Dialysate Composition for Hemodialysis: Changes and Changing Risk

Abstract

Dialysate composition is a critical aspect of the hemodialysis prescription. Despite this, trial data are almost entirely lacking to help guide the optimal dialysate composition. Often, the concentrations of key components are chosen intuitively, and dialysate composition may be determined by default based on dialysate manufacturer specifications or hemodialysis facility practices. In this review, we examine the current epidemiological evidence guiding selection of dialysate bicarbonate, calcium, magnesium and potassium, and identify unresolved issues for which pragmatic clinical trials are needed.

Over the past 60 years, hemodialysis has evolved from a rare treatment modality used for otherwise healthy people with acute kidney failure to a maintenance treatment for patients with a wide range of comorbid conditions, with the anticipation of survival for years or even decades. While survival has improved in recent years for dialysis patients¹, optimal treatment strategies remain uncertain, often guided by guesswork and gestalt. Ironically, although the term dialysis specifically refers to the movement of substances by diffusion, little evidence is available to guide the optimal concentration of substances in the dialysate, leaving dialysate composition guided by the intuitively reasonable goal of normalizing serum chemistries. In steady state, this goal appears appropriate; however, dialysis patients are never at steady state, indicating that, to operationalize this principle, a specific time point needs to be selected for when electrolytes are normalized. For many reasons, most notably convenience, the chosen time point is most often at the start of a dialysis session, following a 48 to 72 hour period during which considerable metabolic changes likely have occurred. However, accumulating evidence suggests that selection of dialysate prescriptions by these criteria may be suboptimal or possibly even harmful, as rapid solute movement may cause dramatic fluxes in solute concentrations, and these rapid shifts may be associated with adverse outcomes.

In this article, we review the relationship between the dialysate composition and patient outcomes, focusing specifically on dialysate potassium, calcium, magnesium and bicarbonate in patients treated with thrice weekly maintenance hemodialysis. Dialysate sodium is the focus of a different article in this issue of Seminars in Dialysis².

Dialysate Bicarbonate and Metabolic Acidosis

Metabolic acidosis is a frequent complication of kidney disease due to retention of acid phosphates and sulfates and decreased generation of ammonia. Plasma bicarbonate often falls below 20 mEq/L in patients with untreated chronic kidney disease (CKD), despite maximal intracellular and bone buffering³. Acidosis in CKD has been shown to have deleterious effects upon bone health, nutritional parameters, and protein metabolism^4–5; causes chronic ventilatory compensation; and is associated with malaise^4,6. Correction of metabolic acidosis therefore is a goal of dialysis therapy, and the 2000 NKF-KDOQI (National Kidney Foundation-Kidney Disease Outcomes Quality Initiative) guidelines suggested that therapy should be adjusted to maintain serum bicarbonate levels at 22 mEq/L or greater⁴, a recommendation that resulted in a gradual increase in dialysate bicarbonate concentration over time.

Acetate was the primary buffer in dialysate when maintenance hemodialysis originated in 1960, because of convenient physicochemical properties and lower cost. Recognition of acetate-induced hypotension led to widespread use of bicarbonate dialysate⁷, aided by the introduction of three-stream proportioning mechanisms that could assemble dialysate from purified water, bicarbonate solution and an ‘acid’ concentrate that contained potassium and other ions, including a small amount of acetate that acted as a buffer⁸. Three-stream proportioning systems prevented formation of precipitates by ensuring that dilution was sufficient before the mixing of concentrates occurred⁸. An additional increase in dialysate total alkali occurred when dry dialysate concentrates containing sodium diacetate were introduced in 2003. Substitution with a dry concentrate results in up to 8 mEq/L of alkali above and beyond the listed bicarbonate concentration of the dialysate, whereas liquid acetic acid and acetate preparations most often provided approximately 4 mEq/L. The contribution of the ‘acid’ concentrate to the total alkali exposure of the patient, due to hepatic conversion of acetate to bicarbonate, is an underappreciated but important aspect of dialysate prescribing⁹.

