Management of severe hyperkalemia in the post-Kayexalate era

Introduction

There is increasing recognition that sodium polystyrene sulfonate (Kayexalate) is ineffective for the immediate management of severe hyperkalemia (Kamel 2012). With Kayexalate gone, there seems to be a gap in our treatment regimen. I often encounter residents who know that Kayexalate isn't helpful, but aren't sure exactly how to treat hyperkalemia without it.

The good news is that abandoning Kayexalate allows us to focus on a more effective approach to hyperkalemia: renal potassium excretion (kaliuresis). Anyone experienced in diuresis knows that it causes a drop in the potassium level, at times requiring frequent monitoring and aggressive potassium repletion. It's time to use this to our advantage.

PART 1: The Bicarbonate Debate

Theory: Three mechanisms whereby bicarbonate could decrease serum potassium

Mechanism #1: Transcellular shift into skeletal muscle Most of the body's potassium is located in skeletal muscle cells, so small shifts of potassium between the serum and muscle cells can strongly affect the serum potassium. Sodium bicarbonate may cause shifting of potassium into muscle cells via various mechanisms. By alkalinizing the serum, bicarbonate may indirectly cause movement of potassium into cells via an H+/K+ exchange mechanism (figure below). Additionally, bicarbonate may be directly transported into muscle cells along with potassium (Aronson 2011).

0
Mechanism #2: Renal Excretion Acute metabolic acidosis impairs potassium excretion by the kidney, whereas metabolic alkalosis facilitates potassium excretion. This is largely due to regulation of potassium channels in the distal nephron, which are down-regulated by acidosis and up-regulated by alkalosis. Metabolic alkalosis additionally inhibits proximal tubule reabsorption of sodium bicarbonate, which facilitates potassium excretion by increasing distal sodium concentration and flow rate (Aronson 2011):

Potassium secretion as measured in a rat nephron micropuncture model, in response to luminal fluid of varying composition (Amorim 2003). Alkalosis caused a slight increase in potassium secretion. Increasing the luminal bicarbonate concentration at a fixed pH of 7.0 caused a greater increase in potassium secretion. Thus, potassium secretion may be independently stimulated by both alkalosis and also by increased sodium bicarbonate excretion in the urine.

Mechanism #3: Dilution If a large volume of isotonic bicarbonate is given, there may be a decrease in potassium concentration due to a dilutional effect. Consider for example a hypothetical 70-kg man with a potassium of 8 mM and an extracellular fluid volume of 15 liters. Temporarily ignoring any effect of potassium shifts, infusion of two liters of isotonic bicarbonate would be expected to decrease his potassium to 7.1 mM simply by expanding his extracellular fluid volume to 17 liters.

Hypertonic bicarbonate appears ineffective

Traditional management of hyperkalemia has involved using ampules of hypertonic 8.5% sodium bicarbonate (which has an osmolality of 2000 mOsm, about seven times higher than plasma). Unfortunately, hypertonic bicarbonate has been uniformly ineffective in multiple studies (Blumberg 1988, Blumberg 1992, Kim 1996, Kim 1997).

Hypertonic fluids are known to increase serum potassium levels due shifting potassium out of cells (Aronson 2011, Conte 1990). One of the mechanisms explaining this is a phenomenon called solute drag. Increasing the plasma tonicity causes cells to shrink, which increases the intracellular potassium concentration. Equilibration with serum then causes potassium to leave the cells (thereby being “dragged” out of the cells following water).

Precisely why hypertonic bicarbonate fails to work is unclear. It is likely that its hypertonic effects negate any benefit of the bicarbonate. It could also be that evaluating the immediate effect of hypertonic bicarbonate relies solely on mechanism #1 above (thus failing to take advantage of mechanisms #2-3).

