NDT Advance Access published December 13, 2014 Nephrol Dial Transplant (2014) 0: 1–8 doi: 10.1093/ndt/gfu376
Original Article Ranking of factors determining potassium mass balance in bicarbonate haemodialysis and Carlo Lomonte1 1
Division of Nephrology, Miulli General Hospital, Acquaviva delle Fonti, Italy and 2Division of Nephrology, Perrino Hospital, Brindisi, Italy
Correspondence and offprint requests to: Carlo Basile; E-mail: [email protected]
A B S T R AC T Background. One of the most important pathogenetic factors involved in the onset of intradialysis arrhytmias is the alteration in electrolyte concentration, particularly potassium (K+). Methods. Two studies were performed: Study A was designed to investigate above all the isolated effect of the factor time t on intradialysis K+ mass balance (K+MB): 11 stable prevalent Caucasian anuric patients underwent one standard (∼4 h) and one long-hour (∼8 h) bicarbonate haemodialysis (HD) session. The latter were pair-matched as far as the dialysate and blood volume processed (90 L) and volume of ultrafiltration are concerned. Study B was designed to identify and rank the other factors determining intradialysis K+MB: 63 stable prevalent Caucasian anuric patients underwent one 4-h standard bicarbonate HD session. Dialysate K+ concentration was 2.0 mmol/L in both studies. Blood samples were obtained from the inlet blood tubing immediately before the onset of dialysis and at t60, t120, t180 min and at end of the 4- and 8-h sessions for the measurement of plasma K+, blood bicarbonates and blood pH. Additional blood samples were obtained at t360 min for the 8 h sessions. Direct dialysate quantification was utilized for K+MBs. Direct potentiometry with an ion-selective electrode was used for K+ measurements. Results. Study A: mean K+MBs were significantly higher in the 8-h sessions (4 h: −88.4 ± 23.2 SD mmol versus 8 h: −101.9 ± 32.2 mmol; P = 0.02). Bivariate linear regression analyses showed that only mean plasma K+, area under the curve (AUC) of the hourly inlet dialyser diffusion concentration gradient of K+ (hcgAUCK+) and AUC of blood bicarbonates and mean blood bicarbonates were significantly related to K+MB in both 4- and 8-h sessions. A multiple linear regression © The Author 2014. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.
output with K+MB as dependent variable showed that only mean plasma K+, hcgAUCK+ and duration of HD sessions per se remained statistically significant. Study B: mean K+MBs were −86.7 ± 22.6 mmol. Bivariate linear regression analyses showed that only mean plasma K+, hcgAUCK+ and mean blood bicarbonates were significantly related to K+MB. Again, only mean plasma K+ and hcgAUCK+ predicted K+MB at the multiple linear regression analysis. Conclusions. Our studies enabled to establish the ranking of factors determining intradialysis K+MB: plasma K+ → dialysate K+ gradient is the main determinant; acid-base balance plays a much less important role. The duration of HD session per se is an independent determinant of K+MB. Keywords: acid-base balance, diffusion concentration gradient, haemodialysis, plasma potassium, potassium mass balance
INTRODUCTION Cardiovascular diseases account for 38–40% of all deaths in dialysis patients with a large proportion (around 25%) attributed to sudden cardiac death [1, 2]. Dialysis treatment per se can be considered as an arrhytmogenic stimulus [1]. One of the most important pathogenetic elements involved in the onset of intradialysis arrhytmias is the alteration in electrolyte concentration, particularly potassium (K+) and calcium. The control of plasma K+ is still one of the most severe problems in the global treatment of haemodialysis (HD) patients. One of the main goals of HD is the removal of K+ that has accumulated in the body in the interval between two dialyses. A correct K+ mass balance (K+MB) during HD is crucial: it should be negative and of the same order of magnitude of 1
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
Carlo Basile1, Pasquale Libutti1, Piero Lisi1, Annalisa Teutonico1, Luigi Vernaglione2, Francesco Casucci1
the positive interdialysis K+MB, in order to prevent both dangerous intradialysis hypokalaemia and fatal interdialysis hyperkalaemia. The main aim of the present study was to identify and rank the factors determining the intradialysis K+MB.
Study design This study included two protocols (Studies A and B) in which K+ kinetics were performed in two selected groups of our HD population: Study A was designed to investigate above all the isolated effect of the factor time t on the removal and kinetic behaviour of K+; Study B was designed to identify and rank the other factors determining the intradialysis K+MB: mean plasma K+, Δ plasma K+, area under the curve (AUC) of the hourly inlet dialyser diffusion concentration gradient during every single HD session of K+ (hcgAUCK+), mean blood pH, Δ mean blood pH, AUC of blood pH, mean blood bicarbonates, Δ mean blood bicarbonates, AUC of blood bicarbonates, age, gender, dialysis vintage, volume of ultrafiltration (VUF), and pre- and post-dialysis body weight. The present studies were approved by our institutional review board and were conducted in accordance with good clinical practice guidelines and ethical principles of the Helsinki Declaration. The inclusion criteria were (i) standard bicarbonate HD treatment since at least 6 months; (ii) uncomplicated HD sessions. All the HD sessions in both studies A and B used the GENIUS® single-pass batch dialysis system (Fresenius Medical Care, Bad Homburg, Germany). The characteristics of the GENIUS® dialysis system have been described elsewhere [3, 4]. Briefly, it uses a double-sided roller pump that generates equal blood and dialysate flows up to 350 mL/min. The system consists of a closed dialysate tank of 90 L and although fresh and spent dialysate are stored together, there is no mixing of fresh and spent dialysate. A volumetric ultrafiltration pump drains the excess body water (the programmed weight loss) from the spent dialysate volume and collects it into the ultrafiltrate recipient. Thus, the ultrafiltrate in this single-pass batch dialysis system is part of the spent dialysate volume drained out from the closed dialysate tank of 90 L [3, 4]. Highflux FX80 dialysers (Fresenius Medical Care) were used in all sessions. Characteristics of the dialyser were described elsewhere [3]. Dialysate composition was as follows: magnesium 0.5.0 mmol/L; potassium 2.0 mmol/L; sodium 140 mmol/L; total calcium 1.5 mmol/L; bicarbonate 35 mmol/L; chloride 113 mmol/L; glucose 5.55 mmol/L and citrate 0.10 mmol/L. Kinetic studies Study A. A group of 11 stable prevalent Caucasian anuric patients (9 males and 2 females, mean age 54.1 ± 17.8 SD years, dialysis vintage 78.0 ± 60.2 months) was enrolled. After obtaining informed consent, they underwent one standard (∼4 h) and one long-hour (∼8 h) slow-flow bicarbonate HD session in a random sequence, always at the midweek interval, at least one week apart. The sessions were pair-matched as far as the dialysate and blood volume processed (90 L) and VUF
2
Study B. A group of 63 stable prevalent Caucasian anuric patients (40 males and 23 females, mean age 55.3 ± 14.7 years, dialysis vintage 55.6 ± 50.9 months) was enrolled for this study. After obtaining informed consent, participants underwent intradialysis kinetic studies. They consisted of one data recording session (at the midweek interval) for each of the 63 patients. The sessions lasted 4 h and were standardized as far as the dialysate and blood volume processed (90 L) are concerned. Blood and dialysate sampling Study A. Blood samples were obtained from the inlet blood tubing immediately before the onset of dialysis and at t60, t120, t180 min and at end of the 4 and 8 h sessions for the measurement of plasma K+, blood bicarbonates and blood pH. Additional blood samples were obtained at t360 min for the 8 h sessions. Dialysate was collected from the inlet dialysate tubing at the same time points as blood samples. Study B. Blood samples were obtained from the inlet blood tubing immediately before the onset of dialysis and at t60, t120, t180 min and at end of the sessions (t240 min) for the measurement of plasma K+, blood bicarbonates and blood pH. Dialysate was collected from the inlet dialysate tubing at the same time points as blood samples. Furthermore, at the end of dialysis (studies A and B), two samples were taken, respectively, from the ultrafiltrate recipient and from the dialysate tank, after thorough mixing, in order to quantify solute concentration in total spent dialysate. Particular attention was paid to the thorough mixing of the spent dialysate, strictly adhering to the ad hoc instructions present in the GENIUS® operator’s manual. Measurements Blood samples collected at the times described were used for the immediate measurement of plasma K+ concentrations by means of direct potentiometry with an ion-selective electrode and pH by means of an ABL 800 series blood gas analyser (Radiometer Medical ApS, Bronshoj, Denmark). K+ concentrations were measured in the fresh and spent dialysate and in the ultrafiltrate recipient by using the dialysate mode of the same instrument. Blood bicarbonates were calculated using the Henderson-Hasselbalch equation, assuming a pK of 6.1 and a solubility coefficient for carbon dioxide of 0.0301. The above parameters allowed the calculation of, in each patient, the area under the curve (AUC) of the hourly inlet dialyser diffusion concentration gradients between K+ concentrations in the plasma and in the dialysate (hcgAUCK+), AUC of blood bicarbonates and AUC of blood pH by using the trapezoidal rule [6, 7]. To calculate K+MBs, K+ concentrations in the inlet dialysate were considered those given by the average of the hourly