Metabolic alkalosis induced by high dialysate base concentrations could have important health consequences, including respiratory depression, reduced cerebral blood flow, neuromuscular excitability, and increased calcium flux from the blood that could predispose to arrhythmia and hypotension and might also promote metastatic calcification¹⁰. Furthermore, increased alkali exposure may result in a high dialysate to serum gradient for bicarbonate that contributes to the arrhythmogenic risks of hemodialysis by promoting rapid movement of potassium into cells¹¹. In addition, very high bicarbonate levels in dialysate force the dialysate proportioning equipment to create dialysate with a reduced potassium concentration¹². Safety concerns associated with dry dialysate acid concentrate resulted in FDA mandated additional safety labeling in 2012¹³.

Bicarbonate and Clinical Outcomes

No adequately powered clinical trials examine the associations among different dialysate alkali concentrations and outcomes, resulting in reliance on observational data (Table 1). In early studies, a J-shaped relationship between mortality and serum bicarbonate was observed^14–15. In two very large studies from a large dialysis provider database, the association of serum bicarbonate with outcomes was dependent on protein-energy wasting and inflammation, reflecting that patients with better nutritional status tend to have lower serum bicarbonate levels. Before adjustment for clinical characteristics, nutrition and inflammation status, mortality risk was lowest among patients with pre-dialysis bicarbonate levels between 17 and 23 mEq/L, with progressively higher mortality risk seen with serum bicarbonate levels of 23 mEq/L and above. After adjustment for characteristics that reflect protein-energy wasting and inflammation, thereby likely accounting for additional acid generation in patients with better nutritional status, the association between higher bicarbonate levels and increased death risk was attenuated while lower serum bicarbonate level became a stronger predictor of mortality risk.^16–17

Table 1.

Epidemiologic studies of metabolic acidosis and dialysate bicarbonate

Study	N	Population	Outcomes	Major Results	Strengths	Limitations
Lowrie14, 1990	19,746	National Medical Care	All-cause death	↑risk: serum HCO3 >22.5, < 17.5	Large US national provider database	No adjustments for nutritional or inflammatory factors
Bommer15, 2004	7,140	DOPPS Phase 1	All-cause death, hospitalization	↑risk: serum HCO3 >27, < 17	Multinational research database Adjusted for nPCR, Kt/V	‘Correction’ of values to approximate midweek values; no inflammatory factors
Wu16, 2005	56,385	DaVita	All-cause death, CV death	↑risk with ↑ serum HCO3≥23, but association reversed after adjustment for MICS-related lab panel	Large US national provider database	Lack of direct inflammatory markers
Vashistha17, 2013	110,951	DaVita	All-cause death	↑risk with ↑serum HCO3 <22, with adjustment for MICS	Large US national provider database	Assays on shipped samples
Tentori10, 2013	17,031	DOPPS Phases 2–4	All-cause death, CV death, hospitalization, infection death	↑ACM, IM, and ACH but not CVM with ↑HCO3 in dialysate. No association with pre-HD serum HCO3	Multinational research database	Exclusion of Japanese data; no direct inflammatory markers
Yamamoto18, 2015	15,132	Renal Data Registry, Japanese Society of Dialysis	All-cause death, CV death	↑pre-HD pH associated with increased ACM, CVM, but no association with pre or post HD HCO3	Japanese national registry	Exclusion of all catheter patients and all patients in first year of treatment

CV, cardiovascular; MICS, malnutrition-inflammation complex syndrome; DOPPS, Dialysis Outcomes and Practice Patterns Study

In a more recent report of patients from eleven countries participating in phases 2–4 of the Dialysis Outcomes and Practice Patterns Study (DOPPS), dialysate bicarbonate concentration was prospectively determined in 17,031 adults undergoing hemodialysis¹⁰. Separate assessments of the alkali contributions of the ‘acid’ concentrates were not made. Cox regression was used to define the associations between dialysate bicarbonate and mortality, stratifying by country and by DOPPS study phase. Higher all-cause mortality was seen with higher dialysate bicarbonate concentration, although the 8% increase seen with each 4-mEq/L higher dialysate bicarbonate was driven almost entirely by a 40% increase in deaths due to infection, which was not the hypothesized cause of increased risk. Cardiovascular (CV) mortality and sudden death did not differ significantly, although all-cause and CV hospitalization were higher with higher dialysate bicarbonate¹⁰. Unlike previous studies^14–15, higher pre-dialysis serum bicarbonate did not have an association with cardiovascular mortality, and serum bicarbonate level was only weakly correlated with dialysate choice; this latter result likely reflects the impact of other factors on serum bicarbonate levels, such as dietary practices, residual kidney function, and respiratory status.

