Heart failure (HF) is a major public health burden and is often a clinically silent process, with progressive cardiac remodeling that eventually leads to symptomatic presentation late in the course of disease progression. The severity and prognosis of HF vary substantially, ranging from mild disease that is easily managed with neurohormonal blockade to advanced illness requiring mechanical support or heart transplantation.1 Physicians use biomarkers as additional tools to aid clinical diagnosis and treatment and to identify high-risk subjects.2

The progression of HF is complex and is driven by multiple biological processes, including inflammation, oxidative stress, neurohormonal activation, vascular remodeling, myocyte injury, and renal impairment.3 Current guidelines recommend testing B-type natriuretic peptide (BNP) or N-terminal pro-BNP (NT-proBNP).4

The progression of HF is associated with left ventricular (LV) remodeling, which manifests as gradual increases in LV end-diastolic and end-systolic volumes, wall thinning, and a change in chamber geometry to a more spherical, less elliptical shape, with a continuous decrease in LV ejection fraction (LVEF).5 When ventricular remodeling is advanced, it is self-sustaining, leading to disease progression, regardless of neurohormonal status.

However, in some situations, there may be LV reverse remodeling (LVRR), characterized by decreases in LV dimensions, normalization of LV shape and improvement of systolic function.

In this work, we set out to find associations between biomarkers and clinical severity and echocardiographic parameters. We also sought predictors of LVRR after optimal pharmacological therapy. We used available biomarkers that reflect diverse biological pathways in HF: adrenaline, noradrenaline, plasma renin, aldosterone and BNP (neurohormonal activation), high-sensitivity C-reactive protein (hs-CRP), cancer antigen CA-125 (inflammation), uric acid and lipoprotein(a) [Lp(a)] (oxidative stress), creatinine and cystatin C (renal function), and 25-hydroxyvitamin D [25(OH)D] (extracellular remodeling), in a cohort of HF outpatients.

We included consecutive adult patients with dilated cardiomyopathy (DCM) followed in an HF outpatient clinic in a tertiary care center, diagnosed less than 24 months previously and with two values of LVEF of <40% more than one year apart. This is a biomarker substudy of a previously published work6 by the same authors. In this new cohort we excluded patients with atrial fibrillation, in order to avoid measurement errors in the assessment of LVEF. We also excluded DCM patients with secondary etiologies including ischemic, valvular, inflammatory and toxic cardiomyopathy.

At baseline, patients underwent clinical assessment of symptoms and signs of HF. Peripheral congestion (ankle edema) and pulmonary congestion were assessed, the latter by lung auscultation and chest X-ray. All patients underwent an electrocardiogram (ECG) and transthoracic echocardiogram and blood laboratory tests. Patients were managed according to current clinical practice guidelines4 and efforts were made to reach the recommended target doses for all therapies. During follow-up, periodic clinical assessment, laboratory tests and echocardiography were performed at three and six months. All patients gave their written informed consent. This study was performed in accordance with the recommendations of the Declaration of Helsinki7 and approved by the hospital's ethics committee.

We studied 50 patients, 28 men (56%), aged 59±10 years and followed for 39±22 months. The majority of patients (62%) were in New York Heart Association (NYHA) class II. At baseline, mean LVEF was 25.4±9.8%, LVEDD was 62.4±7.4 mm, LVEDD/body surface area (BSA) was 34.2±4.5 mm/m2 and grade >II/IV mitral regurgitation was present in 34% of patients. Table 1 details the patients’ baseline characteristics.

On the ECG, 66% had left bundle branch block (LBBB), 22% had LV hypertrophy criteria and all were in sinus rhythm.

At the beginning of the study, 76% of patients were being treated with angiotensin-converting enzyme inhibitors (ACEIs)/angiotensin receptor blockers (ARBs), 62% with beta-blockers and 14% with aldosterone receptor antagonists (Table 1).

