIGM-HF trial who contributed to the sST2 analyses, the independent predictors of higher baseline sST2 levels were higher NT-proBNP, male sex, a history of atrial fibrillation, New York Heart Association class III/IV, diabetes, antiarrhythmic use, LVEF and time since HF diagnosis. Age was thus not independently related to sST2 levels in that study.14 Table 1 in the publication by Aimo et al.17 illustrates that absolute sST2 levels increase with age but as shown in Table 2, age is not an independent predictor of sST2 levels (P = 0.065). Furthermore, the relations between sST2 and outcomes were linear in the PARADIGM-HF study; consistent with no major differences in thresholds identified across age-groups by Aimo et al. for all-cause death, CV death, or HF hospitalization (online supplementary Table 3). Aimo et al. clearly demonstrate the additional contribution of each of the three biomarkers in terms of prognostication, on top of a multivariable model that included demographics and comorbidities associated with adverse outcomes in previous HF trials (Table 4, Model 3).17 It would have been interesting to see the added prognostic value of the biomarkers on top of internationally recognized prognostic models (such as the MAGGIC score), but most interesting is the ability of these three biomarkers to predict early (1, 3 and 6 months) rehospitalization for HF, especially in HFrEF and in the elderly (online supplementary Table 2). Predicting morbidity is especially important in the HF population, where outcomes beyond mortality are very relevant from the patient's perspective. The findings by Aimo et al.17 highlight the imminent need to revise the current entry criteria for HF clinical trials (Figure 1). Recent and current HFrEF clinical trials require participants to have elevated NT-proBNP levels, with different thresholds used for atrial fibrillation and sinus rhythm. Based on the data provided here, age-specific thresholds should be considered for both NT-proBNP and hs-TnT. As seen here, atrial fibrillation is also an independent predictor of the levels of hs-TnT and sST2. Should hs-TnT and sST2 levels be part of entry criteria in future HF trials? They do provide very interesting additional information on future HF hospitalizations, an important outcome both for resource utilization and from a patient's perspective. Should a risk score be used in addition to biomarker levels to select patients for clinical trials assessing efficacy and safety of HF therapies? This type of approach would certainly identify patients at higher risk and could eventually be applicable to clinical reality. The three biomarkers studied by Aimo et al.17 have repeatedly been shown to provide additional prognostic value1 in HFrEF and represent three different (although related) pathophysiologic mechanisms.1, 14, 19 These may collectively help further refine or tailor HFrEF therapy, if we knew which ones (and for some at which level) best identify responders to a given therapy, considering mechanisms of action and predictors of adverse events related to treatment in such models (Figure 1). More treatments will soon be available for patients with HFrEF and it is unlikely that all patients will be responders (with few adverse events) to all proposed therapies. The potential role of machine learning is certainly attractive in the new era of HF therapy.20 Conflict of interest: C.H. reports clinical trial participation with AstraZeneca, American Regent, Boehringer Ingelheim, Merck, Novartis and Pfizer. L.M. reports a Heart and Stroke Foundation of Canada (HSFC) grant for the Bio-AIMI-HF study (co-applicant); speakers and consulting fees from AstraZeneca, Bayer, Servier and Novartis. E.O. reports a HSFC grant for the Bio-AIMI-HF study (co-applicant); clinical trial and steering committee participation for Amgen and American Regent; clinical trial, speakers and consulting fees paid to her or her institution from AstraZeneca, Bayer, Boehringer Ingelheim, Merck, Novartis and Pfizer.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.