Hyperkalemia (HK) is a common electrolyte disorder in chronic kidney disease (CKD), especially advanced stages of CKD.1–5

The main pathogenic mechanisms involved in the development of HK in CKD are6–12: an intake with intestinal absorption of K that exceeds the excretion capacity by the kidneys, which may be limited by an interference in the mechanisms of tubular adaptation to a high load of K, especially those related to the concentration and / or action of aldosterone in the distal tubule. In addition, a redistribution of intracellular K in the case of metabolic acidosis, especially the mineral form — not the one caused by an excess organic acids.13,14

Despite the accumulation of factors predisposing to the development of HK, the kidney is able to maintain the K balance until the glomerular filtrate decreases below 10–15ml/min.8–11 Adaptations to normalize K levels include an increase in tubular secretion of K that eventually leads to a new equilibrium state of K levels, which often remain higher than normal (chronic HK).6–12 An increase in intestinal excretion of K is another adaptive mechanism that develops gradually in patients with CKD.6,15,16

The information on the management of K in CKD comes mainly from the translation of experimental research6–12; results from this investigation do to always answer questions frequently asked by physicians in the daily clinical practice, such as: is the HK in CKD solely the result of a greater (dietary) load of K?, how much is the safe daily load of K in relation to a given glomerular filtration?, what compensation levels of renal excretion of K can be achieved in advanced CKD? What are the main determinants of chronic HK in these patients? Research on these issues could be useful to develop strategies for the management of HK's in advanced CKD.

To investigate on the management of potassium under real clinical conditions in patients with advanced CKD, a cross-sectional study was conducted, in which the main differences between those who present or not HK are analyzed.

This is an observational cross-sectional study in a cohort of adult predialysis patients with CKD stage 4–5. Among the incident patients in the advanced CKD outpatient clinic during the period from September 2009 to May 2015, we selected those >18 years, with estimated glomerular filtration rate <30mL/min/1.73m2, and able to correctly collect the 24-hour urine (see the definition of this criterion below).

Patients not clinically stable (heart, liver or respiratory failure, cancer, active infection, hypercatabolic states or severe gastrointestinal disorders), with a recent acute renal failure, on treatment with K-sparing diuretics (spironolactone or eplerenone), corticosteroids, fludrocortisone or cation exchange resins were excluded.

Included patients attended regularly the advanced CKD outpatient clinic. In addition to the conventional biochemical blood parameters - urea, creatinine, albumin, sodium, chloride, potassium and bicarbonate - creatinine, urea, sodium and potassium concentrations were also determined in all patients in urine collected during the 24h prior to the collection of blood samples.

Biochemical determinations were performed by conventional laboratory methods (Advia Chemistry, Siemens Healthcare Diagnostics).

Glomerular filtration rate was estimated using the abbreviated formula MDRD.17

The concentration of K was measurement in serum, not plasma. To detect possible interferences and errors in the measurement of serum K, a systematic review was performed by hemolytic index (HI), and all samples with a value considered as significant interference (HI>40) were discarded.

To verify the reliability of the 24-hour urine collection, the measured total excretion of creatinine was compared with that expected according to the anthropometric characteristics according to the formula of Ix et al.18 As a selection criterion for inclusion in the study, total creatinine in 24h urine should not differ ± 20% from the estimated creatinine excretion.

Through the total urinary excretion of urea nitrogen, the protein catabolism rate was calculated using the formula of Maroni et al.19

The amount of K collected in the 24-hour urine was taken as a total K load, this parameter referring to the approximate amount of the K ingested-absorbed, minus the amount excreted in the intestine, assuming in all patients, as a constant, both the amount excreted by the intestine and the endogenous balance.

The fractional excretion of K (FEK) was calculated according to the formula:

Ku/GFR=(K in 24h urine)/ GFR, where GFR was measure by the Brochner-Mortensen formula20:

The net endogenous acid generation was also calculated according to the formula of Frassetto et al.21

HK was defined as a concentration of serum K≥5.5mEq/l.

The data were recorded at the first visit to the advanced CKD clinic, before the patients were instructed to make any diet restriction, although it is worth to mention the tendency to restrict some foods rich in K by the own patients or following recommendations by other nephrologists or primary care doctors.

