Serum potassium in patients with chronic heart failure: once we make a U-turn where should we go?

This editorial refers to ‘Associations of serum potassium levels with mortality in chronic heart failure patients’†, by M. Aldahl et al., on page 2890. There has been an increased focus on the association between serum potassium (K+) and mortality over the past few years.1 Several studies have pointed out a U-shaped relationship with an increased risk of death both below and above the normal range of serum K+. While there appears to be agreement that there is a U-shaped relationship between serum K+ and death, the actual serum K+ values at which the risk becomes evident appears to vary depending upon whether or not various comorbidities such as chronic kidney disease (CKD), diabetes mellitus (DM), heart failure (HF), and hypertension are or are not present, as well as age. For example, a recent analysis of a relatively large US insurance database showed that the U-shaped relationship of serum K+ with death was much flatter in normal individuals than in those with comorbidities and age >65 years.2 However, there remains controversy as to the exact level of serum K+ in a patient with HF at which the clinician should become concerned and consider altering therapy. Some clinicians become concerned for a patient with HF when the serum K+ is < 3.5 or > 5.5 mmol/L, whereas others become concerned and consider altering cardiovascular drug therapy at a serum K+ <4.0 or > 5.0 mmol/L. Investigators from the Danish National registries had previously analyzed the relationship between serum K+ and death in patients with hypertension3 and in patients with acute HF,4 noting a U-shaped relationship. In this issue of European Heart Journal, these investigators now report the relationship between serum K+ and mortality in patients with chronic HF.5 They identified 19 549 patients who had a measurement of serum K+ within a minimum of 30 days after initiating medical therapy with a loop diuretic and a renin angiotensin system inhibitor (RAAS-I). Similar to their study in patients with acute HF, they found that deviations in serum K+ above and below ‘normal’ were associated with an increased risk of death: a U-shaped relationship. Of particular interest in this relatively large and well characterized group of patients with chronic HF was that the lowest risk for death occurred with a serum K+ between 4.2–4.4 mmol/L. At a serum K+ between 3.9 and 4.1 mmol/L, a value many clinicians would believe to be within the normal range, the hazard ratio (HR) for death was 1.29 [confidence interval (CI) 1.08–1.55]. Similarly, a slight increase in serum K+ between 4.8–5.0 mmol/L, also considered by many clinicians to be within the normal range (or at least a range that would not cause concern), the HR for death was 1.34 (CI 1.10–1.63). At a serum K+ <3.9 or greater than 5.1 mmol/L there were further increases in the risk of death, reaching a HR of 3.16 (CI 2.43–4.11) for a K+ between 2.8–3.4 mmol/L and 3.31 (CI 2.61–4.20) for a K+ between 5.6–7.4 mmol/L. However, it should be pointed out that they provide no information as to whether patients had a reduced or preserved left ventricular ejection fraction. This may be important since the therapeutic options may differ between them (Figure 1). A U-way of reshaping heart failure management? Adapted from Aldahl et al., Figure 3.5 HFrEF, HFmrEF, HFpEF: heart failure with reduced, mid-range, preserved ejection fraction; RAAS-I, renin angiotensin system inhibitor. The finding in the present study that patients with chronic HF in whom there is even a slight deviation in serum K+ <4.2 or >4.4 mmol/L are at increased risk of death should prompt a reconsideration of our current approach to patients with chronic HF. One should no longer feel complacent with a patient with chronic HF and a serum K+ between 3.5–4.0 mmol/L or between 5.0–5.5 mmol/L. However, while the present study has pointed out the increased risk of both mild hypo- and hyperkalaemia, the question as to how we can avoid these levels, what should be done when one encounters these values, and how to blunt this U-shaped relationship is less certain and warrants future evaluation (Figure 1). The risk factors for hypokalaemia are well known and include the use of loop diuretics. While diuretics have not been conclusively shown to reduce cardiovascular death in patients with chronic HF there is general agreement that their use is essential to relieve congestion. Thus, it may be difficult to avoid mild hypokalaemia in a patient with HF. One commonly used approach is to add a K+ supplement such as KCl. However, it should be pointed out that >80% of patients with hypokalaemia