This report specifically excluded data from Japan, where dialysate bicarbonate levels are systematically lower than in all other DOPPS countries. In data from the renal registry of the Japanese Society of Dialysis Therapy, including measured pH and calculated serum bicarbonate levels from blood gas analyses of mixed arterial blood obtained from hemodialysis fistulas before and after treatments, there was an increased hazard of all-cause mortality with elevated pre-dialysis pH and with greater change in pH over the course of dialysis; however, there was no significant relationship between mortality and bicarbonate levels in either dialysate or blood¹⁸. Neither dialysate bicarbonate nor total dialysate alkali was associated with post-dialysis serum bicarbonate level. In an editorial, Gennari suggested that this finding indicated that the conversion of acetate and other organic anions to bicarbonate reduces the gradient for movement of bicarbonate from dialysate to plasma, and therefore does not influence the alkali exposure during dialysis¹⁹. The differences in practice between Japan and the remainder of the DOPPS cohort theoretically provide a natural experiment in dialysate choice, but this comparison was not included in the DOPPS analysis, in large part reflecting other substantial differences in dialysis practice and populations between Japan and other DOPPS countries.

Taken together, it is challenging to draw a unifying conclusion from these studies. Some evidence suggests that dialysate bicarbonate concentrations above 35 mEq/L may have unintended adverse consequences, with or without the contribution of other organic anions. Others studies suggest that the primary risks associated with high serum bicarbonate are related to nutritional and inflammatory factors, rather than to the dialysis prescription^19–20. This is plausible given the known effects of protein intake on serum total bicarbonate in CKD²⁰, with some authors suggesting that higher serum bicarbonate levels represent underlying burden of protein-energy wasting and chronic disease, that mild acidosis is the laboratory manifestation of higher protein intake, and therefore that the association between bicarbonate levels and mortality is an example of residual confounding²¹. This possibility was considered by the authors of the DOPPS report and rejected on the grounds that similar associations were seen from dialysis facilities in which dialysate choice was uniform and not adjusted for clinical indication¹⁰. A clinical trial would be necessary to determine whether interventions to manipulate the serum bicarbonate concentration have important clinical effects. Regardless, clinicians must be aware of the controversy surrounding the potential contributions of dry or liquid acid concentrates to the alkali exposure of their patients^18–20.

Dialysate Calcium

Interpretation of the serum or plasma calcium concentration requires an interpretation of the units of measurement, as well as an understanding of whether total or ionized calcium is being measured (Table 2). The calcium concentration in serum or plasma can be expressed as mg/dL, mmol/L or mEq/L. Measurements may sometimes refer to the total serum calcium concentration, or the physiologically relevant fraction (normally 45%) of total calcium that is ionized and not bound to either albumin or anions such as sulfate, phosphate, lactate, and citrate.

Table 2.

Typical normal ranges for total and ionized calcium

	Total calcium	Ionized calcium
mg/dL	8.8–10.3	4.4–5.4
mmol/L	2.2–2.6	1.1–1.35
mEq/L	4.4–5.2	2.2–2.7

Dialysate calcium prescriptions have evolved over time, starting from an intuitively reasonable initial concentration of 2.5 mEq/L, which is similar to a normal serum ionized calcium level²². Dialysate calcium levels gradually increased to 3.0–3.5 mEq/L during the 1970’s in response to widespread hypocalcemia, hyperparathyroidism, and bone disease among dialysis patients. In the 1980’s, calcium supplanted aluminum as the phosphate binder of choice and the use of vitamin D analogues to treat secondary hyperparathyroidism became widespread^23–24. Hypercalcemia was frequently encountered, which drove dialysate calcium concentrations back downwards.

The trend back to 2.5 mEq/L calcium baths or even lower calcium concentrations may have been further motivated by the 2009 Kidney Disease Improving Global Outcomes (KDIGO) Clinical Practice Guideline for the Diagnosis, Evaluation, and Treatment of CKD–Mineral and Bone Disorder (CKD-MBD), which stated that ‘In patients with CKD stages 3–5D, we suggest maintaining serum calcium in the normal range (2D)’²⁵. This suggestion, based entirely on observational data, was subsequently reinforced by inclusion of hypercalcemia as a quality metric in the Centers for Medicare and Medicaid Services (CMS) Quality Incentive Program (QIP), creating the risk of financial penalties for dialysis facilities with patients who have serum calcium levels above 10.2 mg/dL²⁶. The QIP metric therefore provided a substantial motivation to reduce calcium in dialysate, despite the similarly evidence-poor statement in the KDIGO guideline that: ‘In patients with CKD stage 5D, we suggest using a dialysate calcium concentration of 1.25–1.50 mmol/L (2.5–3.0 mEq/L) (2D)’²⁵.

