The HOMAGE clinical trial was a multicenter, prospective, randomized, open-label, blinded endpoint (PROBE) trial that investigated the effect of spironolactone on serum markers of collagen metabolism, as well as cardiac structure and function in people at risk of developing HF (ClinicalTrials.gov identifier: NCT02556450). Informed consent was obtained from all participants and the trial was approved by the relevant ethics committees and regulatory bodies.18,19 Detailed methods, follow-up and the primary outcome measures have been published.18 Due to slow recruitment and fixed funding duration, the planned follow-up at nine months was shortened to between three and eight months in 161 patients (32%).19

Briefly, people aged >60 years with established coronary artery disease or at least two of the following risk factors for cardiovascular disease (type 2 diabetes, hypertension, microalbuminuria or an abnormal electrocardiogram), as well as elevated plasma natriuretic peptides (N-terminal brain natriuretic peptide [NT-proBNP] between 125 and 1000 ng/L or B-type natriuretic peptide [BNP] between 35 and 280 pg/mL), were enrolled at hospital clinical research clinics. Patients with left ventricular ejection fraction (LVEF) <45%, a diagnosis of HF, treated with loop diuretics or with atrial fibrillation were excluded. In total, 527 participants were enrolled and randomly allocated to either (1) usual care alone (“control group”) without treatment with an MRA or (2) treatment with spironolactone 25 mg/day (up-titrated to 50 mg/day, if tolerated) in addition to usual care. Information was collected at baseline and at the end of the trial (nine months for most participants), including clinical information, health status questionnaires, blood samples, an electrocardiogram and an echocardiogram. Serum biomarkers measured in the HOMAGE clinical trial included procollagen type III N-terminal pro-peptide (PIIINP), procollagen type I C-terminal pro-peptide (PICP), collagen type-1 C-terminal telopeptide (CITP), galectin-3 and matrix metalloproteinase-1 (MMP-1).18

The primary outcome was the interaction between changes in serum PIIINP concentrations from baseline to final visit and baseline serum galectin-3 concentrations. Estimated sample size required 800 participants to be powered to detect an interaction between PIIINP and median galectin-3. During the trial period, 877 patients consented to participate in the trial, of whom 561 were eligible and 527 were randomized.19

Data regarding health status was collected during the trial period as a non-registered outcome, and a post-hoc secondary analysis of the data was performed. Health status data was available for 519 out of 527 participants (98%) and was assessed using the EQ-5D-3L, EQ-VAS and HOMAGE symptom questionnaire (Supplementary Table 1) at baseline and at the final visit. As the HOMAGE clinical trial included patients without symptomatic HF, the presence of symptoms was also characterized separately from the health status assessments.

The EQ-5D-3L measures health status on five dimensions (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression), stratified in three levels of severity, namely 1 – no impact, 2 – moderate impact, and 3 – severe impact.20 For the EQ-VAS, patients were asked to rate their perception of their health state between 100 (best imaginable health state) and zero (worst imaginable health state) on a VAS.21 The HOMAGE symptom questionnaire (as a supplement to the EQ instruments) assesses self-reported patient perception regarding breathlessness on moderate exertion (the response to which was also assigned to the corresponding New York Heart Association [NYHA] class), tiredness, and ankle edema, using a numerical scale from 0 to 9, with anchor points of 0, standing for “no problems”, 5 – “some problems”, 9 – “severe problems”, and also uses an overall QoL and health rating questionnaire where 1 stands for “very good”, 5 – “average” and 9 – “very bad” (Supplementary Table 1). As NYHA functional class was evaluated by a physician, this value was considered separately from the patient-reported health status parameters.

For each patient, the baseline EQ-5D-3L 5-digit score was used to derive their corresponding EQ index value, considering the European-based VAS value set.22,23 For instance, should a patient report no problems (i.e., level 1 in all dimensions), the corresponding EQ index value would signify an optimal health status (in this case, EQ index value=1).

Study participants were not categorized according to their randomization groups for this analysis. Instead, patients were stratified into three groups according to their EQ European index and categorized as: (1) “optimal health status” if they had an index value of 1 (218 patients, 42%); (2) “intermediate health status” if they had an index value below 1 but higher than or equal to 0.71 (cut-off value for the lower-most quartile Q1; 34%); or (3) “poor health status” if they had an EQ index value lower than 0.71 (24%). Stratification of the EQ index value according to these values, as opposed to the use of equal tertiles, was necessary as a significant number of patients (42%) were considered to have optimal health status (i.e., EQ European index=1). Worsening health status was measured as a decline in the EQ European index from baseline to end of trial.

