Complex Relationships Between Co-Morbidity, Outcomes, and Treatment Effect in Heart Failure

This article refers to ‘Contribution of cardiac and extra-cardiac disease burden to risk of cardiovascular outcomes varies by ejection fraction in heart failure’, by E. Wolsk et al., published in this issue on pages 504–510. It has long been recognized that cardiovascular (CV) co-morbidities such as coronary artery disease are risk factors for ‘incident’ heart failure (HF), especially HF with reduced and mid-range ejection fraction (EF) (HFrEF and HFmrEF).1 Non-CV co-morbidities such as diabetes mellitus and kidney disease are also thought to contribute directly to the gradual development of incident HF, particularly HF with preserved EF (HFpEF).2 For ‘prevalent’ HF, the 2016 European Society of Cardiology HF guidelines emphasize the importance of co-morbidities that may affect both treatment and outcome in HF.3 The prevalence of co-morbidities in HFrEF, HFpEF, and more recently the newly designated HFmrEF have been well described and non-CV co-morbidities generally become more common with higher EF.1, 4, 5 Cause-specific outcomes have been well described and also become more non-CV related with higher EF.6-8 However, the contribution of the former to the latter is difficult to assess and less well known.4 Indeed, there is uncertainty regarding the pathophysiological and clinical role of both CV and non-CV co-morbidities and their relationship to both CV and non-CV outcomes.5 Are co-morbidities risk markers or also risk factors for poor outcomes? How do co-morbidities contribute to progression of and outcomes in HF? Does treatment of CV and non-CV co-morbidities affect CV and non-CV outcomes in HF (Figure 1)? Do these aspects differ by EF? In this issue of the Journal,9 the CHARM trial database was used to assess the association between baseline CV and non-CV ‘disease burden’ and CV outcomes in HFrEF and HFpEF (non-CV outcomes were not studied). Disease burden was grouped into demographic, CV, and non-CV based on 13 variables in the well-validated MAGGIC risk score.10 In assessing the association between these variables and outcomes, the authors did not simply estimate hazard or incidence rate ratios, but elegantly estimated the population attributable risk (PAR) for each variable in relation to each outcome. A hazard ratio estimates how much a variable is associated with increased or decreased risk but does not put its risk into the context of other risk markers or risk factors. PAR indicates what proportion of total events can be attributed to the particular variable and as such considers both how common and how serious a particular characteristic is and thus provides a more useful clinical context. In this selective CHARM trial population, the total MAGGIC risk was greater in HFrEF, consistent with greater risk of CV outcomes and all-cause death in HFrEF vs. HFpEF previously reported from CHARM.7, 11 Baseline CV burden was higher in HFrEF and demographic burden higher in HFpEF. Contrary to what is often assumed, non-CV burden was as high or even higher in HFrEF vs. HFpEF. However, in non-selective settings non-CV burden may be higher in HFpEF1 and mortality is also more similar between HFpEF and HFrEF.1, 12 In HFrEF, there is a broad armamentarium of evidence-based HF interventions.3 The present CHARM analysis9 indicates that in HFrEF, CV co-morbidities represent a high PAR for CV outcomes. Thus a primary challenge in HFrEF is not to treat non-CV co-morbidities per se, but to optimize utilization of HF treatment in the face of limiting non-CV co-morbidities such as chronic kidney disease, hyperkalaemia, symptomatic hypotension, frailty, and polypharmacy.13-15 In contrast, in HFpEF (and HFmrEF), there is no evidence-based therapy. The CHARM analysis indicates that in HFpEF, the high PAR of non-CV co-morbidities may explain why HFpEF trials have failed. However, the non-CV co-morbidities in HFpEF still contributed to CV outcomes. Therefore, treating co-morbidities such as iron deficiency and diabetes may be particularly promising in HFpEF, and ongoing trials with intravenous iron and sodium–glucose co-transporter 2 (SGLT2) inhibitors in HFpEF may pave the way for a new approach in HFpEF. A limitation of the CHARM-Preserved trial was the dichotomy of HFrEF and HFpEF at EF 40% that neglects the grey zone or mid-range 40–49%. This EF category is increasingly well understood. It is intermediate between HFrEF and HFpEF in many respects but distinctly resembles HFrEF in other respects.16 The dichotomization also disregards the limitations of EF as a classifier,17 which the authors addressed in part by spline analyses with EF as a continuous variable. Nevertheless, EF is the most commonly used classifier, and the HFrEF, HFmrEF, and HFpEF categories appear to differentiate patient characteristics and response to therapy quite well. The time now is not to dwell on failed HFpEF trials but perhaps to design trials that recognize the potential for conventional HF drugs in HFmrEF and novel interventions targeting co-morbidities in HFpEF. Conflict of interest: grants to authors institution from AstraZeneca, Relypsa, Novartis, Boehringer; consulting from AstraZeneca and Vifor Pharma.