Serum potassium and risk of death or kidney replacement therapy in older patients with advanced CKD

Serum potassium. Acute derangement is undoubtedly the bane of the middle-grade nephrologist. As a registrar, I don’t remember a single on-call where I wasn’t contacted by either the biochemistry lab, a dialysis unit, another specialty, the emergency department or somewhere else (sometimes all!) with an abnormal potassium to sort out. With so many calls, it is easy to become blasé about potassium readings that are “not too bad”. It is also fair to say that a high potassium reading tends to engender rather more fear and urgency than a low one, even though significant derangement in either direction potentially puts people at increased risk of cardiac arrhythmia and morbidity. 

Similarly, when managing patients with CKD or heart failure (or both), serum potassium towards the higher end of (or just above) normal can cause difficulty when it comes to optimising cardiovascular and renal protective medications, particularly ACE inhibitors, angiotensin receptor blockers, mineralocorticoid receptor antagonists and now SGLT2 inhibitors. Low potassium readings in clinic, when looking after chronic but stable patients don’t cause the same problems and, perhaps, are not considered concerning in the same way, even though chronic hypokalaemia is associated with a faster decline in kidney function (hypokalaemic nephropathy), resulting from a chronic interstitial nephritis.

The excellent study by de Rooij and colleagues, published last month in the American Journal of Kidney Diseases, suggests we should, perhaps, consider derangement of potassium in either direction in a similar way. This international cohort study looked at the outcomes of older patients (>65 years) with CKD stage 4–5, with respect to their serum potassium levels. Once recruited, participants were followed up every 3–6 months, with repeated blood tests until kidney replacement start, death, loss to follow-up, consent withdrawal or end of follow-up – the study ran from 2012 until 2021. The main outcome of interest was initiation of the combined outcome of kidney replacement therapy or death. 

Here is (one of) the important part(s). Potassium levels were monitored at each blood test according to pre-specified categories: ≤3.5, >3.5 to ≤4.0, >4.0 to ≤4.5, >4.5 to ≤5.0, >5.0 to ≤5.5, >5.5 to ≤6.0 and >6.0 mmol/L. The relationships between categorical serum potassium levels, and the combined outcome of death and kidney replacement therapy initiation, were assessed with modelling that included serum potassium as a time-dependent variable updated every 3–6 months. Models were adjusted for important covariates, including age, sex, smoking, diabetes, use of RAAS inhibition and kidney function. 

The study included over 1700 people, with nearly 6100 potassium measurements taken during follow-up. Median time until death or kidney replacement therapy start was 2.6 years. The “safest” serum potassium was 4.9 mmol/L, but the real take-home message was the clear U-shaped relationship between time-dependent serum potassium and the combined endpoint of death or kidney replacement therapy start over 8 years of follow-up. For example, patients with a time-dependent serum potassium ≤3.5 mmol/L had an adjusted HR for the combined endpoint of 1.6 compared to patients with serum potassium >4.5 and ≤5.0 mmol/L, whereas serum potassium >6.0 mmol/L resulted in an adjusted HR of 2.2.

This is a great study, without the limitations of predecessors. Serum potassium readings in this study were taken when patients were well, routinely, not just as part of clinical care, thereby reducing selection bias, while the prospective design and recruitment of incident patients reduced the chance of survivor bias.

But what can we do with the results? Well, I think these data tie in nicely with prevailing discussion about optimising pharmacotherapy in patients with advanced CKD who may well have under-treated cardiovascular risk. Flooding these patients with medications that raise their serum potassium significantly is not good but, in an era where we have really effective medications for managing hyperkalaemia, we should perhaps consider more liberal use of new binder therapies to optimise cardiovascular and kidney protective treatments, whilst still aiming for normokalaemia. Similarly, for patients with chronically low potassium, should we consider using RAASi therapy (and related drugs) to raise serum potassium and optimise cardiovascular/kidney protective pharmacotherapy? Or should we simply be more open to replacing potassium that is perhaps lost as a consequence of diuretic therapies? Perhaps a bit of both. 

I commend the authors on this excellent study, which seems to show that serum potassium is a modifiable risk factor for older patients with advanced CKD. Working to optimise serum potassium levels has the potential to reduce (early) starts on kidney replacement therapy and death.

A digest of the study can be read here.