Clinical study

The incidence and prevalence of permanent kidney failure have both increased steadily during the past decade ^[1] . At the end of 1997, there were >300 000 Americans with permanent renal failure, 62% of whom were receiving treatment with regular hemodialysis sessions ^[1] . The prevention of life-threatening hyperkalemia is one of the main goals of hemodialysis. Although potassium is removed effectively by dialysis, hyperkalemia can occur ^[2] and can sometimes be fatal ^[3] . Aside from dietary indiscretion, the only known risk factors for the development of hyperkalemia in maintenance hemodialysis patients are prolonged fasting ^[4] , hyperglycemia ^[5] , digoxin toxicity ^[6] , the use of β-blockers ^[7] , and the use of nonsteroidal anti-inflammatory drugs (NSAIDs) ^[8] .

Blockade of the renin-angiotensin system with angiotensin-converting enzyme (ACE) inhibitors ^[9] and angiotensin receptor blockers ^[10] can cause hyperkalemia in patients with renal insufficiency who are not on dialysis, presumably by inhibiting renal potassium excretion. Since hemodialysis patients often have minimal residual renal function, renin-angiotensin system blockade may have minimal effects on the serum potassium level. However, the risk of hyperkalemia from ACE inhibitors and angiotensin receptor blockers in hemodialysis patients is unknown.

Plasma aldosterone levels become elevated with advanced renal failure ^{[11] [12]} . This adaptive response increases the renal excretion of potassium; aldosterone may also affect gastrointestinal excretion and cellular uptake of potassium ^{[11] [12]} . In a study involving anephric dialysis patients, the disposal of an acute potassium load was enhanced by mineralocorticoid administration and impaired by spironolactone, an aldosterone antagonist ^[13] . The changes in the serum potassium concentration occurred without an increase in fecal potassium excretion, suggesting that aldosterone affected the cellular uptake of potassium ^[13] . Other studies have shown that intestinal potassium excretion is increased significantly in patients with renal failure ^[14] and that exogenous mineralocorticoids can increase the colonic excretion of potassium ^[15] . A recent study suggested that the enhanced colonic secretion of potassium in renal failure is mediated by an upregulation in angiotensin II receptors rather than by a direct aldosterone effect ^[16] . These findings led us to hypothesize that renin-angiotensin system blockade by ACE inhibitors or angiotensin receptor blockers could lead to hyperkalemia in hemodialysis patients by inhibiting the residual renal excretion of potassium, the colonic excretion of potassium, or the cellular uptake of potassium.

Material and methods

We performed a prospective cohort study involving adult patients with end-stage renal disease. All patients undergoing hemodialysis treatments at The Ottawa Hospital, Ontario, Canada, 3 times per week for a minimum of 90 days, were eligible to participate. There were no other exclusion criteria. All participants provided written informed consent. The study was approved by The Ottawa Hospital Research Ethics Board.

The use of ACE inhibitors or angiotensin receptor blockers was determined by chart review and verified by patient interview. The use of other medications that might affect the serum potassium concentration (digoxin, β-blockers, NSAIDs, diuretics, insulin, oral hypoglycemic agents, β-agonists, and cation exchange resins) was also recorded. Patients were asked to estimate their residual urine output; anuria was defined as a daily urine output of 200 mL or less. All patients were prescribed a diet containing 60 mmol of potassium per day. Age, sex, duration of dialysis, and the presence of diabetes were abstracted from the chart.

Follow-up began on May 1, 1999, and continued until December 2, 1999, unless the patient died, received a kidney transplant, switched to peritoneal dialysis, or transferred to another center. Patients were contacted, and medical records were reviewed after 3 months and again at the end of the study to verify medication use. As part of their usual care, all patients had monthly midweek (ie, Wednesday for patients receiving dialysis on a Monday-Wednesday-Friday schedule, or Thursday for patients on a Tuesday-Thursday-Saturday schedule) predialysis blood samples taken to measure serum potassium and glucose levels. Immediately following dialysis, serum potassium levels were measured again. The dialysis solution potassium concentration and Kt/V (a measure of dialysis adequacy, where K is the urea clearance of the dialysis session, t is the time of the dialysis session, and V is the volume of distribution of urea) were recorded each month. The renin-angiotensin system blockade group consisted of patients who were taking an ACE inhibitor or an angiotensin receptor blocker at the time that blood samples were obtained.