Concern for hypercalcemia arises from the intuitive but largely unproven hypothesis that positive net calcium balance predisposes towards vascular calcification in kidney disease, and that this in turn contributes to the burden of cardiovascular disease in dialysis patients. The vascular calcification hypothesis has led to increasing use of phosphate binders that do not contain calcium, like sevelamer and lanthanum, and expanded use of calcimimetics, such as cinacalcet^27–29. These shifts in medication practices have not been associated with any widespread changes in dialysate calcium prescriptions. In many facilities, dialysate calcium concentrations are below 2.5 mEq/L, which likely results in negative calcium balance. Hypocalcemia caused by negative calcium balance stimulates parathyroid hormone, which can promote vascular calcification by mobilizing skeletal calcium into the bloodstream and is precisely the opposite of the intended effect. Importantly though, lower dialysate calcium may improve bone turnover in patients with adynamic bone disease^30–32.

Lower dialysate calcium may result in additional risks (Table 3)^33–38. Hypocalcemia has been associated with QTc interval prolongation^34–37, myocardial ischemia and stunning²², and hypotension due to decreased vascular resistance and lower cardiac output^22,39–40. In a case control study of patients enrolled in a large dialysis organization database, 510 individuals who had suffered sudden cardiac arrest were compared to 1560 controls matched for age, dialysis vintage, and calendar year. After adjusting for demographic and clinical variables, as well as medications that could affect the QTc interval, patients with a dialysate calcium concentration of < 2.5 mEq/L had a two-fold greater hazard of cardiac arrest; these associations were also observed with lower serum calcium levels and increased serum-to-dialysate calcium gradients³⁹.

Table 3.

Epidemiologic studies evaluating dialysate calcium and cardiovascular outcomes

Study	N	Population	Outcomes	Major Results	Strengths	Limitations
Nappi33, 1999	12	Single-center	Echocardiography	Impaired ventricular relaxation with D-Ca 1.75 mM compared to 1.50 mM or 1.25 mM	Pre/post HD echocardiograms measurements; D-Ca only variable	Small study
Nappi34, 2000	23	Single-center	QTc dispersion	Increased QTc dispersion with D-Ca 1.25 mM compared to 1.50 mM or 1.75 mM	Pre/post HD examination; D-Ca only variable	Small study
Severi35, 2008	23	Single-center	QTc dispersion	Increased QTc dispersion with D-Ca 1.25 mM vs 2.00 mM	Pre/post HD with electrical modeling	Small study
Genovesi36, 2008	16	Single-center	Holter monitor	Increased QTc with D-Ca 1.25 mM compared to 1.50 mM or 1.75 mM	Randomized design	Variation of both calcium and potassium
Di Iorio37, 2012	22	Single-center pilot study	Hourly ECG	Prolongation of QTc with low D-Ca/low K/high bicarbonate	Randomized controlled crossover with blinding	Simultaneous variation of calcium, potassium and bicarbonate
Pun39, 2013	2,070	DaVita	Sudden cardiac arrest	OR 2.00 for D-Ca < 2.5 mEq/L OR 1.40 per 1 mEq/L of serum-to-dialysate gradient	Large provider database	Total calcium only; Case-control design
Brunelli40, 2015	353 (facilities)	DaVita	Death, CV death, CHF Hosp, afib, MACE, fracture, Hypocalcemia, HD hypotension	↑ CHF hosp, hypocalcemia, and HD hypotension with lower D-Ca. No changes in death or CV death	Large provider database	Facility-level data

D-Ca, Dialysate calcium; ECG, electrocardiogram; HD, hemodialysis; OR, odds ratio; CV, cardiovascular; MACE, major adverse cardiac event (MACE); CHF Hosp, congestive heart failure hospitalization

A more recent retrospective investigation using the same database evaluated the risk of sudden cardiac arrest in 274 facilities that had predominant use of dialysate with 2.5 mEq/L calcium concentrations, compared to 79 facilities that had converted to lower dialysate calcium concentrations. Facilities with lower calcium dialysate concentrations had more hypocalcemia, more intradialytic hypotension, and more hospitalization for congestive heart failure⁴⁰.

