Effects of acute variation of dialysate calcium concentrations on arterial stiffness and aortic pressure waveform

Abstract

Background. Abnormal mineral metabolism in chronic kidney disease plays a critical role in vascular calcification and arterial stiffness. The impact of presently used dialysis calcium concentration (D_Ca) on arterial stiffness and aortic pressure waveform has never been studied. The aim of the present study is to evaluate, in haemodialysis (HD) patients, the impact of acute modification of D_Ca on arterial stiffness and central pulse wave profile (cPWP).

Method. A randomized Latin square cross-over study was used to evaluate the three different concentrations of D_Ca (1.00, 1.25 and 1.50 mmol/L) during the second HD of the week for 3 consecutive weeks. Subjects returned to their baseline D_Ca for the following two treatments, allowing for a 7-day washout period between each experimental HD. cPWP, carotido-radial (c-r) and carotido-femoral (c-f) pulse wave velocities (PWV), plasma level of ionized calcium (iCa) and intact parathyroid hormone (PTH) were measured prior to and immediately after each experimental HD session. Data were analysed by the general linear model for repeated measures and by the general linear mixed model.

Results. Eighteen patients with a mean age of 48.9 ± 18 years and a median duration of HD of 8.7 months (range 1–87 months) completed the study. In post-HD, iCa decreased with D_Ca of 1.00 mmol/L (−0.14 ± 0.04 mmol/L, P < 0.001), increased with a D_Ca of 1.50 mmol/L (0.10 ± 0.06 mmol/L, P < 0.001) but did not change with a D_Ca of 1.25 mmol/L. Tests of within-subject contrast showed a linear relationship between higher D_Ca and a higher post-HD Δc-f PWV, Δc-r PWV and Δmean BP (P < 0.001, P = 0.008 and P = 0.002, respectively). Heart rate-adjusted central augmentation index (AIx) decreased significantly after HD, but was not related to D_Ca. The timing of wave refection (Tr) occurred earlier after dialysis resulting in a linear relationship between higher D_Ca and post-HD earlier Tr (P < 0.044). In a multivariate linear-mixed model for repeated measures, the percentage increase in c-f PWV and c-r PWV was significantly associated with the increasing level of iCa, whereas the increasing level of ΔMBP was not significant. In contrast, the percentage decrease in Tr (earlier wave reflection) was determined by higher ΔMBP and higher ultrafiltration, whereas the relative change in AIx was inversely determined by the variation in the heart rate and directly by ΔMBP.

Conclusion. We conclude that D_ca and acute changes in the serum iCa concentration, even within physiological range, are associated with detectable changes of arterial stiffness and cPWP. Long-term studies are necessary to evaluate the long-term effects of D_Ca modulation on arterial stiffness.

Introduction

The high prevalence of cardiovascular disease among haemodialysis (HD) patients cannot be explained solely by traditional cardiovascular risk factors. More recently, arterial stiffness, as measured by carotido-femoral (c-f) pulse wave velocity (PWV), has been shown to be an independent predictor for cardiovascular morbidity and mortality in a HD population [1,2]. Physiologically, increased stiffness of the large elastic arteries leads to increased central pulse pressure (PP), cardiac workload and left ventricular hypertrophy [3–7]. The mechanism of arterial stiffness in advanced chronic kidney disease (CKD) remains poorly understood. Alteration of the vascular wall by chronic effects of hypertension, accumulation of advanced glycation end-products and abnormal mineral metabolism are some of the mechanisms that have been proposed to explain the increased arterial stiffness in CKD [8–11]. The relationship between calcium containing phosphate binders and vascular calcification has raised concern about the long-term effects of chronic positive calcium balance [12,13]. However, the concerns about calcium containing phosphate binders have overshadowed the importance of dialysate calcium concentration (D_Ca) in the overall calcium balance. Although, DOQI guidelines recommend a D_Ca of 1.25 mmol/L, the optimal D_Ca could also be determined by other factors such as the use of specific phosphate binders, vitamin D or vitamin D analogues.

Although logical, the direct relationship between arterial calcification and arterial stiffness has recently been challenged [14,15]. The aim of the present study is to evaluate the impact of acute modification of calcaemia, within the physiological range, on arterial stiffness, as evaluated by PWV, and on the central pulse wave profile (cPWP) in a HD population. In this experimental protocol, each subject underwent three experimental HD sessions with random allocation of D_Ca of 1.00, 1.25 or 1.50 mmol/L.

