Effect of Hemodialysis With Citrate on Vascular Calcification

Abstract

Introduction

Vascular calcification, specifically medial calcification, is prevalent in patients undergoing hemodialysis (HD) and leads to poor outcomes. Citrate is present in some HD solutions and could potentially inhibit vascular calcification. This was examined by comparing the progression of medial arterial calcification, as measured using mammography, in female patients with end-stage renal disease (ESRD) undergoing HD with and without citrate.

Methods

In a cross-over study, 28 female chronic in-center HD patients with preexisting breast arterial calcification (BAC) underwent HD using dialysate with or without citrate, each for 1 year with BAC measured at 6-month intervals. In addition, citrate was measured in calcified human arteries.

Results

HD with citrate increased post-HD plasma citrate over 200% with no changes in pre-HD levels of citrate, calcium (Ca), or phosphate. Progression of BAC was significantly greater during citrate HD (47, interquartile range [IQR]:10–124 vs. 9 , IQR: −4 to 34 mm/yr, P = 0.0005), with 79% of subjects having greater progression during citrate HD. Citrate was present in calcified human arteries and correlated with the degree of calcification (r = 0.91).

Conclusion

HD with citrate unexpectedly promoted vascular calcification, which could not be explained by other dialysis parameters or changes in mineral metabolism. In addition, citrate was associated with vascular calcifications at levels approaching those reported in bone. The results are consistent with known effects of citrate to promote biomineralization and bone formation. Further studies are needed to determine whether this effect of citrate leads to adverse clinical outcomes in patients undergoing HD and occurs with other therapeutic uses of citrate.

Graphical abstract

Cardiovascular disease is common and severe in ESRD, with a mortality 10 to 500 times higher than in the general population.¹ This increased risk remains after adjustment for sex, race, diabetes, and other risk factors for atherosclerosis; is not reduced by lipid-lowering therapy²; and is linked to altered mineral metabolism3, 4, 5, 6, 7, 8, 9, 10; pointing instead to medial arterial calcification, a lesion histologically, pathophysiologically, and clinically distinct from atherosclerosis. This lesion is common in chronic kidney disease and ESRD11, 12, 13 and associated with particularly poor cardiovascular outcomes.¹⁴

Citrate, a component of some HD solutions, is a direct inhibitor of Ca crystallization, independent of its chelation of Ca ions.^15,16 It is a potent inhibitor of Ca crystallization in urine, and is present in the circulation at concentrations that inhibit hydroxyapatite formation in vitro.^15,16 It is metabolized primarily in the liver, with a half-life of approximately 50 minutes in patients undergoing HD, no different than in normal subjects.¹⁷ In a cross-over study of 34 patients dialyzed once using standard acetate acid concentrate and once using acid concentrate containing 2.4 mM citrate instead, serum citrate did not change in the former but increased by 37% in the latter.¹⁸ HD with citrate decreased the serum calcification propensity score, a measure of Ca phosphate precipitation, compared with dialysis with acetate¹⁹; however, the relevance of this to vascular calcification is unclear.

Only a few studies have examined the effect of citrate on vascular calcification. Citrate reduced calcification in a model of vascular calcification in cultured rat aortas²⁰; however, there was no accounting for the expected decrease in ionized Ca (iCa) that could have explained this. Administration of citrate, presumably orally, to rats with renal failure attenuated vascular calcification²¹ but there were substantial decreases in serum creatinine, urea, Ca, and phosphorus that likely explained this. Thus, there have been no properly performed studies examining direct effects of citrate on vascular calcification, and no studies in humans.

