Hypo- and Hyperkalaemia in Heart Failure. Navigating Between Scylla and Charybdis

This article refers to ‘Association between potassium level and outcomes in heart failure with reduced ejection fraction: a cohort study from the Swedish Heart Failure Registry’ by L.B. Cooper et al., published in this issue on pages 1390–1398. In chronic kidney and cardiovascular disease clinical trials and registries, both hypokalaemia and hyperkalaemia have consistently been found to be associated with poor clinical outcomes.1 In this issue of the Journal, Cooper et al.2 using the nationwide Swedish Heart Failure Registry, report a V-shape relationship between potassium and death in a chronic heart failure population with reduced ejection fraction, with a narrow nadir of ‘normal’ serum potassium concentration of 4.2 mmol/L associated with the lowest death rate. A similarly narrow (4.2–4.5 mmol/L) nadir was observed in a similarly large Danish nationwide registry.3 In their important report, Cooper et al.2 used extensive multivariable adjustment, attempting to mitigate the many expected confounders that have commonly confounded multiple previous similar reports. They have also carefully looked at both sides of the coin, i.e. hypokalaemia and hyperkalaemia and their respective associations with both short-term and long-term outcomes. A major finding was that hypokalaemia was both a risk marker and a risk factor associated with short- and long-term mortality, while hyperkalaemia was a risk factor associated with mortality, only at short term, and a risk marker over time.2 The risk associated with hyperkalaemia in heart failure is mainly a consequence of the frequent discontinuation and/or underdosing of renin–angiotensin–aldosterone system inhibitors (RAASi), caused by incident hyperkalaemia or the concern of thereof.4 The report of Cooper et al.2 rightly conclude that ‘clinicians should be targeting a narrower goal potassium range than what has traditionally been considered within the normal range’ (i.e. 4–5 mmol/L). How can this be achieved is an important practical issue. Indeed, despite its strong rationale, how and whether the specific management of hypokalaemia and hyperkalaemia may ultimately lead to improved clinical outcomes still lacks randomised evidence. How best to prevent mortality related to hypokalaemia, and whether the use of potassium binders may ultimately improve outcomes whilst facilitating the uptitration of RAASi in heart failure patients prone to hyperkalaemia or their maintenance in hyperkalaemic patients warrant appropriate testing. The ongoing DIAMOND study is a randomised trial examining the safety and efficacy (cardiovascular death or hospitalisation as primary endpoint) of the potassium binder patiromer for the enablement of spironolactone and management of hyperkalaemia in subjects receiving RAASi medications for the treatment of heart failure (ClinicalTrials.gov NCT03888066). Interestingly, Nunez et al.,5 in a (Spanish) registry setting, reported that potassium normalisation in initially hypokalaemic or hyperkalaemic chronic heart failure patients ultimately resulted in improved clinical outcomes. A key element is to properly monitor potassium levels. Guidelines recommend close monitoring of serum potassium and renal function during periods of RAASi titration until the patient is stable and at regular intervals thereafter.6 In the landmark randomised trials on mineralocorticoid receptor antagonists (MRAs) such as RALES (Randomised Aldactone Evaluation Study), EPHESUS (Eplerenone Post-Acute Myocardial Infarction Heart Failure Efficacy and Survival Study), and EMPHASIS- HF (Eplerenone in Mild Patients Hospitalisation and Survival Study in Heart Failure), management of RAASi dosing was dynamic and algorithmically driven by a protocolised monitoring of serum potassium and creatinine.7 In the EPHESUS and EMPHASIS-HF trials, the median [interquartile range (IQR)] follow-up was 16 (12–20) and 21 (10–33) months, respectively. The anticipated number of serum potassium measurements according to the protocol was 10 (8–11) and 7 (5–10) in each trial, respectively. The median (IQR) number of actual serum potassium measurements was 11 (9–13) and 8 (5–11), respectively, i.e. some additional measurements were performed after a clinical event or after a non-anticipated medication change. Using these databases, adding time-updated variables including potassium concentrations and MRA intake improved the prediction of cardiovascular death (on top of the MAGGIC score) in patients with heart failure eligible for RAASi and MRA therapy, while hypokalaemia and hyperkalaemia actually carried the same level of risk.8 Importantly, more frequent potassium measurements improved the association between risk score and outcome. A web-based calculator was created to allow easy determination of this new risk score (http://cic-p-nancy.fr/CardiovascularriskscoreCalculator/), from which it was hypothesised that this new risk score including MRA usage and potassium may be of value in helping physicians to better use MRAs, avoid unnecessary and potentially detrimental permanent discontinuations, and therefore improve cardiovascular outcomes in patients with chronic heart failure with reduced ejection fraction or heart failure after acute myocardial infarction with left ventricular dysfunction.8 Unfortunately, close serum potassium monitoring is uncommon in real-life settings, with less than one-third of patients starting MRA therapy being monitored for the ‘minimum minimorum’ (i.e. early post-initiation then within 3 months) in surveys performed in the US and in Sweden.9, 10 In the US survey, 5782 (55.4%) and 2328 (22.3%) patients received no testing in early or extended follow-up, respectively.9 Disease management programmes may play a key role.11 Further empowering disease management programmes with remote monitoring tools is highly desirable. Strikingly, in a recent prospective 5-day-a-week home monitoring survey in heart failure patients post-discharge combining haemoglobin/haematocrit, potassium and creatinine measurements using capillary blood measurements with the i-STAT analyser, home monitoring was able to capture a substantial number of clinically relevant cardio-renal (congestion/decongestion, worsening renal function) and electrolyte events (hypokalaemia/hyperkalaemia), which are frequently overlooked and potentially actionable (Figure 1).12 Recent attempts to non-invasively detect hyperkalaemia with two electrocardiographic leads have been developed.13 However, whether such devices are also able to detect hypokalaemia is unknown. Furthermore, whether such an approach would be sufficiently sensitive to detect subtle and potentially relevant potassium variations around the narrow nadir of 4.2 mmol/L is questionable, owing to the well-known lack of sensitivity of the electrocardiogram. Such uncertainties along with the unmet clinical need of properly monitoring potassium in patients prone to experience dyskalaemia clearly opens new and interesting avenues for developing novel technologies. P.R. and F.Z. are supported by the RHU Fight-HF, a public grant overseen by the French National Research Agency (ANR) as part of the second ‘Investissements d'Avenir’ programme (reference: ANR-15-RHUS-0004), and by the French PIA project ‘Lorraine Université d'Excellence’ (reference: ANR-15-IDEX-04-LUE). Conflict of interest: P.