Subjects and methods

Study design and patient population

This study took place at the Centre Hospitalier Universitaire de Québec—L’ Hôtel-Dieu de Québec Hospital, over a 7-month period in 2007. This was a Latin square cross-over study using three different concentrations of D_Ca for the second HD of the week for 3 consecutive weeks. Subjects returned to their baseline D_Ca for the following two treatments, allowing for a 7-day washout period between each experimental HD. Each subject was randomly assigned to one of the three different sequences of D_Ca: sequence 1 (D_Ca of 1.00, 1.25, 1.50 mmol/L), sequence 2 (D_Ca of 1.25, 1.50 and 1.00 mmol/L) and sequence 3 (D_Ca of 1.50, 1.00 and 1.25 mmol/L). No other parameters of the HD prescription or medication were modified during these 3 weeks. Baseline and post-HD measurements of arterial haemodynamic and biochemical parameters were obtained just prior to the beginning and after termination of experimental HD. The study protocol was approved by the ethics committee of the institution and written consent was obtained from all study participants.

Patients were included if they were 18 years or older, were on chronic HD for more than 3 months with stable dry weight and BP, stable doses of antihypertensive medications and phosphate binders and without any changes in dialysis prescription over the preceding month. Patients were excluded if they had any clinical conditions that would hamper pre- or post-dialysis haemodynamic measurements such as arterial fibrillation, multiple intradialytic hypotensive episodes, severe vascular disease or interdialytic weight gain of >5% of total body weight. Patients with a history of parathyroidectomy or PTH levels of >800 ng/L were also excluded. Twenty-one chronic HD patients were enrolled. Three were excluded because of hypotension (n = 1), need to change blood pressure (BP) medication (n = 1) and withdrawal of consent (n = 1). Eighteen patients completed the study. HD was performed 3-times weekly with a filter of 2.1 m² surface area, a dialysis duration of 3–4 h per session and a blood flow of 350–400 mL/min. A bicarbonate-based buffer dialysis solution was used with sodium concentrations of 138–142 mmol/L, potassium concentrations of 1–4 mmol/L and a dialysate flow rate of 500–750 mL/ min. The causes of CKD were glomerulonephritis (n = 4), diabetic nephropathy (n = 2), obstructive nephropathy (n = 4), interstitial nephritis (n = 4), hypertensive nephroangiosclerosis (n = 2) and unknown (n = 2). Patients suffered from hypertension (n = 14), atherosclerotic coronary disease (n = 4) and peripheral atherosclerotic vascular disease (n = 1). One had had a stroke and three had type 2 diabetes. The patients used ACE inhibitors (n = 5), calcium-channel blockers (n = 10), AT1 receptor blockers (n = 5), β-blockers (n = 6), central antihypertensive agents (n = 3), antiarrhythmics (n = 1) and α₁-receptor antagonist (n = 1). Eight patients took a mean of 6.25 tablets of sevelamer HCl (800 mg) daily, and 15 patients took a mean of 2.5 tablets of oral calcium carbonate (500 mg) daily.

Haemodynamic measurements

The patient was positioned in the supine position and allowed to rest for 15 min prior to their haemodynamic measurements. Brachial artery BP was recorded using an automatic sphygmomanometer BPM-100 (BP-Tru, Coquitlam, Canada). BP was recorded six times, with a 2-min interval between each measurement, and the average of the last five measurements was used to determine the brachial systolic and diastolic blood pressure (SBP) (DBP) [16].

The radial pulse wave profile (rPWP) was recorded by applanation tonometry using the SphygmoCor system® (AtCor Medical Pty. Ltd, Sydney, Australia). The tonometer probe was positioned over the radial artery, and a recording of rPWP was obtained for a 10-s period. The rPWP was recalibrated with the systolic and diastolic brachial BP. Three consecutive recordings were performed. A cPWP was then derived from the rPWP using a generalized transfer function as previously validated [17]. Central SBP, DBP, mean BP (MBP), PP and the time of return of the reflected wave (Tr) were derived. Pressure and time of first peak (P1 and T1) and second peak (P2 and T2), and central augmented pressure (AP) were obtained. Central augmentation index (AIx) was computed (AP = P2−P1; AIx = (AP/PP) × 100) and corrected for a heart rate of 75 beats/minute. The mean cPWP for the entire group under each pre- and post-HD with varying calcium concentration was then constructed as a method to visually assess the modification of cPWP under each experimental HD. This was performed by computing group mean DBP, pressure at the end of systole (ESP), pressure of the first and second pressure peak and their respective timing, Tr and duration of cardiac cycle in pre- and post-HD.