Vascular calcification in humans is usually measured in coronary arteries or aorta; however, this is expensive and involves significant radiation exposure. More importantly, these vessels develop atherosclerotic calcification, which cannot be reliably distinguished radiologically from medial calcification. We have developed mammography as a tool for clinical studies of medial calcification. BAC is exclusively medial,²² easily discerned on routine mammograms (MGs), and correlates with peripheral arterial calcification²³ and clinical outcomes,²⁴ particularly in ESRD,²⁵ indicating that it is a marker of systemic medial calcification. We have shown that BAC can be reliably quantified to measure progression and discern the effect of risk factors.^26,27

Therefore, mammography was used to compare the progression of vascular calcification in patients undergoing chronic HD with and without citrate. Furthermore, citrate content was determined in calcified human arteries.

Methods

Patient Selection

The study population was females with ESRD undergoing maintenance HD at Emory outpatient dialysis centers who had baseline BAC. Exclusion criteria were current warfarin use, severe hyperparathyroidism (likely to undergo parathyroidectomy in the next 2 years), difficulty controlling serum Ca (likely to require changes in dialysate Ca concentration in the next 2 years), or a life expectancy < 2 years. After informed consent, subjects without a MG in the past 6 months underwent a screening MG. Subjects without any visible arterial calcification on the screening MG or one performed within the past 6 months were excluded from the study.

Study Protocol

This was a cross-over study comparing the effect of a standard acetate-based acid concentrate (Naturalyte, Fresenius Medical Care) and a citrate-based acid concentrate (Citrasate, Fresenius Medical Care) on the rate of BAC (ClincalTrials.gov # NCT04956120). Citrasate is the standard dialysate used for all patients at Emory dialysis centers. MGs with BAC performed within the past 6 months served as baseline studies and these subjects were necessarily assigned to the Citrasate arm. The remaining subjects with BAC were randomized so that half the final cohort started on each arm for the first year. Dialysate assignment was monitored by unit supervisors and checked 2 to 3 times/mo on clinical rounds. MGs were scheduled at 6 and 12 months, after which subjects were switched to the other acid concentrate for 1 year with MGs scheduled at 6 and 12 months. The 12-month MG served as the last study of the first arm and the baseline study of the second arm. This allowed for 3 measurements of BAC during each study period, thereby providing a more accurate determination of the progression. The primary outcome was the rate of progression of BAC during each period. The secondary outcome was changes in mineral metabolism as measured using blood chemistries. Pre- and post-HD blood samples were scheduled twice during each arm for measurement of serum magnesium (Mg) and iCa, and plasma citrate. The study was approved by the Institutional Review Board of Emory University.

Sample Size

Using previous data from 16 stable patients with 5 sequential measurements of BAC and assuming linear progression,^26,27 an SD of 11.9 mm/breast/yr was calculated for comparison of BAC rates of the first 3 and last 3 MGs (the design for this study). Based on the average rate of progression of 20.1 mm/breast/yr we previously determined in patients undergoing HD,²⁷ a cohort of 47 women was calculated to provide 80% power to detect a 25% change in rate at P = 0.05.

Blood Chemistries

Predialysis serum Ca, phosphorus, bicarbonate, and urea were measured monthly at an outside clinical laboratory per standard clinical care. Predialysis and postdialysis serum Mg and iCa were measured at the same outside clinical laboratory. Dialysis adequacy was assessed as kT/V, calculated using the Daugirdas II equation.²⁸

Predialysis and postdialysis blood samples for measurement of citrate were obtained in heparinized tubes and rapidly centrifuged to obtain plasma, which was immediately frozen. Citrate was measured in the investigator’s laboratory after centrifugation through a 30 kDa filter to remove interfering proteins, using an enzymatic assay based on conversion of citrate to oxaloacetate by citrate lyase and conversion of oxaloacetate to malate by malic dehydrogenase with reduction of nicotinamide adenine dinucleotide to reduced nicotinamide adenine dinucleotide, which was measured spectrophotometrically.