At the end of follow-up, 94% of the patients were being treated with ACEIs/ARBs, 98% with beta-blockers and 60% with aldosterone antagonists. Optimal recommended doses of ACEIs/ARBs were reached in 42% and optimal doses of beta-blockers were reached in 48%. Only 4% needed urgent transplantation or died (two deaths), 22% were hospitalized for worsening HF and 40% received an implantable cardioverter-defibrillator (ICD). At the end of follow-up only 15 patients had LBBB and LVEF ≤35%. Four (8%) received a cardiac resynchronization therapy device (CRT-D); the others did not receive a CRT-D because of NYHA class I, technical problems, patient refusal, transient LVEF recovery, presence of LV diverticulum or old age.

One of the most important areas of biomarker research is the role of biomarker profiling to better characterize the phenotype of patients who might best respond to therapeutic interventions, whether drug or device therapies.10 The newly-defined HF with recovered LVEF represents a distinct HF phenotype with biochemical properties and natural history that differ from the traditional HF population.11 Predictors of LVRR probably distinguish patients in whom LVEF may recover with medical therapy only from patients who may require cardiac devices or more aggressive strategies, including heart transplantation. Patients who have recovered LVEF do not, in theory, have indications for ICD or CRT therapy, thus complicating the timing of implantation of these devices.

Data now suggest that biomarkers may also be useful to predict or monitor LVRR. A new clinical score, the ST2-R2 score, contains five clinical variables (non-ischemic etiology, absence of LBBB, HF duration, LVEF and beta-blocker treatment) and a remodeling biomarker, soluble toll-like receptor 2 (ST2). A significant relationship was observed between ST2-R2 scores and changes in LVEF, indexed LV sizes, and percentage reduction in LV end-systolic volume index; a similar trend was observed with diastolic parameters.11

Our study set out to find predictors of LVRR. In our population LVRR occurred in one third of DCM patients, who were younger and had better renal function and smaller LVEDD. These results are consistent with other studies. In a large study, LVRR was found in 89 of 242 idiopathic DCM patients (37%) and baseline predictors were higher systolic blood pressure and the absence of LBBB.12 Binkley et al. showed that patients who recovered LV function were younger, had higher systolic blood pressure, lower serum creatinine level, shorter QRS interval, a lower prevalence of diabetes and history of hypertension, were more frequently female and had a lower prevalence of ischemic cardiomyopathy.13

In our population LVRR was also associated with lower BNP at the end of follow-up and with favorable outcome and reduced rate of cardiac events. In a recent study with elderly HF patients, intensified medical therapy led to improvement in LVEF and to reverse remodeling. NT-proBNP guided therapy was associated with a greater improvement in LVEF than symptom-guided therapy in both patients aged 60 to 74 and in those aged ≥75 years.14

We found no association between BNP, 25(OH)D, CA-125, hs-CRP or Lp(a) and LV reverse remodeling. As expected, patients with poorer functional class, pulmonary congestion, or ankle edema had higher BNP levels. Atrial natriuretic peptide and BNP are produced in response to myocardial stretch due to pressure or volume overload.15 In the ICON study, increasing severity of HF, as measured by NYHA functional class16 and symptom severity, correlated directly with increasing BNP concentrations.17 In our study, BNP also correlated with LV dimensions, LV volumes, LVEF, PASP and measures of diastolic dysfunction (E/e′ and e′ velocities). Since BNP is primarily synthesized by cardiomyocytes, it is not surprising that the highest levels are secreted by the LV. Studies have demonstrated that BNP and NT-proBNP levels correlate positively with LV dimensions, volumes, and mass and are inversely related to LVEF; the strongest correlations have been reported for BNP with LV diastolic wall stress consistent with stretch-mediated BNP secretion.18,19 BNP levels increase with greater severity of overall diastolic dysfunction, correlating with indices of filling pressure as well as with indices of compliance and myocardial relaxation.20,21