Baseline characteristics were summarized as counts and percentages for categorical variables and median and IQR (interquartile range; from second to fourth quartile) for continuous variables. The p trend between each characteristic and EQ European index was calculated by fitting a linear regression with the EQ European index value as the dependent variable and corresponding characteristics as independent variables. Statistical significance was considered if p <0.05.

Missing values were infrequent and imputed with the median of all values for continuous variables and mode for categorical variables. The absolute frequencies and relative percentages of missing values are listed in Supplementary Table 2.

The association between baseline clinical characteristics, plasma concentrations of NT-pro-BNP, serum markers of fibrosis and echocardiographic variables with declining health status (as assessed by EQ index and VAS), was evaluated through an analysis of covariance (ANCOVA). For this, a linear regression model was fitted with either EQ index value or VAS change (calculated as the respective value at baseline subtracted from that at the end of the trial) as the dependent variable, and each clinical variable plus the EQ index or VAS value at baseline as independent variables. The change in each QoL dimension (rated as none, moderate or severe) and in responses to the HOMAGE symptom questionnaires (with scores ranging from 0 – meaning no problems, 5 – some problems and 9 – severe problems) from baseline to the end of the trial was defined as a binary variable, i.e., “worse” vs. “the same or improved”. The association between clinical characteristics and worsening scores in each dimension was assessed by logistic regression. Variables found to affect the worsening of each QoL parameter, along with their β coefficient (95% CI), z-score and p-value were obtained.

The effect of treatment with spironolactone on the worsening of each QoL measure was also assessed through logistic regression, adjusted to the respective QoL baseline score and treatment arm. Odds ratios (OR), z-scores and p-values for treatment with spironolactone, compared to standard of care were obtained.

Statistical analysis was performed using R statistical software, version 4.1.2.24,25

The baseline characteristics of the 519 patients included in this secondary analysis are described in Table 1. The median EQ European index value in this population was 0.78 (IQR 0.71–1.00). Of the 519 patients, 218 (42%) had an EQ index equal to 1 (optimal health status), 179 (34%) had an index of 0.71–0.99 (intermediate health status) and 122 (24%) had an index below 0.71 (poor health status).

Compared to patients with optimal health status as measured by the EQ index, those with a poor or intermediate health status were more likely to be women, more likely to have exertional breathlessness (self-reported NYHA class ≥II), had a higher body mass index (BMI) as well as a higher waist circumference, were more likely to use two or more pillows while sleeping and were more likely to live alone (Table 1).

When compared to standard of care, treatment with spironolactone did not change health status when considering the EQ European index (OR 1.01, 95% CI 0.68–1.52, p=0.93), the EQ-VAS (OR 1.24, 95% CI 0.84–1.84, p=0.27), patient-reported QoL (OR 0.99, 95% CI 0.66–1.49, p=0.95) or any of the sub-measures evaluated (Table 2).

Despite the exclusion of people with a formal HF diagnosis in the HOMAGE trial, a high proportion of participants had poor health status at baseline. When directly inquired, one in three participants reported frequent breathlessness with effort, and approximately 16% of patients had a self-reported NYHA functional class of greater than or equal to II. The presence of HF symptoms, when combined with the inclusion criteria of elevated natriuretic peptide levels in the HOMAGE trial, make it likely that a significant proportion of enrolled patients had clinical HF.

Health status deterioration was more common in patients with worse baseline symptoms and evidence of diastolic dysfunction. Evidence of left ventricular diastolic dysfunction is associated with a greater risk of incident HF,26 while increased LVMI and E/e′ ratio are associated with impaired quality of life and poor metabolic health (including higher BMI) in asymptomatic populations.27,28

As in several other studies,29–31 we found that self-reported NYHA functional class is a predictor of overall health status and QoL scores. In patients at increased cardiovascular risk, the presence of comorbidities such as anemia,32,33 impaired renal function,34–36 hypertension,37–39 chronic pulmonary disease, cerebrovascular disease38,40 and diabetes38,41,42 have also been associated with impaired health status, which is consistent with our results.