Measurement of Vascular Calcification

BAC was defined as calcifications overlying vessels that followed the path of the vessel and did not extend beyond the vessel wall (Figure 1). Potential calcifications were excluded when there was any uncertainty. Calcified vessels were traced and lengths measured on standard screening MGs as previously described²⁶ using Sectra PACS software (Sectra Medical), with blinding to the associated treatment arm. Measurements were performed after completion of the final MG for all the patients. Where possible, the right breast was used (14 subjects) with additional measurements of the left breast included when the total amount of BAC on the right was < 100 mm (6 subjects). The left breast was used in 6 subjects when there was inconsistent coverage of the right breast (3), a right chest catheter (1), or BAC that was too extensive on the right (2). In 2 subjects, a different breast was used for each arm due either to missing images or to mastectomy. Cranial-caudal views were used exclusively except when BAC was visualized only on the medial-lateral oblique images (2 subjects). Images were cropped before measurements when necessary to achieve consistency in breast coverage.

Figure 1.

Breast arterial calcification. First, third, and fifth mammogram from 1 subject who started on the acetate arm and finished on the citrate arm.

Citrate in Calcified Arteries

Arterial segments were obtained from lower extremity amputations in 3 patients, none of whom had ESRD. After flushing of the lumen and removal of adventitia, the segments were immediately frozen. After thawing, the segments underwent micro–computed tomography (microCT) using a Scanco μCT40 scanner that was calibrated weekly with a factory-supplied phantom (Scanco Medical, Bassersdorf, Switzerland). Scanning was at a voxel size of 6 μm (45 kVp and 177 μA, and 200 ms integration time). Total volume of mineral within the vessel wall was quantified from proximal to distal end (∼0.5 cm) using approximately 800 slices at a voxel size of 6 μm. Thresholding was performed using visual inspection and comparison of preview and slice-wise grayscale 2D images, per the manufacturer’s recommendations; and the same threshold value was used for all samples. The final result was bone volume / total volume, expressed as a percentage. Following microCT, arterial segments were placed in 0.5 – 1 ml of 0.1 M HCl for one week at 4^o C. The citrate concentration was then measured colorimetrically using an assay kit based on citrate lyase (Sigma-Aldrich, St. Louis, MO). The amount of calcification in each segment was estimated by multiplying the weight by the bone volume / total volume determined from the microCT.

Statistical Analysis

All data except for BAC are presented as means ± standard errors, with analysis using paired t testing. BAC data were not normally distributed and are presented as medians and IQR, with paired analysis using the Wilcoxon signed-rank test.

Results

Subjects

A total of 118 patients were consented for the study and 87 underwent screening, of whom 66 had BAC. Of these, 54 began the study and 25 withdrew or were withdrawn before the last MG of the second treatment arm. The reasons for withdrawal were death (8 patients); transplantation (3 patients); unit transfer (6 patients); prolonged hospitalization (2 patients); unable to travel to MG (2 patients); and 1 patient each due to illness, noncompliance, incorrect dialysate assignment, and unwillingness to complete study. All withdrawals were unrelated to the treatment. One subject expired before the fifth MG and was included in the analysis, resulting in a final cohort of 29 subjects who completed the study. The first subject was enrolled in May 2021, with the last subject completing treatment in March 2025, when the trial ended. One subject had inconsistent breast coverage that precluded measurement of BAC progression, resulting in a final cohort of 28. Of these, 12 subjects started on the citrate arm (Group 1) and 16 on the acetate arm (Group 2).

The characteristics of the subjects are presented in Table 1. Almost all the subjects were prescribed oral phosphate binders, with 21% receiving a Ca-based binder. All but 1 subject received a vitamin D analog, either paricalcitol or doxercalciferol, with no subjects prescribed calcitriol. Twenty subjects were treated with a calcimimetic (either cinacalcet or etecalcide). The number of subjects prescribed phosphate binders or vitamin D analogs did not change between treatment arms. There were 3 fewer subjects prescribed calcimimetics during the acetate arm. There were no significant differences in doses of any of the medications between the 2 arms. Dialysis parameters for each arm are shown in Table 2. Three subjects had changes in dialysate Ca concentration during the citrate arm and 1 during the acetate arm. In each case, the dialysate Ca for the entire arm was prorated based on the amount of time on each concentration. No changes were made in dialysate bicarbonate or Mg concentration. There were no differences between the arms. Corresponding data for each group are shown in Table 3. There were no significant differences between the groups.

