Dialysate Potassium, Dialysate Magnesium, and Hemodialysis Risk

Dialytic Factors Modulating Potassium Removal

HD removes potassium from the extracellular fluid compartment, which contains of only 2% of total body potassium; the remainder is found in the intracellular space. Diffusion accounts for 85% of dialytic potassium clearance,⁸ and the rate and amount of potassium removal are largely a function of the potassium serum-dialysate potassium gradient.

As serum potassium falls toward dialysate potassium levels (which are fixed) during the course of the treatment, the gradient is reduced, which in turn, decreases the rate of potassium removal. During the first hour of dialysis, rate of potassium decline is the most rapid when the serum-dialysate potassium gradient is largest; a 1-mEq/L fall is typical, but this fall can be greater with larger serum-dialysate gradients. The acute fall in the first hour is followed by a more gradual decline of an additional 1 mEq/L over next 2 hours as the serum-dialysate potassium gradient narrows (Figure 1). During the final hour, serum potassium levels remain steady, although diffusive clearance is still occurring, indicating equilibrium between the rates of potassium removal and the rate of re-equilibration from intracellular space.⁹ After HD, a subacute “rebound” of serum potassium levels occurs as continued mobilization of potassium from intracellular space to extracellular space occurs. The effect of the serum-dialysate gradient on the rate and amount of intradialytic fall in serum potassium as well as the rate of postdialysis rebound in serum potassium levels was studied by Blumberg et al.¹⁰ A higher serum-dialysate gradient results in both more rapid fall in serum potassium levels during treatment and a rapid postdialysis rebound of potassium levels compared with smaller gradients (Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Comparison of intradialytic and postdialytic potassium levels between a serum-dialysate gradient of 5.8 mEq/L (solid line; predialysis serum potassium 6.8 mEq/L and dialysate potassium 1 mEq/L) and a gradient of 4.7 mEq/L (dashed line; predialysis serum potassium 5.7 mEq/L and dialysate potassium 1 mEq/L). The high serum-dialysate gradient condition results in a total excursion of serum potassium levels of approximately 5 mEq/L (3 mEq/L fall and 2 mEq/L rebound) in the 10 hours after the start of treatment compared with approximately 3 mEq/L with the lower gradient condition. Modified from Blumberg et al.,¹⁰ with permission.

Other aspects of the dialysis prescription other than dialysate potassium concentration influence the rate of potassium removal. Increases in serum bicarbonate levels during a dialysis treatment enhance the activity of Na/K/ATPase channels and result in larger shifts of potassium into the intracellular space, further lowering serum potassium levels during HD. This was shown experimentally in a randomized crossover study comparing the effect of three different dialysate bicarbonate concentrations on the rate of serum potassium decline, keeping dialysate potassium concentration stable.¹¹ High dialysate bicarbonate (39 mEq/L) was associated with the greatest decline in serum potassium levels compared with standard (35 mEq/L) and low (27 mEq/L) bicarbonate concentrations, but there was no significant difference in the total potassium removal, concordant with enhanced acute intracellular potassium shifts rather than enhanced potassium removal. Convective clearance of potassium plays a small but not insignificant role in total dialytic potassium removal; recent mass-balance studies have shown that potassium mass removed by ultrafiltration accounts for approximately 6% of the total potassium mass removed.⁹ Finally, compared with glucose-free or low glucose–containing dialysates, high glucose–containing dialysate solutions can also lead to higher potassium removal via osmotic shifts of intracellular potassium to the extracellular space.^12,13