Breathlessness and fatigue are cardinal features of symptomatic HF but often go unrecognized by patients until symptoms are severe. As breathlessness progresses, patients may modify their lifestyle and limit activities they would have previously been able to complete without discomfort to avoid breathlessness. This may be interpreted by the patient as worsening health status but may not be captured in a simplistic breathlessness scale such as NYHA class. Furthermore, these symptoms may not even be identified by the patient, underlining the importance of a multidimensional tool for assessing health status in these patients. However, whether patients with deteriorating health status and evidence of diastolic dysfunction meet diagnostic criteria for HF with preserved ejection fraction is not clear. Further evaluation with objective measures including echocardiographic parameters and circulating natriuretic peptide levels may increase diagnostic sensitivity in patients with poorly recognized or subtle HF symptoms.43

Mineralocorticoid receptor antagonists have been shown to improve cardiac function and prognosis in patients with left ventricular dysfunction. However, results in patients with HF with preserved ejection fraction had been inconsistent44–46 until the recent publication of the Finerenone Trial to Investigate Efficacy and Safety Superior to Placebo in Patients with Heart Failure (FINEARTS-HF). This study showed that finerenone significantly reduced the composite endpoint of death from cardiovascular causes and total HF events in patients with a LVEF ≥40%.47,48 The HOMAGE trial, performed in individuals at increased risk of developing HF, found that spironolactone may influence type I collagen metabolism and was associated with favorable cardiac remodeling.18,49 However, in this population, nine months of treatment with spironolactone did not appear to change health status. These results are in line with studies in patients with mildly symptomatic HF with preserved ejection fraction, such as the Aldo-DHF trial and the Treatment of Preserved Cardiac Function Heart Failure With an Aldosterone Antagonist (TOPCAT) trial, where treatment with spironolactone did not have a lasting effect on QoL.46,50

The HOMAGE trial measured PIIINP as a marker of type III collagen synthesis; PICP and CITP as markers of type I collagen synthesis and degradation, respectively; galectin-3 was used as a marker of mineralocorticoid receptor activation; and the ratio of CITP to MMP-1 as a marker of collagen cross-linking.18 Galectin-3 is a novel biomarker of inflammation and fibrosis, especially in cardiac tissue.51,52 High serum galectin-3 concentrations are associated with impaired QoL, reduced functional status, worsening HF and all-cause mortality in patients with or without known cardiovascular disease.28,53–58 Additionally, both baseline and longitudinal increases in galectin-3 have been associated with incident HF in several populations.59–62 Some authors consider that serum galectin-3 may have a role in the risk stratification of HF patients,56 particularly those with preserved ejection fractions.63

Mechanistically, galectin-3 plays an important role in HF progression primarily through its effect on ventricular remodeling.58,64 Increased galectin-3 expression enhances fibroblast activation in the extracellular matrix, leading to increased expression of cytoskeletal proteins, increased synthesis of extracellular matrix components (such as type I collagen) and inhibition of matrix degradation.58,65

In this manner, galectin-3 levels may help identify individuals with subclinical cardiac dysfunction who are at increased risk for incident HF and progressive deterioration in health status.58 We found that higher serum PIIINP and galectin-3 levels at baseline and increased markers of collagen degradation were associated with declining health status and worsening HF symptoms, in patients not diagnosed as having HF.

Some limitations need to be considered when interpreting the results of this study. First, this study represents a secondary and exploratory analysis of an unregistered secondary outcome from a clinical trial. Furthermore, as multiple comparisons were conducted, the possibility of type I errors cannot be excluded. Accordingly, these results should be considered hypothesis-generating rather than confirmatory, as some findings may have occurred by chance. Second, the HOMAGE clinical trial could not enroll the prespecified number of patients, nor was the planned follow-up time of nine months completed for all patients. Nonetheless, most patients included in this analysis were followed for a period of approximately nine months, and no interaction was found between the effect of treatment and follow-up length (p for interaction=0.7) on worsening patients’ EQ-5D score from baseline to end of trial. It must also be emphasized that the trial was not sufficiently powered to find a difference in the primary outcome, and as such a significant difference in health status change with spironolactone, a secondary outcome, is unlikely to be found. It could be postulated that should the prespecified number of patients had been enrolled, the clinical effect size could have been meaningful, but one should also underline that larger, fully powered clinical trials did not find a significant change in QoL with spironolactone, as has been previously discussed.46,50 Furthermore, no other clinical trials have studied the effect of spironolactone on health status in patients at increased risk of developing clinical heart failure. Third, the use of a hospital-based cohort, with a predominance of older male participants and a high prevalence of coronary artery disease may limit the generalizability of our results. Nonetheless, these features are consistent with the epidemiology of populations at high cardiovascular risk and, therefore, at elevated risk for HF.13 Fourth, study participants were not categorized according to their randomization groups in this analysis, and as such it is unclear whether the original HOMAGE randomization groups were equally balanced with regard to their health status as measured in this paper. Fifth, although the EQ-5D tools used in the primary health status analysis have undergone extensive validation, the HOMAGE symptom questionnaires used to secondarily determine health status and patient-reported outcomes in this population have not yet been externally validated. Lastly, although some patients had poor health status, in the overall HOMAGE population the health status was moderate or good; thus, it would be difficult to further improve health status with spironolactone treatment.