Table 1.

Patient characteristics

Characteristics	All	Group 1	Group 2
Age (yr)	58.2 ± 2.1	55.7 ± 2.3	60.1 ± 3.1
Race (%AA)	100	100	100
ESRD duration (yr)	6.3 ± 0.8	7.3 ± 1.4	5.6 ± 0.9
Diabetes (%)	54	58	50
Phosphate binder (%)	96	100	94
Sevelamer	54	83	50
Ca acetate	17	17	25
SucroFe oxyhydroxide	14	8	19
Fe citrate	7	8	6
Vitamin D analog (%)	96	92	100
Calcimimetic (%)	61	58	63

AA, African American; Ca, calcium; ESRD, end-stage renal disease; Fe, iron.

There were no significant differences between groups. Means ± standard errors.

Table 2.

Relevant parameters on each arm of the study for the entire cohort

Parameter	Citrate arm	Acetate arm
Duration (d)	401 ± 9	413 ± 18
Age at end of arm (yr)	59.9 ± 2.1	59.8 ± 2.1
Dialysis time (min)	198 ± 4	199 ± 4
Mean kT/V	1.54 ± 0.03	1.55 ± 0.03
Mean dialysate Ca (mEq/l)	2.6 ± 0.04	2.6 ± 0.03
Dialysate Mg (mM)	0.5	0.5
Dialysate HCO3 (mM)	34.6	33.0
Dialysate acetate (mM)	0.3	4.0
Dialysate citrate (mM)	2.4	0

Ca, calcium; HCO3, hydrogen carbonate; Mg, magnesium.

There were no significant differences between arms by paired testing. Means ± standard errors.

Table 3.

Relevant parameters on each arm of the study for each group

Study arm	Group 1		Group 2
	Citrate	Acetate	Citrate	Acetate
Duration (d)	396 ± 14	438 ± 40	405 ± 11	394 ± 7
Age at end of arm (yr)	56.7± 2.3	57.9± 2.4	62.2 ± 3.2	61.1± 3.1
Dialysis time (min)	203 ± 6	204 ± 6	194 ± 5	196 ± 5
Mean kT/V	1.58 ± 0.06	1.59 ± 0.06	1.50 ± 0.04	1.53 ± 0.03
Mean dialysate Ca (mEq/l)	2.7 ± 0.1	2.6 ± 0.1	2.5 ± 0.0	2.5 ± 0.0

There were no significant differences between groups. Means ± standard errors.

Chemistries

Blood chemistries are shown in Table 4 for the entire cohort and Table 5 for each group. Postdialysis serum iCa was not measured in 1 subject during the citrate arm. On the acetate arm, predialysis serum iCa was missing on 1 subject, postdialysis iCa on 2 subjects, and both predialysis and postdialysis serum Mg on 2 subjects. Plasma citrate was not measured in 3 subjects who were at the furthest outlying dialysis center, because of logistical issues. Postdialysis iCa was significantly lower during the citrate arm, with no difference in predialysis iCa. Both predialysis and postdialysis serum Mg were slightly lower during dialysis with citrate. Predialysis serum bicarbonate and parathyroid hormone (PTH) were slightly greater during the citrate arm. PTH levels were higher in Group 2, significant only for the citrate arm. There was no correlation between serum bicarbonate, Mg, or PTH and progression of BAC on either arm (not shown).

Table 4.