Serum and Dialysate Potassium Levels and Clinical Outcomes

Whereas the normal range for serum potassium levels is typically reported between 3.5 and 5.0 mEq/L in the general population, the optimal range of potassium concentration in patients on dialysis is higher. In a study of 2134 patients on HD, a predialysis serum potassium level of 5.1 mEq/L was associated with the lowest risk of peridialytic sudden cardiac arrest, whereas potassium levels above and below 5.1 were associated with increasing risk.¹⁴ In another study examining predialysis serum potassium levels and survival within a cohort of 81,013 patients on HD, potassium concentrations between 4.6 and 5.3 mEq/L were associated with the lowest incidence of all-cause mortality.¹⁵ In the latter study, potassium levels ≥5.6 mEq/L were associated with significantly increased mortality after adjustment for confounders related to comorbidities and nutritional status, whereas risk relationships between potassium levels <4.6 mEq/L and mortality were abrogated after accounting for confounding factors related to malnutrition. A more recent study of 55,183 patients in the Dialysis Outcomes and Practice Patterns Study (DOPPS) multinational cohort confirmed these findings, with the lowest risk of death among patients with predialysis serum potassium levels between 4 and 5.5 mEq/L, a significant increase in the risk of death and arrhythmia outcomes at levels ≥5.6, and attenuation of risk associations with potassium levels <4 mEq after accounting for potential confounding from malnutrition indicators.¹⁶ Even if risks associated with hypokalemia are not entirely mediated through malnutrition and inflammation-related factors, predialysis hypokalemia is encountered less frequently compared with hyperkalemia, making hyperkalemia management a more pressing public health concern. In two studies surveying predialysis serum potassium values obtained within a large dialysis organization, approximately 20% of all measurements were ≥5.5 mEq, and about 12.5% of measurements were ≥6.0,¹⁷ whereas predialysis potassium levels <4.0 accounted for only 9% of all measurements.¹⁵

The prevalence and prognostic significance of immediate postdialysis potassium levels and in particular, postdialysis hypokalemia are unknown, because postdialysis levels are not routinely measured. Regardless, the potential effect of falling potassium levels during and after the HD procedure is compelling. In a study of patients with ESRD who had wearable defibrillators in place, 70.0% of the captured arrhythmia events occurred during the session, and 2.8% were captured immediately after the dialysis procedure.¹⁸

Optimal Dialysate Potassium Level and Outcomes

The challenge of selecting appropriate dialysate potassium levels is balancing the need to achieve sufficient potassium removal to avoid interdialytic hyperkalemia, while minimizing the potential hazards posed by lowering potassium too rapidly during treatment. The lack of consensus on the ideal dialysate potassium concentration to achieve these goals is reflected in the large variation in worldwide dialysate potassium prescription; recent data from the DOPPS report the range of prevalent use of potassium dialysate <2 mEq/L from as low as 3% in the United States to as high as 62% in Spain.¹⁶ The challenge of selecting the appropriate dialysate potassium concentration is further compounded by a commonly applied standard of care to monitor predialysis serum potassium levels only once monthly, with no other data to guide dialysate potassium choice for the remainder of treatments throughout the month. The safety of low-potassium dialysate has been a focus of concern given the possibility that dialysis-induced potassium lowering may provoke cardiac arrhythmias and sudden cardiac death (SCD). An early study documented that higher efficiency dialysis could achieve dramatic removal of potassium with low-dialysate potassium concentration, but this study also noted that “only” one of the 11 patients studied with zero potassium bath experienced “high grade ventricular ectopy.”¹⁹ Since then, multiple large retrospective studies have investigated associations between dialysate potassium levels, predialysis serum potassium levels, and risk of sudden death, cardiac events, and all-cause mortality. These studies and their significant findings are summarized in Table 1. Although subject to indication bias and other sources of confounding, in general, large cohort studies have identified increased risks of SCD associated with use of low-potassium dialysate <2 mEq/L,^14,20 with one study identifying an increased risk of SCD among patients exposed to dialysate potassium <3 mEq/L compared with ≥3 mEq/L.²¹ The risks associated with low-potassium dialysate are principally seen among patients with low to normal predialysis serum potassium; importantly, no study has shown significant risk associated with low-potassium dialysate among patients with predialysis serum potassium levels ≥5 (Figure 2). A smaller retrospective study noted a significant reduction of SCD rates associated with a change in dialysis unit policy toward use of low-potassium (1 mEq/L) dialysate in all patients with predialysis serum potassium levels >5.5 mEq/L, but this study was potentially confounded by other changes in dialysis care that occurred during the same time period.²² Although no long-term prospective controlled studies have been conducted to examine the effect of low-potassium dialysate on hard outcomes, several other short-term crossover studies using cardiac monitoring devices have been conducted to assess subclinical arrhythmic events, such as ventricular ectopy,^19,23,24 premature ventricular complexes,²⁵ and changes in electrocardiographic conduction parameters, such as the QT interval and QT dispersion.^26,27 Most but not all studies observed higher rates of ventricular ectopy and QTc prolongation associated with lower-potassium dialysates of 0 or 1 mEq/L compared with higher-potassium dialysates. However, the validity of these markers as predictors of SCD and mortality is questionable at best, with long-term studies finding no association between these ECG markers and mortality.²⁸

