Heart failure and co-morbidity revisited; the elephant in the room.

Abstract

This article refers to ‘Prevalence and prognostic impact of non-cardiac co-morbidities in heart failure outpatients with preserved and reduced ejection fraction: a community-based study’ by A. Iorio et al., published in this issue on pages 1257–1266. Declining mortality rates of heart failure (HF) patients over the last decades is a testament to the progress and efforts made by the HF community to change medicine through randomized trials and clinical vigilance. This progress is evident in the HF guidelines that are updated and published every few years. However, one aspect of HF management has not experienced a similar progress in scientific findings—the management of co-morbidities in HF patients. It is thoroughly documented through registries and trial data that patients with HF more often than not are burdened with cardiac (e.g. atrial fibrillation) and non-cardiac (e.g. diabetes) co-morbidities. Despite this fact, substantial gaps in our knowledge of how co-morbid burden affects HF patients exist. The European Society of Cardiology HF guidelines acknowledge the importance of co-morbidity and HF, but the evidence to suggest differences in treatment regimens between patients with a co-morbid condition ± HF is generally poor.1 Guidelines generally describe how modifications to the standard HF treatment may be warranted in case of a select few concurring co-morbid conditions (e.g. caution with angiotensin-converting enzyme inhibitors in HF patients with renal disease). But should co-morbidity and HF merely be viewed as separate entities, or is the picture more nuanced? Importantly, many HF patients have more than one co-morbid condition—in fact HF patients have on average three–four co-morbid conditions in excess of their HF.2 This makes inferences on the impact of each individual co-morbid condition and HF over-simplified, considering the vast number of possible combinations of co-morbid diseases and HF and their possible interactions. This analytical issue is not easily resolved, and most studies have chosen to view each co-morbidity separately or assign all co-morbid conditions a similar weight in their analyses, with the obvious advantage of being able to make general inferences on total co-morbid burden and outcomes. Using this methodology, Iorio and co-workers provide a welcome addition to our understanding of the prognostic impact of co-morbid burden in a contemporary cohort of HF patients in this issue of the Journal.3 Their objective was to assess how co-morbid burden relates to adverse outcomes in HF patients across the left ventricular ejection fraction (LVEF) spectrum in a community-based ambulatory setting. Their analysis included 2314 HF patients followed for 31 months, with a 41%/59% split between patients with a LVEF below or above 50%. Fifteen different co-morbidities were chosen based on the catalogue of diseases included in the Charlson co-morbidity index with the addition of obesity and hypertension. Although commonly used as a framework for quantifying co-morbid burden, many of the diseases included in the Charlson co-morbidity index (published in 1987) may not have a great impact on outcomes in contemporary HF patients (e.g. peptic ulcer disease).4 The issue of including co-morbid diseases that do not confer any risk of adverse outcomes in analyses is that risk estimates are likely confounded by other concurrent co-morbidities not adjusted for. Conversely, when the association between additive numbers of co-morbid conditions and outcomes is reported, the estimates are diluted due to diseases included that do not confer any significant risk. In the MAGGIC study, data from approximately 39 000 HF patients [with reduced (HFrEF) and preserved ejection fraction (HFpEF)] included via clinical trials and registries, were used to extract which co-morbid conditions were independent predictors of mortality.5 Renal function, body mass index, systolic blood pressure, diabetes, and chronic obstructive pulmonary disease were identified as co-morbid conditions with independent prognostic influence on mortality. These were all included in the analysis from Iorio and co-workers, along with 10 other co-morbidities. Iorio et al.3 report that the prevalence of co-morbidity was similar across the LVEF spectrum, which perhaps challenges perceptions that HFpEF patients are particularly burdened with extra-cardiac co-morbidity compared to HFrEF patients. Their primary findings show that the risk of adverse events (all-cause mortality, HF hospitalization, and non-cardiovascular hospitalization) was proportional to the number of co-morbid conditions observed in patients. This relationship has also been shown previously in other settings, and hence the findings reiterate the importance of this relationship to be true in an ambulatory setting as well.6, 7 Furthermore, the estimates from Iorio et al.3 suggest that the increase in risk was most pronounced for HF hospitalization compared to the other endpoints examined. In short, every additional co-morbidity added 27% to the risk of future HF hospitalization. The coupling of primarily extra-cardiac co-morbidities with cardiovascular endpoints is intriguing, as it suggests that addressing extra-cardiac co-morbidity may translate into improvements in cardiovascular outcomes. A case in favour of this concept is that of iron replacement therapy for HF patients with iron deficiency. Here intravenous iron therapy significantly reduced HF hospitalization—a case where optimization of an extra-cardiac co-morbidity was reflected in lower rates of a cardiovascular endpoint.8 Reversely, it may also suggest that HF patients with a high extra-cardiac:cardiac co-morbid burden are less likely to gain prognostic cardiovascular benefits from medications targeting cardiac issues alone. Although this concept is not well explored, evidence supporting pharmacological interventions to HFpEF patients have been less encouraging perhaps due to a higher extra-cardiac:cardiac co-morbid burden than found in HFrEF patients.6, 9 Using population attributable fractions, Iorio and co-workers also quantified the contribution of individual co-morbidities to outcomes. The co-morbidities that contributed most to risk of mortality were chronic kidney disease (22%), anaemia (21%), diabetes (18%), and chronic obstructive pulmonary disease (12%). This analysis is especially useful in helping us to prioritize our efforts with the patients. However, knowing that these patients were usually burdened with more than one disease, the single-disease estimates are prone to the same confounding errors as the risk estimates discussed earlier. In fairness, their shortlist of the most significant culprits match well with those reported from various HF risk scores (e.g. MAGGIC,5 3C-HF score10). As the featured paper helps us to revisit and quantitate the issue of HF and co-morbidity, this begs the question—should we treat co-morbidities differently in our HF patients compared to patients without HF? In favour of this notion are data recently published that demonstrated a dramatic U-shaped relationship between glycated haemoglobin and mortality in HF patients with diabetes, notably different to the relationship observed in diabetes patients without HF.11, 12 Another example of a possible interaction between HF and co-morbidity is the well documented case that statin therapy reduces the risk of mortality in patients with ischaemic heart disease; however, statin therapy for patients with ischaemic heart disease and overt HF is not equally beneficial.1, 13 Both examples suggest that co-morbidities and HF are intertwined, and managing the individual diseases as separate entities does not necessarily translate into additive benefits for the patients. It is this complex relationship that we need to explore further, fuelled by the data presented by Iorio and co-workers. Their data nicely demonstrate that ambulatory HF patients fare worse with an increasing number of co-morbidities. This seems to be a consistent finding across HF populations. For the time being, changing clinical practices based on observational data does not seem prudent in general. We need to acknowledge that the elephant is still present and is not moving much until randomized trials deliver answers of how best to manage the interactions between HF and co-morbidities. Conflict of interest: none declared.