Blood chemistries on each arm for the entire cohort

Chemistry		Citrate arm	Acetate arm	P
Serum Ca (mg/dl)	Pre-HD	9.0 ± 0.1 (28)	9.0 ± 0.1 (28)
Serum P (mg/dl)	Pre-HD	5.8 ± 0.2 (28)	5.9 ± 0.3 (28)
Serum HCO3 (mEq/l)	Pre-HD	22.8 ± 0.3 (28)	22.2 ± 0.3 (28)	0.04
Serum PTH (pg/ml)	Pre-HD	651 ± 76 (28)	562 ± 57 (28)	0.028
Serum iCa (mg/dl)	Pre-HD	4.8 ± 0.07 (28)	4.8 ± 0.10 (27)
	Post-HD	4.6 ± 0.04 (27)	4.9 ± 0.08 (26)	0.002
Serum Mg (mg/dl)	Pre-HD	2.3 ± 0.04 (28)	2.4 ± 0.07 (26)
	Post-HD	2.0 ± 0.02 (28)	2.1 ± 0.03 (26)
Plasma citrate (μM)	Pre-HD	134 ± 9 (25)	130 ± 7 (25)
	Post-HD	383 ± 16 (25)	187 ± 20 (25)	0.0001

Ca, calcium; HCO3, hydrogen carbonate; HD, hemodialysis; iCa, ionized Ca; Mg, magnesium; P, phosphorus; PTH, parathyroid hormone.

Values for serum Ca, P, HCO3, and PTH are averages of monthly measurements during each arm. Other measurements are averages of 2 measurements during each arm. Data are means ± standard errors of the number of subjects in parentheses. Parentheses indicate number of subjects. P-values from paired t testing.

Table 5.

Blood chemistries on each arm of the study for each group

Chemistry	Group 1		Group 2
	Citrate arm	Acetate arm	Citrate arm	Acetate arm
Ca (mg/dl)	8.74 ± 0.17	8.77 ± 0.18	9.12 ± 0.14	9.16 ± 0.16
P (mg/dl)	5.80 ± 0.31	5.66 ± 0.41	5.79 ± 0.31	6.01 ± 0.35
HCO3 (mEq/l)	22.2 ± 0.3	22.7 ± 0.5	23.3 ± 0.5	21.7 ± 0.4
PTH (pg/ml)	461 ± 48	444 ± 36	793 ± 117a	650 ± 91
iCa (mg/dl) pre	4.73 ± 0.11	4.80 ± 0.11	4.88 ± 0.08	4.83 ± 0.15
Post	4.64 ± 0.09	4.91 ± 0.11	4.65 ± 0.05	4.87 ± 0.11
Mg (mg/dl) pre	2.27 ± 0.07	2.26 ± 0.08	2.28 ± 0.06	2.51 ± 0.10
Post	2.04 ± 0.04	2.02 ± 0.05	2.01 ± 0.03	2.07 ± 0.04
citrate (μM) pre	116 ± 11	131 ± 12	142 ± 12	127 ± 8
Post	358 ± 21	160 ± 29	393 ± 22	199 ± 25

Ca, calcium; HCO3, hydrogen carbonate; HD, hemodialysis; iCa, ionized Ca; Mg, magnesium; P, phosphorus; PTH, parathyroid hormone.

Values for serum Ca, P, HCO3, and PTH are averages of monthly measurements during each arm. Other measurements are averages of 2 measurements during each arm. Data are means ± standard errors.

^aP = 0.027 vs. Group 1 by paired t testing.

Predialysis plasma citrate did not differ between arms; however, as expected, was substantially greater after dialysis in the citrate arm, with a mean increase of 215 ± 25%. The increase in postdialysis plasma citrate in the acetate arm was due to very high levels in some samples from 5 subjects mirroring the levels obtained during the citrate arm, strongly suggesting that the subjects were on the wrong dialysate when the blood samples were obtained. When these samples were omitted, the mean level was 146 ± 10 uM and not significantly different from the predialysis levels. Postdialysis citrate was uniformly high in all the subjects during the citrate arm. Corresponding data for each group are shown in Table 5.