C. Basile et al.
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
ORIGINAL ARTICLE
SUBJECTS AND METHODS
are concerned. It must be underlined that all data of Study A were extracted from the same database from which data related to removal of uraemic retention solutes in standard bicarbonate HD and long-hour slow-flow bicarbonate HD were extracted and already published [5].
measurements. Furthermore, K+ concentrations in total spent dialysate and in the ultrafiltrate recipient were considered those given by the average of five consecutive measurements performed on the same sample. K+MB was calculated as follows: Kþ MB ¼ ½Kþ concentration in the fresh dialysate 90 ½ðKþ concentration in the spent dialysate 90Þ þ ðKþ concentration in the ultrafiltrate recipient VUF Þ where 90 is the volume (L) of both fresh and spent dialysate.
Study A Table 1 shows the comparison between the mean values of the HD data in the two sessions (4 and 8 h) performed on each of the 11 patients. As said, the sessions were pairmatched as far as the dialysate and blood volume processed (90 L) and VUF are concerned. Thus, the duration of treatment was dictated by the achievement of the target dialysate and blood volume processed (i.e. 90 L): for this reason, it was slightly different from 4 and 8 h. As expected, blood flow rate, dialysate flow rate and ultrafiltration rate were statistically significantly different when comparing 4- and 8-h sessions (P = 0.0001). Notably, K+MB was significantly higher in the 8 h sessions (4 h: −88.4 ± 23.2 mmol versus 8 h: −101.9 ± 32.2 mmol; P = 0.02) (Table 2), even if plasma K+ concentrations were not statistically significantly different both at the start and at the end of the 4- and 8-h sessions (Table 2). The same was
Intradialysis potassium mass balance
4h
8h
P*
Treatment time (min) Blood flow rate (mL/min) Blood volume processed (L) Dialysate flow rate (mL/min) Dialysate volume processed (L) Pre-dialysis body weight (kg) Post-dialysis body weight (kg) Ultrafiltration rate (mL/min) Volume of ultrafiltration (L) Inlet dialyser K+ concentration (mmol/L)
257.7 (1.1) 469.1 (2.8) 350 (0) 190 (0) 90 (0) 90 (0) 350 (0) 190 (0) 90 (0) 90 (0) 72.2 (10.8) 71.9 (10.7) 69.2 (10.3) 69.1 (10.1) 11.3 (3.1) 6.2 (1.9) 2.9 (0.8) 2.9 (0.9) 2.0 (0.2) 2.0 (0.2)
0.0001 0.0001 NS 0.0001 NS NS NS 0.0001 NS NS
Mean(SD). NS, not significant. *Student’s t-test for unpaired data.
Table 2. Comparison of dialysis parameters at the start and the end of the 4 and 8 h sessions of Study A Parameter Plasma K+ concentration (mmol/L) at T0 Plasma K+ concentration (mmol/L) at the end of dialysis Blood bicarbonate concentration (mmol/L) at T0 Blood bicarbonate concentration (mmol/L) at the end of dialysis Blood pH at T0 Blood pH at the end of dialysis K+MB (mmol) K+ mass removed by ultrafiltration (mmol)
4h
8h
P*
5.49 (0.70)
5.49 (0.70) NS
3.73 (0.37)
3.75 (0.35) NS
20.0 (2.12)
20.0 (2.05) NS
5.4 (1.2)
25.6 (1.1)
7.38 (0.04) 7.37 (0.04) 7.45 (0.02) 7.45 (0.02) −88.4 (23.2) −101.9 (32.2) −5.51 (0.7) −6.35 (0.3)
ORIGINAL ARTICLE
R E S U LT S
Parameter
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
Statistical analyses Data are reported as mean ± SD, median and interquartile ranges and 95% confidence intervals when needed. Study A: the mean values of plasma K+, blood bicarbonates and blood pH at the start and end of each HD session and K+MBs and HD treatment characteristics of patients enrolled in study A were compared by Student’s t-test for unpaired data. A generalized longitudinal mixed model for repeated measures with an autoregressive covariance structure was built for the study of comparisons of plasma K+ levels, blood pH and bicarbonates trends between 4 and 8 h sessions. The model for kinetic studies included the times of HD sessions (5 levels) and used the Wilks λ as multivariate test for the significance of each effect studied and Mauchly sphericity test of the matrix. In case of no sphericity of the matrix, the lower-bound ε was considered for the adjustment of the degrees of freedom. Study B: the one-way ANOVA for repeated measures followed by Bonferroni’s post hoc test was used for kinetic studies. The significance of the relationships between K+MBs and the variables studied was tested by means of the linear bivariate regression analysis in both studies A and B. All the variables significantly correlated to K+MB were considered as independent variables in a multiple linear regression model with K+MB as dependent variable in both studies. All statistical inferences were made using SPSS, version 11.0 software (SPSS, Inc., Chicago, IL, USA) and a P < 0.05 was considered statistically significant.
Table 1. Dialysis data of the HD sessions of Study A
NS NS NS 0.02 NS
Mean(SD). NS, not significant. *Student’s t-test for unpaired data.
true for blood pH and bicarbonate levels (Table 2). Furthermore, Figure 1 shows the trends of plasma K+ concentrations, blood pH and blood bicarbonate levels in the 4- and 8-h sessions. In order to find out the determinants of K+ removal by HD, bivariate linear regression analyses between K+MB and the parameters quoted in the Subjects and Methods section were performed. As reported in Table 3, only mean plasma K+, hcgAUCK+, mean blood bicarbonates and AUC of blood bicarbonates were significantly related to K+MB both in 4- and 8-h sessions. It must be pointed out that the relationships between K+MBs and mean blood bicarbonates and between K+MBs and AUC of blood bicarbonates were inversely related. However, only mean plasma K+ concentration and hcgAUCK+ beyond the duration of HD sessions per se kept their statistical significance when tested in a multiple linear regression output with K+MB as dependent variable (P = 0.007) (Table 4).
Study B Table 5 shows the dialysis data of the 63 HD sessions including also K+MBs.
3
ORIGINAL ARTICLE values are plotted. By applying the generalized longitudinal mixed model for repeated measures with an autoregressive covariance structure built for this study, the variations of the three parameters resulted smoother in the 8 h sessions even if the values at the end of the sessions were not statistically significantly different. In particular, in the 4 h sessions: the values of plasma K+ concentrations resulted significantly lower at T1, T2 and T3, blood pH significantly higher at T1, T2, T3 and T4, blood bicarbonates significantly higher at T1 and T2.