The primary endpoint was hyperkalemia, defined as a predialysis serum potassium concentration of 5.5 mmol/L or higher. A secondary endpoint was the occurrence of severe hyperkalemia, defined as a predialysis serum potassium concentration of 6.0 mmol/L or higher.

Baseline differences between the groups were compared with the chi-squared test or the Fisher exact test for categorical data, and the Student t test for continuous data. Episodes of hyperkalemia (including multiple episodes per patient) and person-months of follow-up were determined. Rates of hyperkalemia per 100 person-months of follow-up were also calculated.

Unadjusted and multivariate-adjusted odds ratios (ORs) and their 95% confidence intervals (CIs) were determined using logistic regression. In the regression models, each monthly dialysis treatment (in which serum potassium concentration was measured) was considered a separate observation. To account for within-patient correlation, models were performed with a random-person effect. The analyses were carried out using PROC NLMIXED in SAS (version 8.1, Cary, North Carolina). Clinically important variables (age, sex, diabetes, duration of dialysis, residual urine output, dialysis solution potassium concentration, Kt/V, glucose, and the use of digoxin, β-blockers, insulin, NSAIDs, β-agonists, diuretics, and cation exchange resins) that showed a trend toward statistical significance (P <0.2) on univariate testing were included in the multivariate models. Only variables that reached statistical significance (P <0.05) were retained in the final model. To determine whether the association between renin-angiotensin system blockade and hyperkalemia was modified by residual urine output, we conducted a stratified analysis for anuric and nonanuric groups. An interaction term between anuria and renin-angiotensin system blockade was included in the multivariate models to test for a modification of effect based on residual urine output. All tests of statistical significance were two-tailed.

Results

There were 278 patients eligible for the study. Nineteen patients were not enrolled because they were unable to give informed consent, and 8 patients declined to participate, leaving 251 patients in the cohort. At baseline, 71 patients (28%) were in the renin-angiotensin system blockade group, and 180 patients (72%) were in the control group. In the renin-angiotensin system blockade group, 73% (n = 52) were taking enalapril (median daily dose 10 mg), 10% (n = 7) were taking losartan (median daily dose 50 mg), 10% (n = 7) were taking fosinopril (median daily dose 10 mg), and 7% (n = 5) were taking captopril (median daily dose 50 mg). Nine patients (13%) in the renin-angiotensin system blockade group left the study early: 4 died, 4 received kidney transplants, and 1 switched to peritoneal dialysis. Twenty-two patients (12%) in the control group left the study early: 12 died, 4 received kidney transplants, 3 switched to peritoneal dialysis, and 3 transferred to another dialysis center. After the study began, 24 patients were started on an ACE inhibitor or an angiotensin receptor blocker, and 6 patients had these medications discontinued. Because study results did not change whether these patients were included or excluded, these patients were included in all analyses.

Patients in the renin-angiotensin system blockade group were slightly younger than those in the control group, but otherwise had similar characteristics (Table 1 ). Patients in the renin-angiotensin system blockade group were less likely to be treated with a dialysis solution that contained a high potassium concentration (3.0 mmol/L) than were the control patients. The use of other medications that might affect the serum potassium level was also similar in each group (Table 2 ). No patients were taking a cation exchange resin.

There were 367 episodes of hyperkalemia during the 1877 person-months of follow-up (20 episodes per 100 person-months). Patients underwent hemodialysis immediately after the blood samples were taken to measure the serum potassium concentration, and patients with hyperkalemia received additional dietary counseling. No patient required any other specific intervention because of hyperkalemia found during routine monthly testing.

Patients in the renin-angiotensin system blockade group had a nearly twofold higher risk of developing hyperkalemia compared with the patients in the control group (Table 3 ), which persisted after adjustment for age, dialysis solution potassium concentration, Kt/V, anuria, and the use of β-blockers, insulin, and oral hypoglycemic agents. As expected, anuric patients had a substantially greater risk of hyperkalemia than did nonanuric patients (Table 3 ). Renin-angiotensin system blockade was associated with an increased likelihood of hyperkalemia in both anuric and nonanuric patients (Table 3 ). There was no evidence of an interaction between anuria and renin-angiotensin system blockade (P = 0.64). The increased risk of hyperkalemia in the renin-angiotensin system blockade group was not due to inadequate dialysis clearance; the mean (±SD) postdialysis serum potassium concentration was 3.3 (±0.4) mmol/L in the renin-angiotensin system blockade group and 3.4 (±0.4) mmol/L in the control group.