0
Bicarbonate appears ineffective in the absence of acidosis

Bicarbonate also appears to be ineffective in patients without significant pre-existing metabolic acidosis. Blumberg 1988 and Allon 1996 found that even isotonic bicarbonate was ineffective among patients undergoing chronic hemodialysis with an average initial bicarbonate of 22 mEq/L. There are various possible explanations for this. The Na/H exchange channel in skeletal muscle may be up-regulated by acidosis and down-regulated by alkalosis (figure below). HCO3-/K+ cotransport is partially driven by intracellular acidosis which creates a gradient favoring bicarbonate entry into cells, so this mechanism may also be less effective in the absence of acidosis (Aronson 2011).

Isotonic bicarbonate may be effective in patients with metabolic acidosis

Available evidence suggests that large-volume isotonic bicarbonate infusions may benefit patients with pre-existing metabolic acidosis (2). In 1977, Fraley evaluated the effect of infusing one liter of D5W containing 89 mM or 134 mM sodium bicarbonate over 4-6 hours to 18 hyperkalemic patients with baseline metabolic acidosis. Patients with persistent hyperkalemia were treated with additional bicarbonate. As a control, some patients were initially treated with D5W alone (which was ineffective). There was a linear relationship between the decrease in serum potassium and the increase in serum bicarbonate, with serum potassium decreasing by about 0.15 mM for every 1 mEq/L increase in bicarbonate. Patients were retrospectively divided into two groups depending on whether or not serum pH increased during bicarbonate therapy. Both groups demonstrated a decrease in potassium with bicarbonate therapy (figure below). There was no relationship between the renal potassium excretion and the change in serum potassium, suggesting that renal excretion was not primarily responsible. This study has many flaws, including the use of varying concentrations and volumes of serum bicarbonate.
0

Relationship between changes in blood bicarbonate and serum potassium among five patients with increasing serum pH (left) and nine patients with stable pH (right; Fraley 1977). Multiple data points are present for each patient, representing different time points during the bicarbonate infusion. There was no statistical difference between regression lines relating these values in the two patient groups (both regression lines are shown on the right).

In 1991, Gutierrez evaluated the effect of isotonic bicarbonate, hypertonic bicarbonate, normal saline, or hypertonic saline among patients with chronic renal failure and metabolic acidosis (figure below). At a dose of 1 mEq/kg, hypertonic bicarbonate had no effect, whereas isotonic bicarbonate caused an average decrease in serum potassium of 0.35 mM (p<0.05). Although some have interpreted this data to indicate that bicarbonate is ineffective, 1 mEq/kg is a lower dose of isotonic bicarbonate than other investigators used. The 0.35 mM decrease in serum potassium observed here in parallel with a 3 mEq/L increase in serum bicarbonate is consistent with results obtained by Fraley (above) and Blumberg 1992 (below). Note that saline tended to increase the potassium level – this is discussed further below.

In 1992 Blumberg evaluated the effect of bicarbonate in 12 patients with end-stage renal disease and metabolic acidosis on chronic hemodialysis. Patients first received 240 mM of 8.4% bicarbonate over an hour (equal to about five ampules of bicarbonate). This was followed by an infusion of 900ml of isotonic bicarbonate over the next five hours. Hypertonic bicarbonate had little effect on the serum potassium over the first hour. However, the infusion of isotonic bicarbonate over the next five hours did seem to decrease the serum potassium (figure below). These authors calculated that about half of this decrease in potassium may have been due to a dilution effect by expanding the extracellular fluid volume (Mechanism #3 above). Unfortunately this study is flawed because it is unclear whether the decrease in potassium was due to the isotonic bicarbonate infusion or a delayed effect of the hypertonic bicarbonate.

Conclusions about bicarbonate?