During the inclusion period, 608 patients were studied in the advanced CKD outpatient clinic, of which 54 were excluded due to clinical instability or recent acute renal failure, 17 were exclude because samples had high hemolytic index, 27 were not considered because of treatment with cation exchange resins, 10 were on treatment with diuretics sparing K and 288 were not included because they failed to meet the criteria for correct 24-hour urine collection. Thus, the patients finally included in the study were 212 with the demographic, clinical and biochemical characteristics shown Table 1.

The average serum K concentration of the study group was 5.1±0.7mmol/l. A 30% of patients (63 patients) showed HK. The average excretion of K in 24-hour urine was 57±20mmol, with a minimum value of 8mmol and a maximum of 123mmol. The distribution of urine K excretion values followed a normal distribution (Fig. 1).

Fig. 1.

Histogram of frequency distribution of total urinary potassium measured in 24-hour urine samples.

(0.08MB).

Male showed a significantly higher urinary excretion of both K (62±20 vs. 49±17mmol/24h, p<0.0001) and sodium (168±71 vs. 123±53mmol / 24h, P<0.0001) as compared with females.

The main differences between those with and without HK were: a slightly more reduced renal function, with more metabolic acidosis and more frequently treated with inhibitors of the renin-angiotensin system (ACEI/ARB).

No differences in absolute urinary excretion of K and sodium were observed between patients with or without HK. However, the total amount of K collected in urine adjusted to the degree of renal insufficiency, that is, the load of K per ml/min/1.73m 2 of glomerular filtration rate (Ku/FG), was greater in patients with HK than in those with normal K (Table 1).

The values of fractional excretion of K (FEK) were distributed over a wide range, from a minimum of 5% to a maximum of 88%. The mean FEK was significantly lower in patients with HK (Table 1).

The differences in the main study parameters according to the prescribed medication are shown in Table 2.

FEK and Ku / GFR showed a strong linear correlation (R2=0.74) (Fig. 2), so that the higher the K load per unit of renal function the greater the FEK.

Fig. 2.

Linear regression between the fractional excretion of potassium and the potassium load relative to renal function.

(0.1MB).

In Fig. 3 patients with or without HK were divided into separate regressions. It was observed that for the same Ku / GFR load, the FEK was lower in those patients with HK.

Fig. 3.

Partial linear regressions between the fractional excretion of potassium and the potassium load relative to renal function in patients with or without hyperkalemia (HK).

(0.11MB).

By multivariable logistic regression the main determinants of the presence of HK in this group of patients were (Table 3): gender (men would show a lower probability for the development of HK if the rest of the variables are taken into account), serum bicarbonate (inverse relationship with the risk of HK), but especially the load of K per unit of renal function (Ku/GFR) and the FEK.

Although men showed a higher, but not significant, percentage of HK than women (33% vs. 26%), they also had a greater K load per unit of renal function (Ku/GFR 4.06±1. 51 vs. 3.61±1.33mmol/mL/min/1.73m2 ; P=0.027), but a similar FEK (34.7±13.6% vs. 35.7±14.1 %, P=0.581). Thus, if the relationship between the Ku/GFR and the FEK is represented separately according to gender, it is observed that for a given load of K the compensatory response is less in men than females (Fig. 4) (for a more extensive explanation of this finding, see section "Discussion").

Fig. 4.

Partial linear regressions between the fractional excretion of potassium and the potassium load relative to renal function in male and female patients.

(0.1MB).

Through multiple linear regression, the main determinants of serum K concentration were serum bicarbonate, Ku/FG and FEK (Table 4).

Despite the strong correlation between Ku/GFR and FEK, multicollinearity was ruled out in this multiple linear regression model by analysis of the variance inflation factor (Table 4).

The present study, on kidney control of K in daily clinical practice in pre-dialysis patients with advanced CKD, shows that the main determinants of the maintenance of serum K levels are: the load of K (intake and absorption less extrarenal excretion) relative to renal function, and FEK, the latter as an expression of the degree of adaptive compensation. In addition, serum bicarbonate was also shown as an independent determinant of HK and serum K levels in the patients studied.