Assessment of PWV was made using Complior® SP (Artech Medical, Pantin, France). We first recorded the carotido-radial (c-r) PWV by positioning one sensor over the radial artery and a second sensor over the carotid artery. The distance between the two sensors was measured with a measuring tape, and three consecutive recordings of both pulse waveforms were performed (8–10 heart beats for each recording). The Complior software automatically detected the foot of each pulse waveform from the two arterial sites and then measured the mean distance between the two feet as being the travel time of the wave. PWV was then computed using the following formula: PWV = travel distance/travel time, as previously validated [18]. C-f PWV was also determined using the same technique but positioning the second sensor on the femoral artery.

Biochemical measurements

Blood samples were obtained immediately before the start of each experimental HD session and at the end of each of these sessions. Ionized calcium (iCa), intact parathyroid hormone (PTH) and albumin were monitored. iCa was measured with an ion-selective electrode on a Nova PhoxPlus analyser. Intact PTH (1–84) was measured with the PTH stat assay from Roche diagnostics using two antibodies reactive with epitopes in the amino acid regions 26–32 and 37–42.

Data analysis

Data analysis was performed using the SPSS software (version 16.0 for Windows, SPSS Inc., Chicago, IL, USA). Data were expressed as means ± SD. To study the effect of HD, D_Ca and HD–D_Ca interaction on biochemical and haemodynamic variables, we used general linear model two-way analysis of variance for repeated measures. Sphericity was verified by Mauchly's test of sphericity, and in the case of sphericity deviation, the P-value was corrected according to the Greenhouse–Geisser epsilon coefficient for probability adjustment. This procedure was followed by a polynomial within-subject linear contrast to study the linearity between D_Ca and post-HD variation of biochemical and haemodynamic parameters. A general linear mixed model showed no significant sequence or carryover effects on haemodynamic parameters. To account for repeated observations, the general linear-mixed model was used to study the determinants of relative changes of c-f PWV, c-r PWV, AIx and Tr by entering ultrafiltration, Δheart rate, ΔMBP and ΔiCa and log of PTH ratio as covariates. Bonferroni correction was used for multiple comparisons between treatment groups when appropriate. A P-value of <0.05 was considered to be statistically significant.

Results

Eighteen patients with a mean age of 48.9 ± 18 years and a median duration of chronic HD of 8.7 months (range 1–87 months) completed the study. The participants included 5 women and 13 men. Baseline D_Ca were 1.00 mmol/L (n = 1), 1.25 mmol/L (n = 11) and 1.50 mmol/L (n = 6). Pre-HD levels of PTH, iCa, BP, c-r PWV, c-f PWV, AIx and Tr were not significantly different between each experimental HD session. The average ultrafiltration was 2.0 ± 1.0 L, 1.9 ± 1.1 L and 2.1 ± 1.1 L for D_Ca of 1.00, 1.25 and 1.50 mmol/L, respectively (P = 0.78 for repeated measures ANOVA). There were no significant effects of sequence on the extent of haemodynamic and biochemical parameters. Therefore, the residual effects of the previous D_Ca of one experimental HD session were considered to be negligible on the subsequent experimental HD.

A two-way analysis of variance for repeated measures was conducted to study the effects of HD, D_Ca and HD–D_Ca interaction on biochemical and haemodynamic variable. The results are presented in Tables 1 and 2.

Table 1.