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Sudden cardiac arrest (SCA) risk according to serum and dialysate potassium. The risk of SCA remains higher or equivalent (overlapping 95% confidence intervals) with lower-potassium dialysate <2 mEq/L compared with ≥2, even among patients who are hyperkalemic. Reprinted from Pun et al.,¹⁴ with permission.

Overall, largely circumstantial evidence points to the hazards of lower-potassium dialysate <2 mEq/L, and the evidence for risk is strongest for patients with predialysis serum potassium levels <5 mEq/L. Whether lower-potassium dialysate is appropriate or potentially beneficial for patients with higher predialysis serum potassium levels is unclear. In patients who are hyperkalemic, randomized trials are needed to compare the effects of a strategy to reduce hyperkalemia-associated arrhythmias using low-potassium dialysate with a strategy that uses higher-potassium dialysate to avoid the potential arrhythmic hazards of increased serum-dialysate gradients. These studies are made more feasible with the availability of miniaturized implantable loop monitors, which can be easily inserted with minimal risk and provide full capture of all arrhythmias that occur over extended periods.^29,30

The potential hazards of using higher-potassium dialysate concentrations ≥3 should also be noted. A study of >81,000 United States patients examined mortality rates on the basis of various combinations of dialysate potassium concentrations and predialysis serum potassium levels, and the highest mortality rate was noted for patients exposed to high-potassium dialysate >3 mEq/L with predialysis serum potassium levels ≥5.¹⁵ Studies done among Canadian patients on dialysis³¹ and the DOPPS international cohort¹⁶ have also observed an increased risk of death among users of dialysate potassium ≥3 mEq/L; however, in both studies, these risks were attenuated after adjustment for factors relating to malnutrition and inflammation.

In summary, although no randomized, controlled data exist to identify best dialysate potassium choices, current evidence most strongly supports the avoidance of low-potassium dialysate <2 mEq/L among patients with known predialysis serum potassium levels ≤5. Little evidence exists to support the complete avoidance of dialysate potassium <2 mEq or >3 mEq/L at the extremes of predialysis serum potassium concentrations. However, given the reported risk associations, we suggest that use of these dialysate concentrates be accompanied by frequent serum potassium monitoring and dialysate tailoring when serum levels come with the normal range.

Serum and Dialysate Magnesium and Potential Interactions with Dialysate Potassium

The serum concentration of magnesium is recognized in the general population as a risk factor for arrhythmias, but in patients with ESRD, the risk of excursions in magnesium levels has been largely ignored. Magnesium is a divalent ion with a normal plasma concentration between 0.65 and 1.0 mmol/L (1.3–2.0 mEq/L; 1.6–2.4 mg/dl), and extracellular magnesium is estimated to represent only 1% of total body stores.³² In patients who have ESRD, magnesium concentrations tend to be slightly higher and depend on native kidney function, dietary intake, concomitant medications, and dialytic clearance. A typical diet contains approximately 360 mg magnesium, and 70%–75% of this is excreted in the feces. Normal kidneys excrete about 100 mg magnesium per day. This allows balance between serum magnesium and magnesium contained in bone, muscle, and other tissues. During an HD session, the plasma magnesium is not fully “accessible” to shifts across the HD membrane, because magnesium has approximately 25% protein binding (predominantly albumin) and because 5%–16% of magnesium complexes with anions, such as bicarbonate, phosphate, and citrate.³² The amount of magnesium cleared with HD treatments depends on the magnesium concentration in dialysate. With dialysate magnesium 0.75 mmol/L, 565 mg magnesium will be cleared, and with dialysate magnesium 0.25 mmol/L, about 3100 mg will be cleared.^33,34