Trends of hourly plasma K+ concentrations, blood pH and blood bicarbonate concentrations are shown in Figures 2a–c, respectively. In order to find out the determinants of K+ removal by HD, bivariate linear regression analyses between K+MB and the parameters quoted in the Subjects and Methods section were performed. As reported in Table 6 and in Figures 3a and b, K+MB were statistically significantly related with mean intradialysis plasma K+ (R 2 = 0.4993, P = 0.0001, Figure 3a), hcgAUCK+ (R 2 = 0.5879, P = 0.0001, Figure 3b) and mean
4
blood bicarbonates. It must be pointed out that the relationship between K+MBs and mean blood bicarbonates was inversely related. Among these significantly related variables at the bivariate analysis, only mean intradialysis plasma K+ (P = 0.0001) and hcgAUCK+ (P = 0.036) predicted K+MB at the multiple linear regression analysis (Table 7). Finally, the distribution of K+MB, blood bicarbonates and blood pH in our population was analysed in order to verify if the lack of relationship among them was dependent by the small ranges of variations of these parameters during the study
C. Basile et al.
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
F I G U R E 1 : Comparison of hourly (T) trends of plasma K+, blood pH and bicarbonate levels between 4 and 8 h sessions (Study A). Mean
Table 3. Study A: output of the significant bivariate linear regressionsa with K+MB as dependent variable 4h
+
Mean plasma K concentration (mmol/L) Mean blood bicarbonate concentration (mmol/L) hcgAUCK+ (mmol/L × min)b AUC of bood bicarbonatesc (mmol/L × min)
8h
M
B
R2
P
M
B
R2
P
37.95 −11.56 0.17 −0.05
−70.89 363.90 4.70 −9.51
0.61 0.45 0.81 0.33
0.004 0.023 0.0002 0.05
43.31 −19.58 0.09 −0.04
−99.83 567.8 3.32 −29.9
0.80 0.68 0.94 0.61
0.0002 0.0019 0.0001 0.002
a
Model: Y = mX + b. Area under the curve of the hourly inlet dialyser diffusion concentration gradients between K+ concentrations in plasma and dialysate. c Area under the curve of the hourly blood bicarbonate concentrations. b
Table 4. Study A: multiple linear regression output with K+MB as dependent variable
+
P
36.50 −8.71 4.17
0.0002 0.001 0.02
Model: R 2 = 0.75; P = 0.007. a Coefficient of regression. b Area under the curve of the hourly inlet dialyser diffusion concentration gradients between K+ concentrations in plasma and dialysate.
Treatment time (min) Blood flow rate (mL/min) Blood volume processed (L) Dialysate flow rate (mL/min) Dialysate volume processed (L) Pre-dialysis body weight (kg) Post-dialysis body weight (kg) Ultrafiltration rate (mL/min) Volume of ultrafiltration (L) Inlet dialyser K+ concentration (mmol/L) K+MB (mmol) K+ mass removed by ultrafiltration (mmol)
240 (0.0) 350 (0) 90 (0) 350 (0) 90 (0) 79.0 (21.3) 76.4 (19.7) 12.9 (3.7) 3.1 (0.9) 2.0 (0.2) −86.7 (22.6) −5.76 (0.6)
Mean(SD).
procedures. The median values of mean plasma K+ (4.3 mmol/L), blood pH (7.43) and blood bicarbonates (23.5 mmol/L) during the sessions were respectively 49, 5 and 20% of their ranges. Thus, the three distributions were different and that confirms the lack of relationship among them.
DISCUSSION The present study confirms the behaviour of K+ removal described by others [8, 9]. Plasma K+ concentration rapidly decreased during the first 60 min and stabilized during the last 60 min of dialysis. Plasma K+ reached a steady state during the last hour of dialysis, while K+ continued to emerge into the dialysate. Therefore, it can be assumed that K+ removal rate was equal to the intra- to extracellular mass transfer rate at these time points (Figures 1a and 2a). Many are the factors determining the intradialysis K+MB: Sam et al. identified some of them in their review: dialysate K+
Intradialysis potassium mass balance
Duration of HD session per se Study A was designed to investigate the isolated effect of the factor time t and then of the duration per se of the dialysis session (by processing the same total blood and dialysate volume in two different time schedules) on the kinetics of plasma K+, blood pH and bicarbonates. They had different intradialysis kinetics, but the comparison of end-dialysis data showed no difference between 4- and 8-h sessions. However, the analysis of K+MBs showed a statistically significant larger K+ removal in 8-h HD sessions (Δ 13.56 mmol, equivalent to an increased removal of 15.34%, P = 0.02). The finding is novel as far as K+MB is concerned; however, it confirms consistent data about the important role played by the duration of treatment on the removal of small and middle molecules [3, 5].
5
ORIGINAL ARTICLE
Table 5. Dialysis data of the 63 HD sessions of Study B
Plasma K+ → dialysate K+ gradient Among the significantly related variables at the bivariate analyses in both studies A and B, only mean intradialysis plasma K+ and hcgAUCK+ predicted K+MB at the multiple linear regression analyses (Tables 4 and 7). Thus, our study confirms that the rate of K+ removal during dialysis is largely a function of the pre-dialysis plasma K+ concentration. The higher the initial plasma concentration, the greater the gradient between the plasma and the dialysate and, hence, the greater K+ removal [10]. Actually, Zehnder et al. showed in a prospective, randomized, cross-over study that a 0 mmol/L dialysate K+ concentration was able to remove more K+ than 1 or 2 mmol/L dialysate K+ concentration (P < 0.001) [9]. While confirming the data by Zendher et al., i.e. that the rate of K+ removal from the plasma depends on the gradient between plasma and dialysate K+ concentrations, at variance with their results, we demonstrated that mean intradialysis plasma K+ and hcgAUCK+ were able to predict K+MB at the multiple linear regression analysis [9].
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
Mean plasma K concentration (mmol/L) hcgAUCK+ (mmol/L × min)b Duration of HD session (min)
Ba
concentration, the efficiency of the dialyser, the duration and frequency of dialysis [10]. Furthermore, Fissell and Hakim underlined in their recent review that dialysis treatment lowers plasma K+, both by removal of K+ with dialysate and by rapid shift of K+ from the extracellular to the intracellular space as metabolic acidosis is treated [11]. Main aim of the present study was to identify and rank the factors determining the intradialysis K+MB in bicarbonate HD. The present results allow us to formulate the following ranking:
ORIGINAL ARTICLE peated measures followed by Bonferroni’s post hoc test was statistically significant for the effect of time points of HD sessions showing a gradual stepwise decrease in plasma K+ concentration, and a gradual stepwise increase in blood pH and blood bicarbonate levels. Table 6. Study B: output of the significant bivariate linear regressionsa with K+MB as dependent variable M Mean plasma K+ concentration (mmol/L) Mean blood bicarbonate concentration (mmol/L) hcgAUCK+ (mmol/L × min)b
R2
B
P
30.87 −44.34 0.50 0.0001 −4.74 195.6 0.12
0.47 0.03
17.52 0.59 0.0001
a
Model: Y = mX + b. Area under the curve of the hourly inlet dialyser diffusion concentration gradients between K+ concentrations in plasma and dialysate. b
Acid-base balance Unlike the extracellular K+, which can freely pass across the dialysis membrane, the intracellular K+ is slow to move into the extracellular space. This latter movement is influenced by a variety of factors, which can change during a dialysis procedure [12, 13]. Alkalosis causes a shift of K+ into cells and acidosis results in a K+ efflux from cells. Introduction of buffer
6
base into blood during dialysis promotes cellular uptake of K+ and thereby attenuates the dialytic removal of K+ (this is more evident in an acidotic patient). There are case reports describing that dialysis succeeded in reducing plasma K+ concentrations even though the dialysate K+ levels were higher than the pre-dialysis plasma K+ values. The decline in plasma K+ concentration was associated with a corresponding dialysisinduced rise in blood pH [12]. Finally, a randomized controlled trial showed an association between higher dialysate bicarbonate concentration and a faster decrease in intradialysis plasma K+ concentration [14]. The latter study was not able to show any statistically significant difference in K+MBs when utilizing three different dialysate bicarbonate concentrations, but the non-significant data may simply reflect the low statistical power of the study (only eight patients enrolled) [14]. In our hands, the role played by intradialysis blood pH and bicarbonates in determining K+MB seems to be minor: in both studies A and B the following bivariate regression analyses related to acid-base balance were performed utilizing K+MB as