Neither of these studies evaluated the effect of cinacalcet, which simultaneously reduces parathyroid hormone (PTH), fibroblast growth factor-23 (FGF-23) and serum calcium. In a secondary analysis of the EVOLVE trial, a large, placebo-controlled, randomized trial of cinacalcet that did not demonstrate significant reductions in cardiovascular mortality or a composite cardiovascular endpoint in hemodialysis patients with secondary hyperparathyroidism, cinacalcet was shown to reduce serum FGF-23 levels. In this post-hoc analysis, decreases in FGF-23 with cinacalcet, but not with placebo, were associated with lower rates of cardiovascular mortality, heart failure, and sudden death²⁹. Given the null overall relationship between randomization to cinacalcet and cardiovascular disease outcomes in EVOLVE, one could infer that randomization to cinacalcet with a lack of a subsequent reduction in FGF-23 may be associated with increased risk of adverse outcomes as compared to randomization to placebo. This hypothesis was not formally tested. The increasing use of cinacalcet in patients on hemodialysis could potentially modify the relationship between dialysate calcium, serum calcium, and cardiovascular endpoints. Evidence from the PARADIGM trial suggests that the effect of cinacalcet on parathyroid hormone levels is diminished when dialysate calcium is < 2.5 mEq/L⁴¹.

Taken together, the evidence suggests that dialysate calcium levels of < 2.5 mEq/L may have unintended adverse cardiovascular consequences and should be avoided in the setting of non-calcium containing binders. Cardiovascular risks appear associated with both hypercalcemia and hypocalcemia, and the risks of the latter need to be considered despite the regulatory emphasis on the former (Table 4). For the time being, dialysate calcium levels should be adapted individually, albeit likely with caution and within a tight threshold, for avoidance of severe hypocalcemia and severe hypercalcemia, with the hypothesis that this may result in a lower risk of cardiovascular adverse events.

Table 4.

Potential risks and benefits associated with dialysate calcium choices

	Higher Dialysate Calcium	Lower Dialysate Calcium
Potential Benefits	Improved hemodynamic stability Reduced acute arrhythmia potential Lower PTH levels May facilitate cinacalcet use	Reduced substrate for vascular calcification May help overcome adynamic bone disease
Potential Risks	Increased vascular calcification, particularly in the presence of other calcification promoters Suppression of PTH and development of adynamic bone disease	Increased PTH levels Requirement for higher vitamin D analogue doses with subsequent increased phosphorus Increased intradialytic hypotension

PTH, parathyroid hormone

Dialysate Magnesium

Very little attention has been given to factors affecting the selection of dialysate magnesium concentrations, which are often determined by the facility and the dialysate manufacturer rather than by individual prescription. Magnesium is a cofactor in more than 300 human enzymatic reactions⁴¹, and the secondary consequences of abnormal magnesium levels remain to be defined.

Hypomagnesemia is associated with cardiovascular mortality in the general population^42–43, as well as patients with chronic kidney disease^{44, 46}, including those treated with dialysis^47–49, although no trials exist showing improved outcomes with correction of magnesium levels to the normal range. The potential for hypomagnesemia, and associated hypocalcemia and hypokalemia, to promote arrhythmias exists, and low magnesium concentration may prove to be a non-traditional risk factor for cardiovascular mortality. Hypomagnesemia also has longstanding associations with intradialytic hypotension, impaired endothelial function, insulin resistance, and inflammation^50–51.

Magnesium supplementation has been shown to improve coronary arterial calcification scores^52–54 and reduce carotid intima-media thickness^{51, 55}. High magnesium levels activate the calcium-sensing receptor in the parathyroid gland and suppress parathyroid hormone. Magnesium also can be used as a phosphate binder⁵³. Despite these potential advantages, magnesium is rarely given to dialysis patients as supplements, laxatives, or phosphate binders. While there is potential for increased retention of magnesium in patients with kidney failure, the proportion of patients with severe hypermagnesemia is very low. Impaired nutrition can reduce levels of both magnesium and albumin in patients on dialysis. When albumin levels are low, the proportion of free serum magnesium increases, and the removal of magnesium during dialysis is facilitated.