0
Ultimately the literature regarding bicarbonate remains unsatisfying. All studies above, aside from Fraley 1977, investigated patients with chronic end-stage renal disease and moderate hyperkalemia attending routine hemodialysis. Results from this patient population may not be generalizable to patients presenting with acute life-threatening hyperkalemia who often have more severe acidosis and acute renal failure. For example, it is possible that patients undergoing chronic hemodialysis could have chronically elevated intracellular potassium levels, and thus be less able to shift additional potassium intracellularly. Indeed, end-stage renal disease is known to impair extra-renal potassium metabolism in numerous ways, including impaired function of Na-K channels (Ahmed 2001).
0
Overall it is impossible to reach any definite conclusion based on existing evidence. Theoretical and experimental evidence suggest that isotonic bicarbonate may be beneficial among patients with metabolic acidosis. Potassium might decrease by roughly 0.15 mM for every 1 mM increase in bicarbonate, suggesting that a large volume of isotonic bicarbonate may be required (e.g., a sufficient volume to increase serum bicarbonate levels by 5-10 mM, roughly 1-2 liters)(1). This cannot be done in a patient with volume overload. The ideal candidate for bicarbonate therapy would be a patient with volume depletion, hyperkalemia, and metabolic acidosis, because isotonic bicarbonate may improve all three of these problems simultaneously.

PART 2: Avoid normal saline
0

What about volume resuscitation of a patient with hyperkalemia who doesn't have metabolic acidosis? Although normal saline is traditionally used in this situation, it has been proven in three randomized controlled trials to induce a hyperchloremic metabolic acidosis and worsen hyperkalemia (not including Gutierrez 1991 discussed above; evidence explored here). In contrast, Lactated Ringers is safe to use in hyperkalemic renal failure and is proven to cause less hyperkalemia than normal saline. Of everything discussed in this post, the danger of normal saline is supported by the strongest evidence (three independent prospective double-blind RCTs).

PART 3: Diuresis vs. Dialysis

Previously it was believed that there were three routes to emergently remove potassium from the body: stool (using Kayexalate), urine (kaliuresis), and dialysis. Removal of Kayexalate from the treatment algorithm simplifies matters and allows us to focus on kaliuresis, which can be extremely effective and is often under-utilized. For example, a recent review article on hyperkalemia failed to mention diuresis at all (Elliott 2010).

For a patient with life-threatening hyperkalemia, it is often reasonable to make a single attempt at kaliuresis prior to proceeding to dialysis or simultaneously to pursuing dialysis (e.g., while arranging transfer to a hospital with dialysis capabilities). Of course in some situations such as chronic anuric renal failure, kaliuresis is unlikely to succeed, so it may be more sensible to proceed immediately to dialysis.
0

0
Most diuretics cause potassium loss in the urine. A loop diuretic (e.g., furosemide) is the most potent agent, and is generally used as the backbone of the diuretic regimen. For patients with life-threatening hyperkalemia and renal insufficiency, it may be reasonable to use multiple diuretics, as these will operate in a synergistic fashion by blocking potassium reabsorption at different sites in the nephron (figure above). The combination of a loop diuretic and thiazide is commonly used in diuretic-resistant patients, with increased efficacy and potassium loss (Jentzer 2010). Acetazolamide may be especially kaliuretic because it increases bicarbonate delivery to the distal nephron (Weisberg 2008, Goodman & Gillman 12e Chapter 25).

There is no evidence regarding the number or dose of diuretic which should be used. For life-threatening hyperkalemia there is generally time for a single attempt at kaliuresis. Therefore, typically this attempt is fairly aggressive. For a patient with renal dysfunction who is expected to respond poorly, high doses of multiple agents may be considered (e.g., intravenous furosemide plus intravenous chlorothiazide)(3). The risk of over-diuresis and electrolyte depletion may be minimized with close monitoring of electrolytes and repletion as needed. Urine output and volume status must be carefully monitored, with ongoing volume administration to return urinary losses and maintain a euvolemic state.