C. Basile et al.
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
+
F I G U R E 2 : Trends of hourly (T) plasma K concentrations, blood pH and blood bicarbonate levels (Study B). The one-way ANOVA for re-
ORIGINAL ARTICLE
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
F I G U R E 3 : Study B: (a) scatter plot of net K+MB during every single HD session (y axis) versus the mean intradialysis plasma K+ concentra-
tion during every single HD session; (b) scatter plot of net K+MB during every single HD session (y axis) versus the area under the curve (AUC) of the hourly inlet dialyser diffusion concentration gradient during every single HD session ( plasma K+ concentration − dialysate K+ concentration). Best fit linear regressions are shown.
Table 7. Study B: multiple linear regression output with K+MB as dependent variable B +
Mean plasma K 28.88 concentration (mmol/L) hcgAUCK+ (mmol/L × min)a 3.21E-02
SE
P
95% CI
4.246 0.0001 20.35; 37.39 0.015 0.036
0.002; 0.062
a
Area under the curve of the hourly inlet dialyser diffusion concentration gradients between K+ concentrations in plasma and dialysate.
dependent variable and mean blood pH, Δ blood pH, AUC of blood pH, mean blood bicarbonates, Δ blood bicarbonates and AUC of blood bicarbonates as independent variables. Only mean blood bicarbonates and AUC of blood bicarbonates were significantly related to K+MB in Study A and only mean blood bicarbonates in Study B. However, none of them kept its
Intradialysis potassium mass balance
statistical significance when tested into a multiple linear regression output with K+MB as dependent variable. K+ removal during HD can occur through diffusion and convection. Current prescribing practices for chronic intermittent HD rely primarily on diffusive and less on convective losses [8, 9]. Our studies confirm these data (Tables 2 and 5). Thus, intradialysis K+ kinetics is quite different from that of sodium, in which convection accounts for ∼80% of intradialytic sodium mass balance, while the diffusive gradient between plasma and the inlet dialyser sodium concentration is an important factor in the ‘fine tuning’ of sodium mass balance [15]. Limitations of the study Our studies dealt with two small populations (11 patients in Study A and 63 patients in Study B). Thus, results, all the key covariates taken into account notwithstanding, cannot be
7
Conclusions The true challenge in HD patients is to avoid both lifethreatening pre-dialysis hyperkalaemia ( plasma K+ level > 6 mmol/L) and post-dialysis relative hypokalaemia (or at least very rapid decrease of plasma K+ level, and the related risk of lethal arrhythmias): resins (calcium or sodium polystyrene sulfonate) may be used; alternative strategies, such as longer or more frequent HD sessions and/or dialysate K+ profiling [17], may be required in such cases. Our studies enabled the establishment of the ranking of factors determining intradialysis K+MB: plasma K+ → dialysate K+ gradient is the main determinant; acid-base balance plays a much less important role. The duration of HD session per se is an independent determinant of K+MB.
C O N F L I C T O F I N T E R E S T S TAT E M E N T None declared.
REFERENCES 1. Green D, Roberts PR, New DI et al. Sudden cardiac death in hemodialysis patients: an in-depth review. Am J Kidney Dis 2011; 57: 921–929
8
2. US Renal Data System: USRDS 2012. Annual Data Report. National Institutes of Health, National Institutes of Diabetes and Digestive and Kidney Diseases, Bethesda, MD, 2012, accessible from http://www.usrds.org/2012/ view/v2 3. Eloot S, Van Biesen W, Dhondt A et al. Impact of hemodialysis duration on the removal of uremic retention solutes. Kidney Int 2008; 73: 765–770 4. Basile C, Libutti P, Di Turo AL et al. Effect of dialysate calcium concentrations on parathyroid hormone and calcium balance during a single dialysis session using bicarbonate hemodialysis: a crossover clinical trial. Am J Kidney Dis 2012; 59: 92–101 5. Basile C, Libutti P, Di Turo AL et al. Removal of uraemic retention solutes in standard bicarbonate haemodialysis and long-hour slow-flow bicarbonate haemodialysis. Nephrol Dial Transplant 2011; 26: 1296–1303 6. Yeh KC, Kwan KC. A comparison of numerical integrating algorithms by trapezoidal, Lagrange, and spline approximation. J Pharmacokinet Biopharm 1978; 6: 79–98 7. Eckardt K-U, Kim J, Kronenberg F et al. Hemoglobin variability does not predict mortality in European hemodialysis patients. J Am Soc Nephrol 2010; 21: 1765–1775 8. Feig PU, Shook A, Sterns RH. Effect of potassium removal during hemodialysis on the plasma potassium concentration. Nephron 1981; 27: 25–30 9. Zehnder C, Gutzwiller J-P, Huber A et al. Low-potassium and glucosefree dialysis maintains urea but enhances potassium removal. Nephrol Dial Transplant 2001; 16: 78–84 10. Sam R, Vaseemuddin M, Leong WH et al. Composition and clinical use of hemodialysates. Hemodial Int 2006; 10: 15–28 11. Fissell R, Hakim RM. Improving outcomes by changing hemodialysis practice patterns. Curr Opin Nephrol Hypertens 2013; 22: 675–680 12. Weigand C, Davin T, Raij L et al. Life threatening hypokalemia during hemodialysis. Trans Am Soc Artif Intern Organs 1975; 25: 416–418 13. Ward RA, Wathen RL, Williams TE et al. Hemodialysate composition and intradialytic metabolic, acid base and potassium changes. Kidney Int 1987; 32: 129–135 14. Heguilén RM, Sciurano C, Bellusci AD et al. The faster potassium-lowering effect of high dialysate bicarbonate concentrations in chronic haemodialysis patients. Nephrol Dial Transplant 2005; 20: 591–597 15. Basile C, Libutti P, Lisi P et al. Sodium setpoint and gradient in bicarbonate hemodialysis. J Nephrol 2013; 26: 1136–1142 16. Mathialahan T, Maclennan KA, Sandle LN et al. Enhanced large intestinal potassium permeability in end-stage renal disease. J Pathol 2005: 206: 46–51 17. Santoro A, Mancini E, London G et al. Patients with complex arrhytmias during and after haemodialysis suffer from different regimens of potassium removal. Nephrol Dial Transplant 2008; 23: 1415–1421
Received for publication: 21.8.2014; Accepted in revised form: 16.11.2014
C. Basile et al.
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
ORIGINAL ARTICLE
extrapolated to populations not included in these studies, because of the low numerosity of the population studied. Furthermore, it is well known that the colon contributes considerably to K+ removal in dialysis patients, with colonic disposal being ∼30% of the dietary intake, a value that is about three times higher than normal [16]. The colonic K+ excretion was not studied. Thus, we are unable to say if colonic excretion differed among the patients of our studies. Finally, it is also well known that insulin and blood glucose affect K+ shifts. Seven out of the 74 patients under study (9.5%) were insulin-dependent diabetics. Blood glucose evaluation during the experimental HD sessions was not included into the study design. Thus, no blood glucose data are available; anyway, no patient received insulin during or in the peri-dialytic timeframe.