Biochemical and haemodynamic parameters

	DCa (mmol/L)			Repeated measures ANOVA			Linear contrast
	1.00	1.25	Ca × HD	Ca	HD	HD × Ca	HD × Ca
iCa (mmol/L)
Pre-HD	1.17 ± 0.05	1.16 ± 0.05	1.17 ± 0.06	<0.001	0.09	<0.001	<0.001
Post-HD Δ	−0.14 ± 0.04	−0.02 ± 0.05	0.10 ± 0.06
PTH
Pre-HD	223 ± 170	220 ± 150	199 ± 145	<0.001	0.68	<0.001	<0.001
Post-HD Δ	237 ± 196	49 ± 205	−104 ± 103
Brachial SBP (mmHg)
Pre-HD	129 ± 18	125 ± 16	130 ± 19	0.003	0.29	0.001	0.001
Post-HD Δ	−10 ± 19	−5 ± 15	3 ± 14
Brachial DBP (mmHg)
Pre-HD	79 ± 9	79 ± 11	79 ± 11	0.03	0.75	0.005	0.014
Post-HD Δ	−2 ± 9	−3 ± 11	4 ± 8
Heart rate (b.p.m.)
Pre-HD	70 ± 12	70 ± 17	69 ± 10	0.20	0.74	0.46	0.28
Post-HD Δ	2 ± 13	0.4 ± 8	−0.7 ± 7
c-f PWV (m/s)
Pre-HD	11.6 ± 3.1	11 ± 1.8	11 ± 2.7	0.84	0.17	0.001	<0.001
Post-HD Δ	−0.8 ± 1.4	0.9 ± 1.9	0.8 ± 0.9
c-r PWV (m/s)
Pre-HD	10.3 ± 1.9	9.8 ± 1.6	9.8 ± 1.7	0.87	0.003	0.028	0.008
Post-HD Δ	0.16 ± 1.3	0.99 ± 1.4	1.2 ± 1.4

D_Ca, dialysate calcium concentration; iCa, ionized calcium; PTH, parathyroid hormone; SBP, systolic blood pressure; DBP, diastolic blood pressure; c-f PWV, carotido-femoral pulse wave velocity; c-r PWV, carotido-radial pulse wave velocity.

Table 2.

Central pulse wave profile parameters

	DCa (mmol/L)			Repeated measures ANOVA			Linear contrast
	1.00	1.25	1.50	Ca	HD	Ca × HD	Ca × HD
SBP (mmHg)
Pre-HD	120 ± 19	115 ± 15	120 ± 18	0.003	0.15	0.006	0.001
Post-HD Δ	−10 ± 21	−7 ± 16	0.8 ± 15
DBP (mmHg)
Pre-HD	80 ± 10	80 ± 11	80 ± 11	0.024	0.71	0.007	0.018
Post-HD Δ	−2 ± 9	−3 ± 11	3 ± 8
MBP (mmHg)
Pre-HD	97 ± 12	95 ± 12	97 ± 12	0.003	0.229	0.001	0.002
Post-HD Δ	−6 ± 11	−5 ± 13	2 ± 10
ESP (mmHg)
Pre-HD	108 ± 16	105 ± 14	109 ± 16	0.002	0.59	0.001	0.001
Post-HD Δ	−6 ± 17	−4 ± 15	5 ± 13
ED (ms)
Pre-HD	320 ± 20	317 ± 17	320 ± 19	0.75	<0.001	0.98	0.97
Post-HD Δ	−49 ± 37	−48 ± 31	−49 ± 27
AIx (%)
Pre-HD	22 ± 10	22 ± 9	22 ± 9	0.51	0.004	0.73	0.97
Post-HD Δ	−8 ± 14	−9 ± 11	−8 ± 11
Tr (ms)
Pre-HD	143 ± 11	145 ± 13	144 ± 12	0.10	<0.001	0.093	0.044
Post-HD Δ	−4 ± 8	−6 ± 5	−9 ± 6

Pre-HD, pre-haemodialysis value; post-HDΔ, absolute post-dialysis variation of each parameter; D_Ca, dialysate calcium concentration; SBP, systolic blood pressure; DBP, diastolic blood pressure; MBP, mean blood pressure; ESP, end systolic pressure; ED, ejection duration; AIx, heart augmentation index; Tr, time to the start of the reflected wave.

Biochemical parameters

HD per se had no significant effect on the levels of iCa and PTH. D_Ca and HD–D_Ca interaction were statistically significant for iCa and PTH (Table 1).

Arterial stiffness

HD was associated with the post-HD increase in c-r PWV (P = 0.003). D_Ca had a significant effect on systolic and diastolic BP (P = 0.003 and P = 0.03, respectively). There was a significant relationship between D_Ca and post-HD Δc-f PWV and Δc-r PWV (P = 0.001 and P = 0.028 for HD–D_Ca interaction). In addition, the tests of within-subject contrast showed a linear effect of D_Ca on post-HD Δc-f PWV and Δc-r PWV (P < 0.001 and P = 0.008, respectively).