Magnesium concentrations in dialysate have gradually decreased over time, possibly due to early reports raising concerns that hypermagnesemia was a contributory factor in osteomalacia, or due to concerns about hypermagnesemia due to kidney failure⁴². Unlike calcium, magnesium levels are not regulated by any known hormone, so a reduction in dialysate magnesium levels from early values of 1.5 mEq/L to current levels of 0.5–1.0 mEq/L would be expected to result in net removal of magnesium from the extracellular space. Although the dialyzable fraction is <1% of total body magnesium, as magnesium primarily is located within cells, the cumulative effects of small magnesium losses over time may be magnified by a shift towards lower total dietary uptake of magnesium⁴². Proton-pump inhibitors, which are associated with hypomagnesemia in both the general population and in hemodialysis patients, have become widely used^56–59. The net effect of these factors is that hypomagnesemia occurs at least 5% of the time when the dialysate magnesium concentration is 1.0 mEq/L, and up to 33% of the time when dialysate magnesium concentration is 0.5 mEq/L^42,53.

Two recent large observational studies showed an association between lower serum magnesium levels and mortality^48–49. A Japanese registry of 142,555 hemodialysis patients for whom magnesium levels were available showed nearly linear trends for all-cause, cardiovascular, and non-cardiovascular mortality up until a serum level of ≥ 3.1 mg/dL⁴⁸. A study of nearly 50,000 patients in a large US dialysis provider database similarly showed a nearly linear relationship between serum magnesium and one-year mortality, although hazard ratios attenuated with adjustment for case-mix and laboratory variables⁴⁹.

Taken together, the evidence suggests risk of ongoing magnesium depletion when conventional dialysates are used, especially in patients with poor nutrition and those taking proton-pump inhibitors. Clinicians should be aware of this possibility and consider avoiding low magnesium dialysates, monitoring for hypomagnesemia, and treating with adjustments in dialysate or oral supplements.

Dialysate Potassium

The commonality underlying many of the risks associated with dialysate prescription is an effect on serum potassium levels, as derangements in serum potassium in either direction are associated with cardiac arrhythmias. Potassium levels in hemodialysis patients are affected by many factors aside from dialysate potassium levels (Table 5); accordingly, when considering dialysate potassium, clinicians must consider the interplay among all of the components of dialysate.

Table 5.

Epidemiologic studies of dialysate potassium

Study	N	Population	Outcomes	Major Results	Strengths	Limitations
Karnik61, 2001	5,744,708 (HD sessions)	Fresenius	Cardiac arrest during HD	↑risk with DK<=1.0	Large population	No adjustment for comorbid conditions, or medications
Bleyer62, 2006	80	Five linked dialysis programs	Sudden death	↑risk with low SK	Standard definition of sudden death; chart review	No controls; incomplete ascertainment
Kovesdy64, 2007	81,013	DaVita	All-cause death, CV death	↑risk with SK<4 or >5.5	Large population; adjustment for MICS	Potential confounding by indication and by comorbid conditions
Pun63, 2010	2,134	DaVita	Sudden cardiac death	↑risk with DK<2	Adjustment for comorbid conditions and medications	Retrospective; limited to in-center deaths
Jadoul65, 2012	37,765	DOPPS	All-cause death, sudden death	↑risk with DK<1.5 or DK 2.0–2.5	Large population; adjustment for medications	Observational data;
Karaboyas66, 2017	55,183	DOPPS	all-cause death, arrhythmia	DK 2.0 vs. 3.0 equivalent in risk DK not linked to SK	Large population	Observational data with missingness imputed; prescribed DK, not actual

HD, hemodialysis; DK, dialysate potassium; SK, serum potassium; CV, cardiovascular; DOPPS, Dialysis Outcomes and Practice Patterns Study

Dialysate bicarbonate affects the rate at which serum potassium levels fall by promoting shifts of potassium into the intracellular compartment¹¹. This may be advantageous when serum potassium is dangerously high, but may have unintended adverse consequences at other times. In addition, prescription of high dialysate bicarbonate may result in an unintentional decrease in dialysate potassium, due to the constraints of conventional three-stream dialysate proportioning systems¹². Increasing dialysate calcium, and therefore serum calcium levels, may exert their beneficial effect on cardiac arrhythmias analogous to the way that administration of intravenous calcium stabilizes cardiac membranes - by restoring the resting membrane potential. Low dialysate magnesium may exert its effects in part via induction of transient or persistent hypokalemia and hypocalcemia, as well as hypomagnesemia. Magnesium activates the Na-K ATPase pump and increases intracellular potassium, which is crucial for maintaining the resting membrane potential. Membrane depolarization due to hypokalemia promoted by hypomagnesemia has been shown to predispose to ventricular arrhythmias⁶⁰.