Vascular Calcification

All 5 MGs were obtained in 22 of the subjects. Six-month MGs were not obtained for one of the arms in 4 subjects and for both of the arms in 2 subjects. In those cases, the change in BAC was calculated from the yearly MGs. An example of BAC progression is shown in Figure 1. As shown in Table 6, the median change in BAC in mm/yr was significantly greater during the citrate arm irrespective of which arm the subject was started on. The results were similar when expressed as a percentage of the baseline BAC (Figure 2). The median difference between the 2 arms was 21%/yr; IQR: 4%/yr–104%/yr, (P = 0.0005).

Table 6.

Progression of breast arterial calcification

Subjects	Parameter	Citrate	Acetate	Citrate-acetate	P-value
All	mm/yr	47 (10–124)	9 (−4 to 34)	27 (5–81)	0.0005
	%/yr	40 (20–98)	21 (−3 to 33)	21 (4–104)	0.0005
Citrate first	mm/yr	52 (15–141)	9.0 (−4 to 24)	51 (2–83)	0.034
Acetate first	mm/yr	47 (9–120)	7 (−6 to 41)	18 (8–81)	0.011

Values are medians and interquartile ranges. Significance was determined by paired testing using the Wilcoxon signed-rank test.

Figure 2.

The change in breast arterial calcification/yr expressed as a percentage of the baseline breast arterial calcification. Each line represents a different subject. Horizontal lines indicate the medians and vertical lines indicate the interquartile ranges. Paired analysis by the Wilcoxon signed-rank test yielded a P-value of 0.0005.

Of the 28 subjects, 22 had a greater rate of progression on the citrate arm. In the 6 subjects with greater progression on the acetate arm, the median rate was only 14%/yr (IQR: 0.9%/yr–50%/yr) compared with 51%/yr (IQR: 31%/yr–127%/yr) in the 22 subjects with a greater change in BAC during the citrate arm. Ten patients on the acetate arm had a decrease in BAC ranging from 1.1%/yr to 19%/yr, whereas only 5 subjects had a decrease in BAC during the citrate arm, ranging from 3.7%/yr to 20%/yr. These decreases are probably within the margin of error of BAC measurements and likely do not represent reversal of calcification. Individual data points for the subjects starting on each arm are shown in Figure 3. The effect of citrate on progression was significant in both diabetic and nondiabetic subjects.

Figure 3.

Individual breast arterial calcification values for subjects starting on the citrate arm (left) and the acetate arm (right). The dashed lines indicate when subjects were switched to the next arm. The solid black lines represent the mean breast arterial calcification at the dialysate switch extrapolated at each end by the mean slope during that arm.

Citrate in Calcified Arteries

To determine whether citrate is present in vascular calcifications, segments of human arteries obtained from lower extremity amputations were analyzed using microCT to determine the location and extent of calcification and then extracted to remove citrate. The arterial segments were devoid of intimal calcification by visual inspection and confirmed by CT scanning. The amount of citrate correlated directly with the amount of calcifications (Figure 4). Although the amount of citrate within the calcifications could not be directly measured, it was estimated from the slope at about 0.8% weight/weight based on the microCT measurement of bone volume as a surrogate for calcification volume.

Figure 4.

Citrate content in 7 segments of calcified human arteries from 3 patients as a function of the amount of calcification based on micro–computed tomography. Each symbol represents a separate arterial segment. The slope is 0.81% with an r value of 0.91.

Discussion

Progression of vascular calcification was unexpectedly greater during HD with citrate, which could not be explained by differences in serum Ca and phosphate levels or dialysis adequacy. Citrate can chelate Mg, a known inhibitor of vascular calcification. Although predialysis and postdialysis serum levels of Mg were slightly lower during the citrate arm, it is unlikely that this small decrease could fully explain the results. The predialysis serum bicarbonate was slightly but significantly higher during the citrate arm; and though alkalinity can promote vascular calcification,²⁹ this very slight difference cannot explain the results. Lastly, serum PTH levels were 16% higher during the citrate arm; however, PTH does not appear to have any direct effect on vascular calcification.²⁹ Furthermore, because none of these chemistries correlated with BAC progression in either arm, they cannot explain the results.