Central pulse wave profile

Table 2 shows the effects of HD, D_Ca and HD–D_Ca interaction on central haemodynamic parameters as derived from cPWP. HD was associated with the post-HD reduction in heart rate adjusted AIx (P = 0.004), ejection duration (P < 0.001) and Tr (P < 0.001). D_Ca had a significant effect on central SBP (P = 0.003), DBP (P = 0.024), MBP (P = 0.003) and ESP (P = 0.002). The HD–D_Ca interaction was statistically significant for central SBP (P = 0.006), DBP (P = 0.007), MBP (P = 0.001) and ESP (P = 0.001).

A representative example of a cPWP from a middle-age patient suffering from CKD, as derived from the radial PWP, is presented in Figure 1. Using the average BP at various time points of cPWP, we constructed a pre- and post-HD mean cPWP for the sample population under each experimental HD (Figure 2). The use of D_Ca of 1.25 mmol/L did not affect any of the pressures. The tests of within-subject linear contrast showed a statistically significant effect of D_Ca on post-HD ΔDBP, ΔP1, ΔP2, ΔESP, Δmean pressure of diastole, ΔT1 and ΔTr (P < 0.05). However, no relationship was observed between D_Ca and post-HD changes in T2, ejection duration and heart rate.

Fig. 1.

Central pulse wave profile. The central pulse wave profile can be broken into the following parameters: first peak of pressure (P1), second peak of pressure (P2), time at P1 (T1), time at P2 (T2), time of return of the reflection wave (Tr), diastolic blood pressure (DBP), systolic blood pressure (SBP), pulse pressure (PP).

Fig. 2.

Mean pre- and post-haemodialysis central pressure waveform. Using the mean pressure points at corresponding time points of each patient (n = 18), the group's central pulse wave profile (cPWP) was constructed for each treatment group. The solid line and dotted lines indicate the pre- and post-HD cPWP, respectively. The top, middle and lower panels, respectively, refer to dialysate calcium concentrations of 1.00, 1.25 and 1.50 mmol/L.

Determinants of haemodynamic variation

In a multivariate linear-mixed model for repeated measures, the percentage increase in c-f PWV and c-r PWV was significantly associated with the increasing level of iCa, whereas the changes in the level of MBP were not significant (Table 3). However, the percentage decrease in Tr (earlier wave reflection) was determined mainly by higher ΔMBP and higher degree of ultrafiltration. The relative change in AIx was inversely determined by the variation in the heart rate and directly with ΔMBP.

Table 3.

Determinants of variation in arterial stiffness and wave reflection

Estimates of fixed effects
	Estimate (95% confidence interval)	P-value
%Δ c-f PWV
ΔiCa (mmol/L)	0.566 (0.018–1.114)	0.043
ΔMBP (mmHg)	0.002 (−0.002–0.006)	0.279
%Δ c-r PWV
ΔiCa (mmol/L)	0.654 (0.063–1.247)	0.031
ΔMBP (mmHg)	0.003 (−0.001–0.007)	0.131
%Δ Tr
ΔMBP (mmHg)	−0.001 (−0.002–0.0008)	0.013
UF (L)	−0.012 (−0.023 to −0.0003)	0.045
%Δ AIx
ΔMBP (mmHg)	0.597 (0.382–0.812)	<0.001
ΔHR (b.p.m.)	−0.420 (−0.660 to −0.179)	0.001

Linear-mixed model for repeated measure.

Covariates in the model include ultrafiltration (UF), changes in the heart rate (ΔHR), ionized calcium (ΔiCa), mean blood pressure (ΔMBP) and parathyroid hormone levels.