Original Article Ranking of factors determining potassium mass balance in bicarbonate haemodialysis and Carlo Lomonte1 1
Division of Nephrology, Miulli General Hospital, Acquaviva delle Fonti, Italy and 2Division of Nephrology, Perrino Hospital, Brindisi, Italy
Correspondence and offprint requests to: Carlo Basile; E-mail: [email protected]
A B S T R AC T Background. One of the most important pathogenetic factors involved in the onset of intradialysis arrhytmias is the alteration in electrolyte concentration, particularly potassium (K+). Methods. Two studies were performed: Study A was designed to investigate above all the isolated effect of the factor time t on intradialysis K+ mass balance (K+MB): 11 stable prevalent Caucasian anuric patients underwent one standard (∼4 h) and one long-hour (∼8 h) bicarbonate haemodialysis (HD) session. The latter were pair-matched as far as the dialysate and blood volume processed (90 L) and volume of ultrafiltration are concerned. Study B was designed to identify and rank the other factors determining intradialysis K+MB: 63 stable prevalent Caucasian anuric patients underwent one 4-h standard bicarbonate HD session. Dialysate K+ concentration was 2.0 mmol/L in both studies. Blood samples were obtained from the inlet blood tubing immediately before the onset of dialysis and at t60, t120, t180 min and at end of the 4- and 8-h sessions for the measurement of plasma K+, blood bicarbonates and blood pH. Additional blood samples were obtained at t360 min for the 8 h sessions. Direct dialysate quantification was utilized for K+MBs. Direct potentiometry with an ion-selective electrode was used for K+ measurements. Results. Study A: mean K+MBs were significantly higher in the 8-h sessions (4 h: −88.4 ± 23.2 SD mmol versus 8 h: −101.9 ± 32.2 mmol; P = 0.02). Bivariate linear regression analyses showed that only mean plasma K+, area under the curve (AUC) of the hourly inlet dialyser diffusion concentration gradient of K+ (hcgAUCK+) and AUC of blood bicarbonates and mean blood bicarbonates were significantly related to K+MB in both 4- and 8-h sessions. A multiple linear regression © The Author 2014. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.
output with K+MB as dependent variable showed that only mean plasma K+, hcgAUCK+ and duration of HD sessions per se remained statistically significant. Study B: mean K+MBs were −86.7 ± 22.6 mmol. Bivariate linear regression analyses showed that only mean plasma K+, hcgAUCK+ and mean blood bicarbonates were significantly related to K+MB. Again, only mean plasma K+ and hcgAUCK+ predicted K+MB at the multiple linear regression analysis. Conclusions. Our studies enabled to establish the ranking of factors determining intradialysis K+MB: plasma K+ → dialysate K+ gradient is the main determinant; acid-base balance plays a much less important role. The duration of HD session per se is an independent determinant of K+MB. Keywords: acid-base balance, diffusion concentration gradient, haemodialysis, plasma potassium, potassium mass balance
INTRODUCTION Cardiovascular diseases account for 38–40% of all deaths in dialysis patients with a large proportion (around 25%) attributed to sudden cardiac death [1, 2]. Dialysis treatment per se can be considered as an arrhytmogenic stimulus [1]. One of the most important pathogenetic elements involved in the onset of intradialysis arrhytmias is the alteration in electrolyte concentration, particularly potassium (K+) and calcium. The control of plasma K+ is still one of the most severe problems in the global treatment of haemodialysis (HD) patients. One of the main goals of HD is the removal of K+ that has accumulated in the body in the interval between two dialyses. A correct K+ mass balance (K+MB) during HD is crucial: it should be negative and of the same order of magnitude of 1
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
Carlo Basile1, Pasquale Libutti1, Piero Lisi1, Annalisa Teutonico1, Luigi Vernaglione2, Francesco Casucci1
the positive interdialysis K+MB, in order to prevent both dangerous intradialysis hypokalaemia and fatal interdialysis hyperkalaemia. The main aim of the present study was to identify and rank the factors determining the intradialysis K+MB.
Study design This study included two protocols (Studies A and B) in which K+ kinetics were performed in two selected groups of our HD population: Study A was designed to investigate above all the isolated effect of the factor time t on the removal and kinetic behaviour of K+; Study B was designed to identify and rank the other factors determining the intradialysis K+MB: mean plasma K+, Δ plasma K+, area under the curve (AUC) of the hourly inlet dialyser diffusion concentration gradient during every single HD session of K+ (hcgAUCK+), mean blood pH, Δ mean blood pH, AUC of blood pH, mean blood bicarbonates, Δ mean blood bicarbonates, AUC of blood bicarbonates, age, gender, dialysis vintage, volume of ultrafiltration (VUF), and pre- and post-dialysis body weight. The present studies were approved by our institutional review board and were conducted in accordance with good clinical practice guidelines and ethical principles of the Helsinki Declaration. The inclusion criteria were (i) standard bicarbonate HD treatment since at least 6 months; (ii) uncomplicated HD sessions. All the HD sessions in both studies A and B used the GENIUS® single-pass batch dialysis system (Fresenius Medical Care, Bad Homburg, Germany). The characteristics of the GENIUS® dialysis system have been described elsewhere [3, 4]. Briefly, it uses a double-sided roller pump that generates equal blood and dialysate flows up to 350 mL/min. The system consists of a closed dialysate tank of 90 L and although fresh and spent dialysate are stored together, there is no mixing of fresh and spent dialysate. A volumetric ultrafiltration pump drains the excess body water (the programmed weight loss) from the spent dialysate volume and collects it into the ultrafiltrate recipient. Thus, the ultrafiltrate in this single-pass batch dialysis system is part of the spent dialysate volume drained out from the closed dialysate tank of 90 L [3, 4]. Highflux FX80 dialysers (Fresenius Medical Care) were used in all sessions. Characteristics of the dialyser were described elsewhere [3]. Dialysate composition was as follows: magnesium 0.5.0 mmol/L; potassium 2.0 mmol/L; sodium 140 mmol/L; total calcium 1.5 mmol/L; bicarbonate 35 mmol/L; chloride 113 mmol/L; glucose 5.55 mmol/L and citrate 0.10 mmol/L. Kinetic studies Study A. A group of 11 stable prevalent Caucasian anuric patients (9 males and 2 females, mean age 54.1 ± 17.8 SD years, dialysis vintage 78.0 ± 60.2 months) was enrolled. After obtaining informed consent, they underwent one standard (∼4 h) and one long-hour (∼8 h) slow-flow bicarbonate HD session in a random sequence, always at the midweek interval, at least one week apart. The sessions were pair-matched as far as the dialysate and blood volume processed (90 L) and VUF
2
Study B. A group of 63 stable prevalent Caucasian anuric patients (40 males and 23 females, mean age 55.3 ± 14.7 years, dialysis vintage 55.6 ± 50.9 months) was enrolled for this study. After obtaining informed consent, participants underwent intradialysis kinetic studies. They consisted of one data recording session (at the midweek interval) for each of the 63 patients. The sessions lasted 4 h and were standardized as far as the dialysate and blood volume processed (90 L) are concerned. Blood and dialysate sampling Study A. Blood samples were obtained from the inlet blood tubing immediately before the onset of dialysis and at t60, t120, t180 min and at end of the 4 and 8 h sessions for the measurement of plasma K+, blood bicarbonates and blood pH. Additional blood samples were obtained at t360 min for the 8 h sessions. Dialysate was collected from the inlet dialysate tubing at the same time points as blood samples. Study B. Blood samples were obtained from the inlet blood tubing immediately before the onset of dialysis and at t60, t120, t180 min and at end of the sessions (t240 min) for the measurement of plasma K+, blood bicarbonates and blood pH. Dialysate was collected from the inlet dialysate tubing at the same time points as blood samples. Furthermore, at the end of dialysis (studies A and B), two samples were taken, respectively, from the ultrafiltrate recipient and from the dialysate tank, after thorough mixing, in order to quantify solute concentration in total spent dialysate. Particular attention was paid to the thorough mixing of the spent dialysate, strictly adhering to the ad hoc instructions present in the GENIUS® operator’s manual. Measurements Blood samples collected at the times described were used for the immediate measurement of plasma K+ concentrations by means of direct potentiometry with an ion-selective electrode and pH by means of an ABL 800 series blood gas analyser (Radiometer Medical ApS, Bronshoj, Denmark). K+ concentrations were measured in the fresh and spent dialysate and in the ultrafiltrate recipient by using the dialysate mode of the same instrument. Blood bicarbonates were calculated using the Henderson-Hasselbalch equation, assuming a pK of 6.1 and a solubility coefficient for carbon dioxide of 0.0301. The above parameters allowed the calculation of, in each patient, the area under the curve (AUC) of the hourly inlet dialyser diffusion concentration gradients between K+ concentrations in the plasma and in the dialysate (hcgAUCK+), AUC of blood bicarbonates and AUC of blood pH by using the trapezoidal rule [6, 7]. To calculate K+MBs, K+ concentrations in the inlet dialysate were considered those given by the average of the hourly