The total amount of K excreted in the 24h urine reflects, in a stable clinical situation, the amount of K ingested and absorbed from the diet minus the amount excreted by the extrarenal route, mainly intestinal.7–9,11 Assuming a balance of endogenous K (i.e., K released by cell lysis equal to that used by new cells), and a constant intestinal excretion, urinary K (Ku) could be taken as an approximation of K ingested in the diet.

This parameter, here referred to as the K load, presented a normal distribution with a wide distribution range. Taking into account that the absorption of ingested K is 90%, and that the intestinal excretion in patients with CKD is approximately 15mmol/day, it can be estimated that the average intake of K in this group of patients was about 78mEq/day (3.04g).

Although the recommendations on the content of K that should have a healthy diet are about 4.7g/day,22 the measurements obtained in some studies in the general population indicate an average daily intake of K is approximately 3g,23 similar to the value estimated in our patients. Thus, the incidence of HK, which was 30%, was obtained assuming that this group of advanced CKD patients were not on a intentional dietary restriction of K or on any type of intestinal captor of K. However, HK was not strictly related to the absolute load of K, but to the K load relative to its glomerular filtration rate (Ku/ GFR).

Male patients showed a significantly greater load of both K and sodium as compared to female. This data is consistent with other studies in which the highest incidence of HK in men is attributed to a higher dietary intake of this K.23–25

With a greater K load and a significantly higher K load relative to renal function (Ku/ GFR), the incidence of HK was slightly higher in men than in women, although the difference was not significant. But, it was paradoxical that the men had a lower FEK than expected for their Ku/GFR. These results could explain the inclusion of male sex as an independent variable and significant determinant of the HK in the logistic regression model; a biological interpretation of this finding could be a more efficient extrarenal management of potassium in men than in women.

The strong linear correlation between the load of K relative to renal function (Ku/GFR) and FEK allowed us to uncover one of the main characteristics of patients with HK, as was the smaller slope of this line that could indicate the limitation of physiological process of compensation of renal excretion in patients with higher serum K concentration. In this group patients with HK the correlation between compensatory increase of FEK against high K loads was maintained, which points out that the determinants of a limitation in renal excretion of K (e.g., angiotensin inhibitors, hypoaldosteronism, etc.) act only incompletely by promoting a higher threshold of Kaliemia (chronic HK), above which there is still a capacity to increase renal excretion of K and avoid more extreme elevation of serum K levels. This finding is consistent with the physiological mechanisms of tubular handling of K,10,11,26 which can maintain an effective excretion of K even in situations of aldosterone blockade, and that despite promoting chronic HK, the serum K levels rarely reach extreme high levels.

Serum bicarbonate levels was found to be another of the main determinants of the HK. This close association is somewhat surprising because the direct pathophysiological mechanism between HK and metabolic acidosis is through redistribution of intracellular K.13,14 Such a situation occurs in cases of non-organic acid retention (mineral acidosis). A possible explanation for this finding could be the bidirectional relationship between HK and acidosis, that is, acidosis could cause a certain degree of redistribution of K, but at the same time HK could also reduce renal acidification mechanisms (decreased generation of ammonia).27 Moreover, inhibitors of the renin-angiotensin system or situations of low aldosterone secondary to low renin could also limit K excretion and development of tubular acidosis.13,14

This study has limitations. Due to the transversal and retrospective design, definite causal relationships cannot be established, and the interpretation of these static data are being extrapolated to theoretical dynamic responses in individuals. The criteria for inclusion of patients were fixed with the objective of obtaining reliable urine samples. It is possible that due to these criteria, more compliant and rigorous patients were selected which could cause selection bias. Nevertheless, the incidence of HK was similar to that observed in our ERCA population.

The assumptions of the endogenous K balance and constant K intestinal excretion in each individual may not be real, however the exclusion of patients with factors that predispose to an imbalance in K makes unlikely that this assumptions could alter the results and final estimations of this study.

In conclusion, although the K load relative to glomerular filtration rate (Ku/GFR) is a determinant of the probability of development of HK in patients with advanced CKD, the limitation of a compensatory renal excretion of K expressed as a lower FEK characterizes this electrolyte alteration. However, in these patients there could be exceptional mechanisms playing a role on K excretion that can effectively prevent extreme increase serum K levels.

The authors have no conflicts of interest to declare.