Discussion

In this study, we evaluated acute changes in arterial stiffness and cPWP during 3 HD sessions differing only by the D_ca that ranged from 1.00 to 1.50 mmol/L, thereby modulating serum iCa concentrations within the physiological range. Our results show that a rise in serum iCa during one HD session, even within the physiological range of calcaemia, is associated with an increase in PWV of both muscular and elastic type arteries. In addition, higher levels of calcaemia were associated with a lesser decrease in post-HD central BP and a more rapid return of the reflected wave (earlier Tr) without any significant change in AIx. Our findings are in keeping with previous studies comparing the haemodynamic effects of D_ca of 1.25 and 1.75 mmol/L. However, in light of growing concern regarding positive calcium balance, the D_ca of 1.75 mmol/L is no longer recommended for standard three times weekly dialysis sessions [19]. Therefore, there was a need to evaluate the haemodynamic effects of the presently used D_Ca.

PWV is an established method of evaluating segmental arterial stiffness and has repeatedly been associated with clinical outcomes, especially in a HD population [20,21]. The acute changes of PWV in our study, however, were transitory and so are believed to be functional changes as it is highly unlikely that they would induce acute structural changes. They are most likely the result of variation in vascular smooth muscle cell (VSMC) tone that is highly dependent on the extracellular calcium concentration. Earlier Tr could be explained by either an increase in PWV and/or proximalization of the reflection sites. However, using a linear-mixed model, we found no significant association between the degree of increase in either c-rPWV or c-f PWV and the degree of earlier wave reflection. Therefore, it seems reasonable to assume that the proximalization of arterial reflection sites is a greater determinant of earlier wave reflection in this experimental study. The proximalization of arterial reflection sites is also thought to result from enhanced vascular tone, which leads to a greater reduction in the lumen of the branching arteries. The fact that an earlier return of the wave reflection is positively associated with the degree of ultrafiltration during the HD session is in keeping with this hypothesis. In this study, AIx seems to be more influenced by variation in heart rate and MBP than by the change in calcaemia. In fact, AIx is known to be influenced by factors such as reflection sites, reflection coefficient and heart rate, therefore limiting its utility as a reliable marker of arterial stiffness [22,23].

Marchais et al. [24] have previously shown an increase in both aortic and brancial PWVs with a D_Ca of 1.75 mmol/L, but not with a D_Ca of 1.5 mmol/L. In addition, Kyriazis et al. [25] found an increase in stiffness index after one HD with D_ca of 1.75 mmol/L as evaluated by digital pulse volume, and Yoo et al. [26] observed an increase in carotid compliance after switching from D_ca of 1.75 to a D_ca of 1.25 mmol/L for eight consecutive HD sessions. After switching form a D_Ca of 1.75 to 1.25 mmol/L even for only one HD session, some investigators have shown a significant decrease in peripherial BP [27,28]. In contrast, Kyriazis et al. [29] failed to confirm these observations with only one HD session although they observed a reduction in PP and MBP after switching from D_ca 1.75 to 1.25 mmol/L for 10 consecutive HD sessions. There is compelling evidence to support that high extracellular calcium concentrations could induce arterial vasoconstriction in humans [30]. However, higher dialysate calcium concentrations have also been shown to increase cardiac contractility, potentially leading to a higher BP and different pulse profile through increased cardiac output [27,28,31]. In the present study, higher D_Ca was associated with a higher P1 and an earlier T1, suggesting an increase in myocardial contractility within this range of D_Ca. The only potential influence of cardiac contractility on arterial PWV could be mediated through the degree of change in arterial MBP. However, in multivariate analysis, the degree of change in the iCa concentration was the only significant determinant of relative change in c-r PWV and c-f PWV.

The haemodynamic changes that were observed in our study may not be solely attributed to the direct effects of iCa on the cardiovascular system. Since iCa has an acute effect on PTH concentrations, it may be possible that at least some of the effect could be mediated through the actions of PTH. Indeed, based on animal experiments, it has been suggested that acute PTH administration can result in the reduction of BP [32]. In this regard, PTH may act directly as a vasorelaxant on VSMC via cAMP-dependent inhibition of L-type Ca⁺⁺ channel currents [33]. It could be argued that in our study, the suppression of PTH by the acute rise in serum calcium concentration could play a significant role in the rise of BP. However, in multivariate analysis, the change in the PTH level was not significantly related to changes in arterial PWV.

In conclusion, this study showed that D_ca and acute changes in the serum iCa concentration, even within physiological range, are associated with detectable changes in arterial stiffness of both elastic (c-f segment) and muscular-type arteries (c-r segment), and in the PWP of the aorta. More studies are needed to evaluate the consequences of these repetitive effects on the long-term function and structure of arteries in the uraemic milieu.