C. Basile et al.
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
ORIGINAL ARTICLE
SUBJECTS AND METHODS
are concerned. It must be underlined that all data of Study A were extracted from the same database from which data related to removal of uraemic retention solutes in standard bicarbonate HD and long-hour slow-flow bicarbonate HD were extracted and already published [5].
measurements. Furthermore, K+ concentrations in total spent dialysate and in the ultrafiltrate recipient were considered those given by the average of five consecutive measurements performed on the same sample. K+MB was calculated as follows: Kþ MB ¼ ½Kþ concentration in the fresh dialysate 90 ½ðKþ concentration in the spent dialysate 90Þ þ ðKþ concentration in the ultrafiltrate recipient VUF Þ where 90 is the volume (L) of both fresh and spent dialysate.
Study A Table 1 shows the comparison between the mean values of the HD data in the two sessions (4 and 8 h) performed on each of the 11 patients. As said, the sessions were pairmatched as far as the dialysate and blood volume processed (90 L) and VUF are concerned. Thus, the duration of treatment was dictated by the achievement of the target dialysate and blood volume processed (i.e. 90 L): for this reason, it was slightly different from 4 and 8 h. As expected, blood flow rate, dialysate flow rate and ultrafiltration rate were statistically significantly different when comparing 4- and 8-h sessions (P = 0.0001). Notably, K+MB was significantly higher in the 8 h sessions (4 h: −88.4 ± 23.2 mmol versus 8 h: −101.9 ± 32.2 mmol; P = 0.02) (Table 2), even if plasma K+ concentrations were not statistically significantly different both at the start and at the end of the 4- and 8-h sessions (Table 2). The same was
Intradialysis potassium mass balance
4h
8h
P*
Treatment time (min) Blood flow rate (mL/min) Blood volume processed (L) Dialysate flow rate (mL/min) Dialysate volume processed (L) Pre-dialysis body weight (kg) Post-dialysis body weight (kg) Ultrafiltration rate (mL/min) Volume of ultrafiltration (L) Inlet dialyser K+ concentration (mmol/L)
257.7 (1.1) 469.1 (2.8) 350 (0) 190 (0) 90 (0) 90 (0) 350 (0) 190 (0) 90 (0) 90 (0) 72.2 (10.8) 71.9 (10.7) 69.2 (10.3) 69.1 (10.1) 11.3 (3.1) 6.2 (1.9) 2.9 (0.8) 2.9 (0.9) 2.0 (0.2) 2.0 (0.2)
0.0001 0.0001 NS 0.0001 NS NS NS 0.0001 NS NS
Mean(SD). NS, not significant. *Student’s t-test for unpaired data.
Table 2. Comparison of dialysis parameters at the start and the end of the 4 and 8 h sessions of Study A Parameter Plasma K+ concentration (mmol/L) at T0 Plasma K+ concentration (mmol/L) at the end of dialysis Blood bicarbonate concentration (mmol/L) at T0 Blood bicarbonate concentration (mmol/L) at the end of dialysis Blood pH at T0 Blood pH at the end of dialysis K+MB (mmol) K+ mass removed by ultrafiltration (mmol)
4h
8h
P*
5.49 (0.70)
5.49 (0.70) NS
3.73 (0.37)
3.75 (0.35) NS
20.0 (2.12)
20.0 (2.05) NS
5.4 (1.2)
25.6 (1.1)
7.38 (0.04) 7.37 (0.04) 7.45 (0.02) 7.45 (0.02) −88.4 (23.2) −101.9 (32.2) −5.51 (0.7) −6.35 (0.3)
ORIGINAL ARTICLE
R E S U LT S
Parameter
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
Statistical analyses Data are reported as mean ± SD, median and interquartile ranges and 95% confidence intervals when needed. Study A: the mean values of plasma K+, blood bicarbonates and blood pH at the start and end of each HD session and K+MBs and HD treatment characteristics of patients enrolled in study A were compared by Student’s t-test for unpaired data. A generalized longitudinal mixed model for repeated measures with an autoregressive covariance structure was built for the study of comparisons of plasma K+ levels, blood pH and bicarbonates trends between 4 and 8 h sessions. The model for kinetic studies included the times of HD sessions (5 levels) and used the Wilks λ as multivariate test for the significance of each effect studied and Mauchly sphericity test of the matrix. In case of no sphericity of the matrix, the lower-bound ε was considered for the adjustment of the degrees of freedom. Study B: the one-way ANOVA for repeated measures followed by Bonferroni’s post hoc test was used for kinetic studies. The significance of the relationships between K+MBs and the variables studied was tested by means of the linear bivariate regression analysis in both studies A and B. All the variables significantly correlated to K+MB were considered as independent variables in a multiple linear regression model with K+MB as dependent variable in both studies. All statistical inferences were made using SPSS, version 11.0 software (SPSS, Inc., Chicago, IL, USA) and a P < 0.05 was considered statistically significant.
Table 1. Dialysis data of the HD sessions of Study A
NS NS NS 0.02 NS
Mean(SD). NS, not significant. *Student’s t-test for unpaired data.
true for blood pH and bicarbonate levels (Table 2). Furthermore, Figure 1 shows the trends of plasma K+ concentrations, blood pH and blood bicarbonate levels in the 4- and 8-h sessions. In order to find out the determinants of K+ removal by HD, bivariate linear regression analyses between K+MB and the parameters quoted in the Subjects and Methods section were performed. As reported in Table 3, only mean plasma K+, hcgAUCK+, mean blood bicarbonates and AUC of blood bicarbonates were significantly related to K+MB both in 4- and 8-h sessions. It must be pointed out that the relationships between K+MBs and mean blood bicarbonates and between K+MBs and AUC of blood bicarbonates were inversely related. However, only mean plasma K+ concentration and hcgAUCK+ beyond the duration of HD sessions per se kept their statistical significance when tested in a multiple linear regression output with K+MB as dependent variable (P = 0.007) (Table 4).
Study B Table 5 shows the dialysis data of the 63 HD sessions including also K+MBs.
3
ORIGINAL ARTICLE values are plotted. By applying the generalized longitudinal mixed model for repeated measures with an autoregressive covariance structure built for this study, the variations of the three parameters resulted smoother in the 8 h sessions even if the values at the end of the sessions were not statistically significantly different. In particular, in the 4 h sessions: the values of plasma K+ concentrations resulted significantly lower at T1, T2 and T3, blood pH significantly higher at T1, T2, T3 and T4, blood bicarbonates significantly higher at T1 and T2.