During the study period, there were 141 episodes of severe hyperkalemia (predialysis serum potassium concentration ≥6.0 mmol/L). Patients in the renin-angiotensin system blockade group had a higher risk of developing severe hyperkalemia than did patients in the control group in an unadjusted analysis (OR = 1.5; 95% CI: 1.0 to 2.1), but not in an adjusted analysis (OR = 1.5; 95% CI: 0.9 to 2.6). Patients with anuria had about a twofold higher risk of severe hyperkalemia than did nonanuric patients (OR = 2.0; 95% CI: 1.1 to 3.7) in an adjusted analysis. There was no evidence of an interaction between anuria and renin-angiotensin system blockade and the risk of severe hyperkalemia (P = 0.58).

Discussion

In this prospective study of patients receiving maintenance hemodialysis, the use of an ACE inhibitor or an angiotensin receptor blocker was associated with a significant increase in the risk of hyperkalemia. For patients with residual renal function, the risk of hyperkalemia was about twice as great among those taking ACE inhibitors or angiotensin receptor blockers, compared with other patients. This finding is consistent with reports of hyperkalemia following blockade of the renin-angiotensin system in patients with renal insufficiency who are not on dialysis ^{[9] [10] [17]} and indicates that the risk of hyperkalemia continues after regular dialysis therapy has been initiated.

The risk of hyperkalemia was also greater among anuric patients taking an ACE inhibitor or an angiotensin receptor blocker. As these patients had no residual renal function, the increased risk of hyperkalemia was likely due to a reduction in the gastrointestinal excretion of potassium or to a defect in the cellular uptake of potassium ^{[11] [12]} . These results are consistent with previous studies of the effects of the renin-angiotensin system on the colonic secretion of potassium ^{[14] [15] [16]} and the extracellular removal of potassium ^{[13] [18]} .

The results of this study may have clinical relevance owing to the use of ACE inhibitors in dialysis patients that appears to be increasing ^[19] , with about 40% of these patients having heart failure and a similar proportion having coronary artery disease ^[20] . Studies in the general population have shown that ACE inhibitors reduce cardiovascular events in patients with heart failure ^[21] , as well as in patients with vascular disease and normal left ventricular function ^[22] . If the results of these trials are applied to dialysis patients, it is likely that use of ACE inhibitors will increase further. In addition, ACE inhibitors delay the progression of renal disease ^{[23] [24]} , and a recent study demonstrated that the use of ACE inhibitors in patients initiating dialysis was associated with a 32% lower risk of loss of residual renal function ^[25] . Although the expanded use of ACE inhibitors may have overall benefits for hemodialysis patients, our results suggest that there may also be an increase in the frequency of hyperkalemia.

The strengths of this study included a low rate of nonparticipation, a high degree of follow-up, and complete ascertainment of predialysis serum potassium concentrations. The prospective design allowed accurate classification of medication use and potential confounding variables. However, the study also had several imitations. The serum potassium concentration was measured only when routine monthly blood tests were obtained, so that the patients would not be inconvenienced with any additional testing. Had the serum potassium concentration been measured on Mondays or Tuesdays (after an additional dialysis-free day) or during unscheduled visits to the emergency room, the rates of hyperkalemia and severe hyperkalemia may have been substantially higher. As the study participants were not inpatients, the potassium content of their food could not be measured. Although all patients were prescribed a similar diet containing the same amount of potassium and met regularly with a trained renal dietician, we cannot exclude the possibility that some patients did not follow the diet accurately. Physician prescribing may have been influenced by the knowledge that blockade of the renin-angiotensin system can cause hyperkalemia in patients with renal insufficiency who are not on dialysis. As a result, some hemodialysis patients who had recurrent hyperkalemia before the study was initiated may have been prescribed agents other than ACE inhibitors or angiotensin receptor blockers for the treatment of hypertension. However, confounding such as this would tend to weaken the observed association between renin-angiotensin system blockade and hyperkalemia.

In conclusion, we have demonstrated that the use of ACE inhibitors or angiotensin receptor blockers is associated with an increased risk of hyperkalemia in chronic hemodialysis patients. The serum potassium concentration should be closely monitored when prescribing these medications to hemodialysis patients.