in the present analysis were on a K+ supplement and there did not seem to be any significant effect of these supplements on reducing the risk of death associated with the occurrence of hypokalaemia. Another approach would be to use a mineralocorticoid receptor antagonist (MRA) such as spironolactone or eplerenone, which would tend to increase serum K+. However, while their use may minimize the risk of hypokalaemia it has not eliminated it, as is evident from a review of the major randomized studies evaluating the safety and effectiveness of MRAs such as randomized aldactone evaluation study (RALES), eplerenone post-acute myocardial infarction heart failure efficacy and survival study (EPHESUS), and more recently eplerenone in mild patients hospitalization and survival study in heart failure (EMPHASIS-HF).6 Whether there would be an advantage of using a MRA compared to a K+ supplement in reducing the risk of death associated with a given level of hypokalaemia remains to be determined (Figure 1). Similarly, the risk factors for developing hyperkalaemia are well known and include the presence of CKD, DM, and the use of RAAS-Is, especially MRAs. While one might avoid the risk of hyperkalaemia by avoiding the use of a RAAS-I or reducing the dose once it occurred one might be placing the patient with chronic HF at even greater risk of death since RAAS-Is have clearly been shown to reduce mortality in patients with chronic HF, at least those with chronic HF and a reduced ejection fraction (HFrEF). In view of the recent availability of the K+ lowering agent Patiromer both in the US and Europe one might consider adding this agent to a patient with an increase in serum K+ to reduce the level to between 4.2 and 4.4 mmol/L while maintaining their RAAS-I. However, it should be pointed out that while Patiromer is effective in reducing serum K+ in patients with HF and hyperkalaemia maintained on a RAAS-I,7 there is no evidence as yet that it will reduce the risk of death associated with hyperkalaemia. While the occurrence of hyperkalaemia in a patient with chronic HF is clearly associated with an increased risk of 20 and 90 day death, as is evident from the findings in the present study, it is possible that the increase in serum K+ is merely a marker of increased risk rather than the cause. Finally, an increase in RAAS-I dose has been associated with better outcomes in patients with HF [assessment of treatment with lisinopril and survival (ATLAS)8 and heart failure endpoint evaluation of angiotensin II antagonist losartan (HEAAL)9] despite an increased incidence of hyperkalaemia.10 Whether the addition of Patiromer would allow an increase in RAAS-I dose in normokalaemic HF patients and therefore achieve a better prognosis needs future testing (Figure 1). While we know the risk factors for the development of both hypo- and hyperkalaemia, there is relatively little guidance for the clinician in predicting the risk of death once the patient with HF is found to have either hypo- or hyperkalaemia. This information would be of value in selecting the timing and most appropriate therapeutic strategy to prevent the increased risk of death emphasized by the Danish National registry investigators. Given the increased risk of death in patients with chronic HF both with mild hypo- and hyperkalaemia, one might also reconsider current recommendations for monitoring of serum K+. There are guideline recommendations for the frequency of K+ monitoring in patients with HF administered a RAAS-I11 and suggestions regarding the frequency of K+ monitoring in patients with hyperkalaemia receiving a K+-lowering agent such as Patiromer.12 However, recent data in the US suggests that these recommendations are often not followed and that many patients do not have any monitoring of their serum K+ or renal function after initiation of a RAAS-I.13 In view of the known risk factors for both hypo- and hyperkalaemia, one might consider more frequent monitoring of serum K+ and renal function in those with these risk factors. However, the cost effectiveness and efficacy of any new K+ monitoring strategy will require further evaluation (Figure 1). Thus, while we are indebted to the Danish National registry investigators for emphasizing the U-shaped relationship between serum K+ and death in patients with chronic HF, and especially the increased risk of death at levels of serum K+ <4.2 and >4.4 mmol/L, levels currently considered by some clinicians as not warranting concern, the study, as with most good studies, raises many more questions. Once we cross the nadir of serum K+ in a patient with chronic HF and make a U-turn we will need further direction as to where we go from here.