Extreme concentrations of serum magnesium can be common in clinical practice. In a recent survey of hospitalized patients, magnesium levels above 0.9 mmol/L (2.1 mg/dl) were present in 31.5% of patients, and levels below 0.7 mmol/L (1.7 mg/dl) were noted in 20.2% of patients.³⁵ Either extreme was associated with higher risk of adverse clinical outcomes, but the risk of hypermagnesemia was more pronounced. In the case of patients with CKD, why would extremes of magnesium be detrimental? High serum magnesium can cause oversuppression of PTH and other mineral metabolism adverse effects.³⁶ Low magnesium concentrations are speculated to cause endothelial dysfunction, enhance soft tissue calcification, and promote cardiac arrhythmias.^37,38 Hypomagnesemia by itself is capable of prolonging QT interval and increasing risk for ventricular arrhythmias.³⁹ Therefore, it is imperative to avoid overlaying risk exposures, such as hypomagnesemia, with hypokalemia.

Risks associated with serum magnesium concentrations are not straightforward in patients with ESRD. Several studies evaluated risks associated with serum magnesium concentrations in patients who have ESRD, and the observations differ. A retrospective cohort study of 142,555 Japanese patients on HD determined that there was a J-shaped relationship between serum magnesium concentrations and 1-year mortality, with a “sweet spot” around 1.2 mmol/L (2.8 mg/dl)⁴⁰ (Figure 3). Perhaps related to differences in dietary magnesium intake between countries, a study of United States patients on prevalent HD reported slightly different results.⁴¹ Among 27,544 patients followed in the latter study, a Cox proportional hazards model depicted a linear decline in death risk from the lowest to the highest serum magnesium category, with the best survival observed at serum magnesium levels ≥1.25 mmol/L (2.5 mEq/L; hazard ratio, 0.68; 95% confidence interval, 0.56 to 0.82) and no apparent risks associated with the highest magnesium category. Concordant results were reported in a retrospective study of 9359 United States patients on incident HD, but the study also suggested that the risk of low serum magnesium was attenuated when statistical adjustments were made for markers of malnutrition and inflammation.⁴² More recently, a post hoc study of the Convective Transport Study (CONTRAST) evaluated associations of baseline serum magnesium values of patients assigned to either HD or hemodiafiltration.⁴³ In a subset of 365 randomized patients (who had stored blood samples), the CONTRAST analysis observed that, for every increase of serum magnesium concentration by 0.1 mmol/L, the hazard ratio for all-cause mortality was 0.85 (95% confidence interval, 0.77 to 0.94). Very few of the patients in the CONTRAST included in the study had magnesium concentrations above 1.4 mmol/L.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Adjusted associations between the serum magnesium level and all-cause mortality among 142,555 patients on HD. The dashed lines represent the 95% confidence intervals. Reprinted from Sakaguchi et al.,⁴⁰ with permission.

How should a clinician interpret these observations? It is evident that nutritional issues and medications, such as proton pump inhibitors, can lower serum magnesium levels along with nutritional issues, and part of the risks observed related to hypomagnesemia may be attributed to these confounding factors.³² However, given the potential risks associated with hypomagnesemia and a growing awareness that proton pump inhibitors are often inappropriately prescribed,⁴⁴ a focused review of medication lists and a nutritional assessment should be a routine response to hypomagnesemia in patients on HD. However, if supplemental magnesium is to be considered, extremely high serum concentrations should be avoided.

Clearly, the dialysis prescription can affect serum concentrations of magnesium. Recent data suggest that hypomagnesemia occurs 33% of the time at dialysate concentrations of 0.25 mmol/L (0.5 mEq/L) and only 5% of the time among patients exposed to dialysate magnesium of 0.5 mmol/L (1.0 mEq/L).³² In fact, the trends in prescribed dialysate magnesium varied over the past three decades. Perhaps because of concerns about hypermagnesemia causing osteomalacia and an early report on the improvement of uremic pruritus after lowering dialysate magnesium concentration,⁴⁵ dialysate magnesium concentrations decreased from early values of approximately 0.75 mmol/L (1.5 mEq/L) to as low as 0.25 mmol/L (0.5 mEq/L).⁴⁶ However, with recognition that higher serum magnesium levels may limit arterial calcification, the pendulum has swung back in recent years.⁴⁷ Perhaps on the basis of these concepts, current manufacturers commonly provide dialysate products with final magnesium content of 0.5 mmol/L (1.0 mEq/L).⁴⁸