Trends of hourly plasma K+ concentrations, blood pH and blood bicarbonate concentrations are shown in Figures 2a–c, respectively. In order to find out the determinants of K+ removal by HD, bivariate linear regression analyses between K+MB and the parameters quoted in the Subjects and Methods section were performed. As reported in Table 6 and in Figures 3a and b, K+MB were statistically significantly related with mean intradialysis plasma K+ (R 2 = 0.4993, P = 0.0001, Figure 3a), hcgAUCK+ (R 2 = 0.5879, P = 0.0001, Figure 3b) and mean
4
blood bicarbonates. It must be pointed out that the relationship between K+MBs and mean blood bicarbonates was inversely related. Among these significantly related variables at the bivariate analysis, only mean intradialysis plasma K+ (P = 0.0001) and hcgAUCK+ (P = 0.036) predicted K+MB at the multiple linear regression analysis (Table 7). Finally, the distribution of K+MB, blood bicarbonates and blood pH in our population was analysed in order to verify if the lack of relationship among them was dependent by the small ranges of variations of these parameters during the study
C. Basile et al.
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
F I G U R E 1 : Comparison of hourly (T) trends of plasma K+, blood pH and bicarbonate levels between 4 and 8 h sessions (Study A). Mean
Table 3. Study A: output of the significant bivariate linear regressionsa with K+MB as dependent variable 4h
+
Mean plasma K concentration (mmol/L) Mean blood bicarbonate concentration (mmol/L) hcgAUCK+ (mmol/L × min)b AUC of bood bicarbonatesc (mmol/L × min)
8h
M
B
R2
P
M
B
R2
P
37.95 −11.56 0.17 −0.05
−70.89 363.90 4.70 −9.51
0.61 0.45 0.81 0.33
0.004 0.023 0.0002 0.05
43.31 −19.58 0.09 −0.04
−99.83 567.8 3.32 −29.9
0.80 0.68 0.94 0.61
0.0002 0.0019 0.0001 0.002
a
Model: Y = mX + b. Area under the curve of the hourly inlet dialyser diffusion concentration gradients between K+ concentrations in plasma and dialysate. c Area under the curve of the hourly blood bicarbonate concentrations. b
Table 4. Study A: multiple linear regression output with K+MB as dependent variable
+
P
36.50 −8.71 4.17
0.0002 0.001 0.02
Model: R 2 = 0.75; P = 0.007. a Coefficient of regression. b Area under the curve of the hourly inlet dialyser diffusion concentration gradients between K+ concentrations in plasma and dialysate.
Treatment time (min) Blood flow rate (mL/min) Blood volume processed (L) Dialysate flow rate (mL/min) Dialysate volume processed (L) Pre-dialysis body weight (kg) Post-dialysis body weight (kg) Ultrafiltration rate (mL/min) Volume of ultrafiltration (L) Inlet dialyser K+ concentration (mmol/L) K+MB (mmol) K+ mass removed by ultrafiltration (mmol)
240 (0.0) 350 (0) 90 (0) 350 (0) 90 (0) 79.0 (21.3) 76.4 (19.7) 12.9 (3.7) 3.1 (0.9) 2.0 (0.2) −86.7 (22.6) −5.76 (0.6)
Mean(SD).
procedures. The median values of mean plasma K+ (4.3 mmol/L), blood pH (7.43) and blood bicarbonates (23.5 mmol/L) during the sessions were respectively 49, 5 and 20% of their ranges. Thus, the three distributions were different and that confirms the lack of relationship among them.
DISCUSSION The present study confirms the behaviour of K+ removal described by others [8, 9]. Plasma K+ concentration rapidly decreased during the first 60 min and stabilized during the last 60 min of dialysis. Plasma K+ reached a steady state during the last hour of dialysis, while K+ continued to emerge into the dialysate. Therefore, it can be assumed that K+ removal rate was equal to the intra- to extracellular mass transfer rate at these time points (Figures 1a and 2a). Many are the factors determining the intradialysis K+MB: Sam et al. identified some of them in their review: dialysate K+
Intradialysis potassium mass balance
Duration of HD session per se Study A was designed to investigate the isolated effect of the factor time t and then of the duration per se of the dialysis session (by processing the same total blood and dialysate volume in two different time schedules) on the kinetics of plasma K+, blood pH and bicarbonates. They had different intradialysis kinetics, but the comparison of end-dialysis data showed no difference between 4- and 8-h sessions. However, the analysis of K+MBs showed a statistically significant larger K+ removal in 8-h HD sessions (Δ 13.56 mmol, equivalent to an increased removal of 15.34%, P = 0.02). The finding is novel as far as K+MB is concerned; however, it confirms consistent data about the important role played by the duration of treatment on the removal of small and middle molecules [3, 5].
5
ORIGINAL ARTICLE
Table 5. Dialysis data of the 63 HD sessions of Study B
Plasma K+ → dialysate K+ gradient Among the significantly related variables at the bivariate analyses in both studies A and B, only mean intradialysis plasma K+ and hcgAUCK+ predicted K+MB at the multiple linear regression analyses (Tables 4 and 7). Thus, our study confirms that the rate of K+ removal during dialysis is largely a function of the pre-dialysis plasma K+ concentration. The higher the initial plasma concentration, the greater the gradient between the plasma and the dialysate and, hence, the greater K+ removal [10]. Actually, Zehnder et al. showed in a prospective, randomized, cross-over study that a 0 mmol/L dialysate K+ concentration was able to remove more K+ than 1 or 2 mmol/L dialysate K+ concentration (P < 0.001) [9]. While confirming the data by Zendher et al., i.e. that the rate of K+ removal from the plasma depends on the gradient between plasma and dialysate K+ concentrations, at variance with their results, we demonstrated that mean intradialysis plasma K+ and hcgAUCK+ were able to predict K+MB at the multiple linear regression analysis [9].
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
Mean plasma K concentration (mmol/L) hcgAUCK+ (mmol/L × min)b Duration of HD session (min)
Ba
concentration, the efficiency of the dialyser, the duration and frequency of dialysis [10]. Furthermore, Fissell and Hakim underlined in their recent review that dialysis treatment lowers plasma K+, both by removal of K+ with dialysate and by rapid shift of K+ from the extracellular to the intracellular space as metabolic acidosis is treated [11]. Main aim of the present study was to identify and rank the factors determining the intradialysis K+MB in bicarbonate HD. The present results allow us to formulate the following ranking:
ORIGINAL ARTICLE peated measures followed by Bonferroni’s post hoc test was statistically significant for the effect of time points of HD sessions showing a gradual stepwise decrease in plasma K+ concentration, and a gradual stepwise increase in blood pH and blood bicarbonate levels. Table 6. Study B: output of the significant bivariate linear regressionsa with K+MB as dependent variable M Mean plasma K+ concentration (mmol/L) Mean blood bicarbonate concentration (mmol/L) hcgAUCK+ (mmol/L × min)b
R2
B
P
30.87 −44.34 0.50 0.0001 −4.74 195.6 0.12
0.47 0.03
17.52 0.59 0.0001
a
Model: Y = mX + b. Area under the curve of the hourly inlet dialyser diffusion concentration gradients between K+ concentrations in plasma and dialysate. b
Acid-base balance Unlike the extracellular K+, which can freely pass across the dialysis membrane, the intracellular K+ is slow to move into the extracellular space. This latter movement is influenced by a variety of factors, which can change during a dialysis procedure [12, 13]. Alkalosis causes a shift of K+ into cells and acidosis results in a K+ efflux from cells. Introduction of buffer
6
base into blood during dialysis promotes cellular uptake of K+ and thereby attenuates the dialytic removal of K+ (this is more evident in an acidotic patient). There are case reports describing that dialysis succeeded in reducing plasma K+ concentrations even though the dialysate K+ levels were higher than the pre-dialysis plasma K+ values. The decline in plasma K+ concentration was associated with a corresponding dialysisinduced rise in blood pH [12]. Finally, a randomized controlled trial showed an association between higher dialysate bicarbonate concentration and a faster decrease in intradialysis plasma K+ concentration [14]. The latter study was not able to show any statistically significant difference in K+MBs when utilizing three different dialysate bicarbonate concentrations, but the non-significant data may simply reflect the low statistical power of the study (only eight patients enrolled) [14]. In our hands, the role played by intradialysis blood pH and bicarbonates in determining K+MB seems to be minor: in both studies A and B the following bivariate regression analyses related to acid-base balance were performed utilizing K+MB as