Many drugs commonly administered to dialysis patients may cause alterations in serum potassium levels. Digitalis competes with potassium for binding sites on the Na-K ATPase pump, so toxicity is potentiated in the setting of hypokalemia, but serum potassium levels rise because binding of drug prevents potassium from entering cells. Dialysis patients often receive inhibitors of the renin-angiotensin-aldosterone system, which reduce intestinal potassium loss, predisposing to hyperkalemia even in patients without residual kidney function.

Independent of other prescribing practices, there is consistent evidence of increased risk with the prescription of low potassium dialysate (Table 5). In a case-control study comparing 400 hemodialysis treatments complicated by sudden cardiac death to nearly 6 million overall treatments in a large national dialysis provider’s database, treatments complicated by sudden cardiac death had nearly twice the odds of having had a prescribed dialysate potassium of 0 mEq/L or 1 mEq/L. These patients did not have significantly higher pre-dialysis serum potassium values⁶¹.

In a retrospective analysis of 88 sudden cardiac deaths occurring over an eight year period in a localized dialysis system, hypokalemia (defined as serum potassium<4.0 mEq/L) had occurred in 25% of patients who went on to have sudden cardiac death; given the heightened risk of death in the 12 hours following and 12 hours preceding dialysis treatments, the authors suggested that potassium fluxes and hypokalemia during hemodialysis may be complicit⁶².

Another retrospective study of sudden cardiac death in a large dialysis organization database compared 502 patients who suffered cardiac arrests with 1632 controls matched for age and dialysis-vintage. Compared to controls, there was a two-fold increased odds of having been exposed to dialysate potassium<2.0 mEq/L in patients with sudden cardiac arrest. There was a significant interaction between serum and dialysate potassium levels, in which the greatest risks of low potassium dialysate occurred in individuals with lower serum potassium, but there was no level of hyperkalemia at which a benefit of low potassium dialysate could be demonstrated.⁶³

This contrasts with another report of a cohort from the same large dialysis provider. In this study, higher mortality risk was most notably appreciated in patients with serum potassium of 5 mEq/L or higher dialyzed against 3 mEq/L or higher dialysate baths. The lowest risk of death was in patients with pre-dialysis potassium levels 4.6–5.3 mEq/L⁶⁴. In a 2012 analysis of DOPPS data, any dialysate potassium < 3 mEq/L was associated with increased risk⁶⁵. A more recent DOPPS publication compared dialysate potassium of 3 mEq/L to 2 mEq/L, and showed no significant differences in all-cause mortality or arrhythmia at any level of serum potassium; serum potassium was only minimally related to dialysate potassium⁶⁶.

These data highlight the difficulties that maintenance hemodialysis patients face, as both high and low levels of serum potassium carry potentially serious adverse consequences, as do rapid changes in potassium concentration. More frequent monitoring of serum potassium levels with appropriate adjustments of dialysate prescriptions, particularly in patients not at steady-state, could be a prudent response. The roles of dialysate potassium-modeling⁶⁷ and newer oral agents for the non-dialytic management of hyperkalemia remain to be determined. While there are promising recent data for the use of patiromer sorbitex calcium^68–69 and sodium zirconium cyclosilicate^70–71, currently available studies excluded patients treated with dialysis. That stated, there is no reason to suspect that these new potassium-binding agents will have different effects in people treated with dialysis than they have in those with advanced chronic kidney disease.

Conclusions

Clinicians prescribing hemodialysis currently rely largely upon observational data rather than clinical trials. This results in clinical decisions being made in the absence of strong data to guide treatment choices. Given the standardization of many components of dialysate in dialysis facilities, assessment of optimal dialysate composition in cluster-randomized pragmatic trials should be performed. Until then, we are guided by physiologic principles, observational data with their inherent biases, and gestalt.