The only substantial biochemical difference between the 2 arms was the postdialysis plasma citrate level, which increased just over 3-fold, suggesting that this was responsible for the difference in vascular calcification. This was substantially greater than the 37% increase previously reported.¹⁸ However, those measurements were performed in serum, which can interfere with citrate assays. Although an effect of acetate to reduce vascular calcification cannot be ruled out, there is no apparent mechanism.

Promotion of calcification by citrate might seem counterintuitive because of its ability to chelate Ca and prevent crystallization such as in urine. However, there is a large amount of citrate in bone, comprising >5% of the organic material and accounting for up to ≤90% of total body citrate.³⁰ Studies have shown that citrate plays an important role in the formation and stability of bone and in the initial stages of biomineralization of collagen,^31,32 possibly explaining the effect of citrate on vascular calcification in this study. In support of this, we found that citrate is present at significant levels in calcified human arteries and correlates strongly with the amount of calcification, indicating that it is associated with the calcifications. Because these arteries did not come from patients receiving HD, this likely cannot be explained by exogenous citrate. This estimated citrate content of 0.8% compares with 1.6% reported for bone³³ and is the first demonstration of citrate in vascular calcifications or ectopic calcification in general.

This study has several weaknesses. The final sample size was small due to the loss of many patients, unrelated to the treatment. However, the cross-over design with paired analysis and the highly significant differences mitigate this concern. The study was performed within a single dialysis organization, and the study population was entirely female and African American; however, it is highly unlikely that the effect of citrate would be limited to a single population and both populations comprise a substantial portion of patients undergoing HD. Lastly, there were apparent errors in the dialysate assignment during some of the dialysis sessions based on the high postdialysis citrate levels in a few patients during the acetate arm, likely related to the fact that citrate is standardly used in these units. Consistent with this, postdialysis citrate levels were uniformly high during the citrate arm. These errors in dialysate assignment were intermittent because the dialysate was checked periodically and would have minimized the differences in vascular calcification. When the 5 patients with high post-HD citrate levels during the acetate arm were omitted, progression of calcification was still significantly greater during the citrate arm (91%/yr vs. 25%/yr, P = 0.0054).

Our previous studies have shown that BAC predicts cardiovascular events in patients undergoing HD,^23,25 consistent with other studies showing a strong correlation between medial arterial calcification and outcomes.11, 12, 13, 14 Consequently, the effect of citrate demonstrated in this study would be expected to translate into poorer outcomes, which has not been carefully examined. An initial report of increased mortality in patients dialyzed with citrate³⁴ was not confirmed in subsequent studies35, 36, 37, 38; however, all these studies had significant limitations, including assumptions about the dialysate assignment. This was followed by a large international study³⁹ that showed a hazard ratio for mortality of 1.14 associated with citrate dialysis that did not quite reach statistical significance (confidence interval: 0.97–1.34). In all these studies, there was uncertainty about the exposure to citrate, and only mortality was examined. Therefore, the long-term effect of HD with citrate remains unknown and highlights the fact that the safety of different dialysis solutions is based solely on short-term studies.

The results have potential implications beyond HD because plasma citrate levels are increased by PTH and calcitriol.⁴⁰ Furthermore, citrate is frequently administered to individuals without ESRD to alkalize the urine or in the form of a Ca supplement, and can increase plasma levels.⁴¹ Whether citrate can increase vascular calcification in the absence of an underlying predisposition remains to be determined.

Disclosure

During this study, WCO served on the Scientific Advisory Board for Inozyme Pharma and consulted for Rectify Pharma. All the other authors declared no competing interests.

Supplementary Material

Supplementary File (PDF)

CONSORT Checklist.