Serum CA-125, a high molecular weight glycoprotein, is a tumor marker widely used in patients with ovarian cancer.22 Recently, increased serum CA-125 levels, in parallel with catecholamines and natriuretic peptides, have also been documented in patients with HF.23 However, little is known about the biologic role of this substance: whether it simply reflects increased activation of the cytokine pathway or is an active substance actually responsible for myocardial and/or peripheral dysfunction. In our population, CA-125 was not a predictor of LVRR, but did predict more severe presentation as shown by worse functional class and pulmonary and peripheral congestion. It also correlated with measures of diastolic dysfunction (LA volume, E/A ratio and PASP), as well as with BNP levels, hs-CRP and uric acid, suggesting a potential pathogenic link between inflammatory activation and this marker of systemic congestion. Kouris et al. showed that patients in NYHA classes III and IV had significantly higher mean CA-125 values than patients in class II; serum CA125 levels correlated weakly with PASP and renal function.24 A study analyzing CA-125 levels and LV dysfunction in patients on hemodialysis showed they were positively correlated with pro-BNP and C-reactive protein (CRP) levels, as well as with LVEDD, LVESD and LV mass index.25

CRP is a plasma protein that participates in the systemic response to inflammation, an important mechanism in the progression of HF. A landmark review showed that CRP concentrations were higher than normal in 70% of HF patients, and were directly related to severity of HF.26 In a recent study of patients with acute HF, both hs-CRP and NT-proBNP were independent predictors of 12-month mortality.27 In our patients, hs-CRP was associated with clinical severity and marginally correlated with measures of systolic dysfunction (LVEF and LVEDV/BSA). In a previous study, CRP levels increased in parallel with NYHA class and were also related to higher readmission and mortality rates.28

Functional vitamin D receptors are present in cardiac cells and their activation affects gene expression, proliferation and contraction of cardiomyocytes. Vitamin D may thus contribute to the development of cardiac hypertrophy and fibrosis.29 In our population, 25(OH)D was not a predictor of LVRR and was only correlated with PASP and E/A. In a study of subjects with HF, LVEDD, LVESD, LVEDV and LVESV were significantly larger and fractional shortening was lower in patients with 25(OH)D <25 nmol/l than with 25(OH)D ≥25 nmol/l (p<0.05); log values of 25(OH)D were negatively correlated with LVEDD and LVEDV (r=-0.28; p<0.05).29

In a recent study of ambulatory patients with chronic HF, Ky et al.30 tested the hypothesis that a group of seven biomarkers (BNP, soluble fms-like tyrosine kinase receptor, hs-CRP, ST2, cardiac troponin I, uric acid and creatinine) could be combined into a multimarker score that would predict risk of adverse outcome, defined as death, cardiac transplantation or ventricular assist device placement. Patients in the highest tertile of the multimarker score had a 13.7-fold increased risk of adverse outcomes compared with the lowest tertile.

The approximate cost per patient of the biomarkers that we used is currently around €100. In-hospital care is responsible for ∼60% of HF-related costs and median hospitalization cost is €9475,4 but in many cases costs are much higher due to frequent comorbidities and need for intensive care. If the use of biomarkers succeeds in preventing just one hospitalization, this could be a highly cost-effective strategy.

In our center, high-sensitivity troponin I has only been available since last year, so measurements at initial assessment are not available. However, plasma samples have been frozen in liquid nitrogen and a study with emerging biomarkers is ongoing. Due to the small population it was not possible to perform a multimarker score.

In our population, LVRR occurred in one third of DCM patients, especially in younger patients, with better renal function and smaller LVEDD. We can postulate that patients with long-standing disease and larger LV diameters may not recover LV function. CA125, BNP and hs-CRP were predictors of clinical severity but not of reverse remodeling. BNP correlated with parameters of systolic and diastolic dysfunction, while CA-125 correlated with measures of diastolic dysfunction.

The authors have no conflicts of interest to declare.