In the absence of evidence, selection of the number and dosage of diuretics must be based on clinical judgement. For example, at Genius General we once admitted a pleasant elderly man with chronic renal failure complicated by hyperkalemia causing bradycardia and shock. He wished never to undergo dialysis and was not amenable to this therapy even temporarily. Given probable death if he failed to respond promptly to diuretics, he was treated with maximal kaliuresis (200 mg i.v. furosemide, 500 mg i.v. acetazolamide, 1000 mg i.v. chlorothiazide, and isotonic bicarbonate). He responded well, and ultimately required potassium and fluid repletion. In retrospect, he likely would have responded to a less aggressive diuretic regimen. However, in the face of life-threatening hyperkalemia, it may be safer to err on the side of over-treatment followed by meticulous replacement of electrolytes and fluid as needed.

Neither Kayexalate nor hypertonic bicarbonate (i.e., ampules of 8.4% bicarbonate) are effective for emergent treatment of hyperkalemia.
Isotonic bicarbonate may be effective for patients with metabolic acidosis. Unfortunately this requires a large volume of fluid, and cannot be used in patients with volume overload.
Normal saline is proven to worsen hyperkalemia and should be avoided. For a hypovolemic patient without metabolic acidosis, lactated ringers is a reasonable fluid choice.
Kaliuresis (facilitating urinary potassium excretion with diuretics) may be quite effective in patients with residual renal function. Otherwise, emergent dialysis is generally needed.

Additional resources

Podcast by Scott Weingart about treatment of severe hyperkalemia from 2010.
Review article by Weisberg regarding the management of severe hyperkalemia. Although this article is now seven years old, it remains one of the best reviews out there.
Is Kayexalate effective? This has been discussed in EMLyceum, EMCrit, Precious Bodily Fluids, and Kamel 2012. There's not much I can add to this discussion that hasn't already been said, so if you're interested in the Kayexalate issue please see these sources.
Prior post on pH-guided resuscitation describes the rationale for choosing different fluids during resuscitation in order to optimize the final acid-base status.
The effects of pH on renal handling of potassium is reviewed by Aronson 2011. This is a very detailed article with lots of information about various potassium channels.

Notes

(1) The volume of isotonic bicarbonate required may be estimated by calculating a patient's bicarbonate deficit using MDCalc and then dividing by 150 mEq to calculate the number of liters of isotonic bicarbonate this equals. For example, a 70-kg man with a bicarbonate of 15 mEq/L has a bicarbonate deficit of 252 mEq. Given that every liter of isotonic bicarbonate contains 150 mEq of bicarbonate, this deficit correlates to roughly 1.7 liters of isotonic bicarbonate. Therefore, for this patient 1.7 liters of isotonic bicarbonate would be expected to increase his bicarbonate from 15 mEq/L to 24 mEq/L, an increase of 9 mEq/L which would be expected to decrease potassium by roughly 9 mEq/L x 0.15 = 1.35 mM. These are all very rough calculations, but may provide a general concept about how much bicarbonate is required.

(2) Isotonic bicarbonate contains 150 mEq/L of sodium bicarbonate. This is commonly obtained by adding 3 ampules of sodium bicarbonate (containing 50 mM/ampule) to a liter of D5W. Further discussion of isotonic bicarbonate may be found on a prior post regarding pH-guided resuscitation.

(3) I'm not aware of any direct evidence upon which to base this selection. Theoretically, acetazolamide may be expected to be more kaliuretic than a thiazide diuretic. However, acetazolamide overall may be a less powerful agent, and less effective at eliciting diuresis in a patient with renal dysfunction. A common practice of nephrologists and intensivists at Genius General Hospital has been to combine intravenous furosemide and chlorothiazide, and this seems to be effective.

Image credits:

Image of bicarbonate ampule: http://dailymed.nlm.nih.gov/dailymed/fda/fdaDrugXsl.cfm?setid=c1ab9fff-c97b-4fca-b7a2-2378045bc799&type=display
Diagram of nephron: http://www.boomer.org/c/p2/Exam/Exam9905/Exam9905-1.html