It is important to recognize, however, that important interactions exist among dialysate magnesium content and other dialysate constituents. For example, magnesium concentration in dialysate is likely influenced by citrate content. One study showed that citrate-containing dialysate created a mildly negative magnesium balance with routine HD.⁴⁹ In addition to biochemical interactions, low-magnesium dialysate could potentiate risks associated with low-potassium dialysate given the effects of hypomagnesemia on risk of hypokalemia via promotion of intracellular potassium shifts.⁴⁶ Given the high prevalence of hypomagnesemia, risk associations with mortality, and potentiating effect on potassium shifts, we suggest standard prescription of higher-magnesium dialysate levels of 0.5 mmol/L (1.0 mEq/L), particularly among patients who are hypokalemic and patients subject to large serum-dialysate potassium gradients.

Nondialytic Treatments to Improve Potassium Homeostasis

It is important not to discount the potential utility of nondialytic management strategies to control serum potassium levels and decrease the need to expose patients to the potential hazards of low dialysate potassium. The involvement of renal dieticians to provide dietary counseling is indispensable, and as part of their dialysate potassium adjustment algorithm, Lee and Mendelssohn⁷ also engaged renal dieticians in reviewing all dialysate potassium levels, alerting providers to potential serum-dialysate potassium mismatches. Attention to other potential contributing factors to abnormal potassium levels, such as ongoing gastrointestinal losses or poor nutrition among patients with acute or chronic hypokalemia, is also necessary. Studies have shown that angiotensin-converting enzyme inhibitors and spironolactone influence potassium levels even in patients on HD who are oligoanuric, presumably through their effects on colonic secretion; discontinuing these agents may have a small but appreciable effect on lowering predialysis serum potassium levels by 0.2–0.6 mEq/L.^55,56 Large prospective clinical trials are underway to evaluate whether mineralocorticoid receptor blockers will improve clinical outcomes in patients on HD.⁵⁷ However, any positive result from use of these agents may be tempered by lowering of dialysate potassium concentrations in patients with resulting cases of predialysis hyperkalemia.

Potassium binding resins, including sodium polystyrene sulfonate and calcium polystyrene sulfonate, have been available for nearly 60 years; however, their efficacy in treating hyperkalemia has never been rigorously tested, and potentially fatal risk of colonic necrosis is well known.⁵⁸ The use of these agents for chronic management of hyperkalemia in patients on HD is limited by poor gastrointestinal tolerability. A retrospective study of 70 chronically hyperkalemic patients on HD who were prescribed calcium polystyrene sulfonate found that only 40% of patients were adherent after 3 months, despite intensive adherence reinforcement; additionally, in the medication-adherent group, predialysis serum potassium levels did not decrease after 3 months of therapy and intensive dietary education.⁵⁹ Loop diuretics have been used to manage hyperkalemia, but their efficacy in patients on HD is limited to only patients with residual renal function and associated with high, potentially ototoxic doses required to produce kaliuresis.⁶⁰ Thus, these agents have poor utility for the management of chronic hyperkalemia.

However, two new oral potassium binding agents (sodium zirconium cyclosilate and patiromer) have been shown to be effective in reducing serum potassium levels among patients with moderate CKD.^61,62 These new medications have only been tested in small populations of patients on HD,⁶³ but in contrast to the well known gastrointestinal side effect profile of sodium polystyrene sulfonate and potential risks of fatal colonic necrosis that make chronic use impractical in patients on HD, these new agents appear to be well tolerated. If proven effective and safe in patients on HD, the potential for novel potassium binding agents to reduce predialysis hyperkalemia and allow for the use of higher dialysate potassium levels could be an important advance in reducing dialysis-induced arrhythmias.

Disclosures

P.H.P. and J.P.M. have served on advisory boards for Relypsa, Inc. J.P.M. has received research support from Janssen and ZS Pharma and served on advisory boards for Astra Zeneca.