C. Basile et al.
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
+
F I G U R E 2 : Trends of hourly (T) plasma K concentrations, blood pH and blood bicarbonate levels (Study B). The one-way ANOVA for re-
ORIGINAL ARTICLE
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
F I G U R E 3 : Study B: (a) scatter plot of net K+MB during every single HD session (y axis) versus the mean intradialysis plasma K+ concentra-
tion during every single HD session; (b) scatter plot of net K+MB during every single HD session (y axis) versus the area under the curve (AUC) of the hourly inlet dialyser diffusion concentration gradient during every single HD session ( plasma K+ concentration − dialysate K+ concentration). Best fit linear regressions are shown.
Table 7. Study B: multiple linear regression output with K+MB as dependent variable B +
Mean plasma K 28.88 concentration (mmol/L) hcgAUCK+ (mmol/L × min)a 3.21E-02
SE
P
95% CI
4.246 0.0001 20.35; 37.39 0.015 0.036
0.002; 0.062
a
Area under the curve of the hourly inlet dialyser diffusion concentration gradients between K+ concentrations in plasma and dialysate.
dependent variable and mean blood pH, Δ blood pH, AUC of blood pH, mean blood bicarbonates, Δ blood bicarbonates and AUC of blood bicarbonates as independent variables. Only mean blood bicarbonates and AUC of blood bicarbonates were significantly related to K+MB in Study A and only mean blood bicarbonates in Study B. However, none of them kept its
Intradialysis potassium mass balance
statistical significance when tested into a multiple linear regression output with K+MB as dependent variable. K+ removal during HD can occur through diffusion and convection. Current prescribing practices for chronic intermittent HD rely primarily on diffusive and less on convective losses [8, 9]. Our studies confirm these data (Tables 2 and 5). Thus, intradialysis K+ kinetics is quite different from that of sodium, in which convection accounts for ∼80% of intradialytic sodium mass balance, while the diffusive gradient between plasma and the inlet dialyser sodium concentration is an important factor in the ‘fine tuning’ of sodium mass balance [15]. Limitations of the study Our studies dealt with two small populations (11 patients in Study A and 63 patients in Study B). Thus, results, all the key covariates taken into account notwithstanding, cannot be
7
Conclusions The true challenge in HD patients is to avoid both lifethreatening pre-dialysis hyperkalaemia ( plasma K+ level > 6 mmol/L) and post-dialysis relative hypokalaemia (or at least very rapid decrease of plasma K+ level, and the related risk of lethal arrhythmias): resins (calcium or sodium polystyrene sulfonate) may be used; alternative strategies, such as longer or more frequent HD sessions and/or dialysate K+ profiling [17], may be required in such cases. Our studies enabled the establishment of the ranking of factors determining intradialysis K+MB: plasma K+ → dialysate K+ gradient is the main determinant; acid-base balance plays a much less important role. The duration of HD session per se is an independent determinant of K+MB.
C O N F L I C T O F I N T E R E S T S TAT E M E N T None declared.
REFERENCES 1. Green D, Roberts PR, New DI et al. Sudden cardiac death in hemodialysis patients: an in-depth review. Am J Kidney Dis 2011; 57: 921–929
8
2. US Renal Data System: USRDS 2012. Annual Data Report. National Institutes of Health, National Institutes of Diabetes and Digestive and Kidney Diseases, Bethesda, MD, 2012, accessible from http://www.usrds.org/2012/ view/v2 3. Eloot S, Van Biesen W, Dhondt A et al. Impact of hemodialysis duration on the removal of uremic retention solutes. Kidney Int 2008; 73: 765–770 4. Basile C, Libutti P, Di Turo AL et al. Effect of dialysate calcium concentrations on parathyroid hormone and calcium balance during a single dialysis session using bicarbonate hemodialysis: a crossover clinical trial. Am J Kidney Dis 2012; 59: 92–101 5. Basile C, Libutti P, Di Turo AL et al. Removal of uraemic retention solutes in standard bicarbonate haemodialysis and long-hour slow-flow bicarbonate haemodialysis. Nephrol Dial Transplant 2011; 26: 1296–1303 6. Yeh KC, Kwan KC. A comparison of numerical integrating algorithms by trapezoidal, Lagrange, and spline approximation. J Pharmacokinet Biopharm 1978; 6: 79–98 7. Eckardt K-U, Kim J, Kronenberg F et al. Hemoglobin variability does not predict mortality in European hemodialysis patients. J Am Soc Nephrol 2010; 21: 1765–1775 8. Feig PU, Shook A, Sterns RH. Effect of potassium removal during hemodialysis on the plasma potassium concentration. Nephron 1981; 27: 25–30 9. Zehnder C, Gutzwiller J-P, Huber A et al. Low-potassium and glucosefree dialysis maintains urea but enhances potassium removal. Nephrol Dial Transplant 2001; 16: 78–84 10. Sam R, Vaseemuddin M, Leong WH et al. Composition and clinical use of hemodialysates. Hemodial Int 2006; 10: 15–28 11. Fissell R, Hakim RM. Improving outcomes by changing hemodialysis practice patterns. Curr Opin Nephrol Hypertens 2013; 22: 675–680 12. Weigand C, Davin T, Raij L et al. Life threatening hypokalemia during hemodialysis. Trans Am Soc Artif Intern Organs 1975; 25: 416–418 13. Ward RA, Wathen RL, Williams TE et al. Hemodialysate composition and intradialytic metabolic, acid base and potassium changes. Kidney Int 1987; 32: 129–135 14. Heguilén RM, Sciurano C, Bellusci AD et al. The faster potassium-lowering effect of high dialysate bicarbonate concentrations in chronic haemodialysis patients. Nephrol Dial Transplant 2005; 20: 591–597 15. Basile C, Libutti P, Lisi P et al. Sodium setpoint and gradient in bicarbonate hemodialysis. J Nephrol 2013; 26: 1136–1142 16. Mathialahan T, Maclennan KA, Sandle LN et al. Enhanced large intestinal potassium permeability in end-stage renal disease. J Pathol 2005: 206: 46–51 17. Santoro A, Mancini E, London G et al. Patients with complex arrhytmias during and after haemodialysis suffer from different regimens of potassium removal. Nephrol Dial Transplant 2008; 23: 1415–1421
Received for publication: 21.8.2014; Accepted in revised form: 16.11.2014
C. Basile et al.
Downloaded from http://ndt.oxfordjournals.org/ at McGill University Libraries on January 6, 2015
ORIGINAL ARTICLE
extrapolated to populations not included in these studies, because of the low numerosity of the population studied. Furthermore, it is well known that the colon contributes considerably to K+ removal in dialysis patients, with colonic disposal being ∼30% of the dietary intake, a value that is about three times higher than normal [16]. The colonic K+ excretion was not studied. Thus, we are unable to say if colonic excretion differed among the patients of our studies. Finally, it is also well known that insulin and blood glucose affect K+ shifts. Seven out of the 74 patients under study (9.5%) were insulin-dependent diabetics. Blood glucose evaluation during the experimental HD sessions was not included into the study design. Thus, no blood glucose data are available; anyway, no patient received insulin during or in the peri-dialytic timeframe.
