A noninferiority trial comparing a heparin-grafted membrane plus citrate-containing dialysate versus regional citrate anticoagulation: results of the CiTED study

ABSTRACT

Background. Anticoagulation is a prerequisite for successful haemodialysis. Heparin and low-molecular weight heparins are routinely used despite increased bleeding risk. Regional citrate anticoagulation (RCA) is efficacious, but is laborious and may induce metabolic disturbances. Heparin-grafted membranes are less efficacious. It is not known whether combining citrate-containing dialysate and a heparin-grafted membrane is a valid anticoagulation strategy.

Methods. We performed a randomized crossover noninferiority trial, with a prespecified noninferiority threshold of 10% in maintenance dialysis patients (n = 25). We compared the combination of citrate-containing dialysate plus a heparin-grafted membrane [CiTrate and EvoDial (CiTED) protocol] with RCA. The primary endpoint was completion of dialysis without significant clotting. Secondary endpoints included time to clotting, achieved Kt/V_urea, loss of total cell volume, venous air chamber clotting score and systemic-ionized calcium concentration.

Results. In total, 1284 sessions were performed according to study protocol, 636 in the CiTED arm and 648 in the RCA arm. The primary outcome of preterm interruption due to clotting occurred in 36 (5.7%) of sessions in the CiTED arm, and in 40 (6.2%) sessions in the RCA arm, thereby meeting noninferiority criteria (P < 0.0001). Most of the clotting events occurred in the fourth hour of dialysis. Repetitive clotting occurred in four patients in the CiTED arm and one patient in the RCA arm. Time to preterm interruption due to clotting and achieved Kt/V_urea was not significantly different. Systemic-ionized calcium levels during treatment were significantly lower in the RCA arm and clinically relevant hypocalcaemia was noted only in the RCA arm.

Conclusion. The combination of citrate-containing dialysate and a heparin-grafted membrane is a valid alternative to RCA.

anticoagulation, haemodialysis, heparin-grafted membrane, regional citrate anticoagulation

INTRODUCTION

Anticoagulation is a prerequisite for haemodialysis (HD) [1]. The optimal-anticoagulant regimen provides full anticoagulation of the extracorporeal circuit with minimal systemic effects and comes at reasonable cost. Unfractionated heparin (UFH) has been the standard of care for many years. In several countries, UFH has gradually been replaced by low-molecular weight heparins (LMWHs). LMWHs are easy to use as they are administered as a bolus injection, and reduce membrane fibrin and platelet deposition [1, 2].

Both UFH and LMWH provide adequate anticoagulation of the extracorporeal circuit, at the price of systemic anticoagulation. Repetitive systemic heparinization contributes to the high-bleeding risk in HD patients [3, 4], especially when combined with other anticoagulants [5]. Apart from the risk for bleeding, systemic anticoagulation with UFHs has been associated with dyslipidaemia, hypoaldosteronism and hyperkalaemia, thrombocytopenia, osteoporosis, pruritus and hypersensitivity reactions. Several alternative anticoagulation regimens have been proposed including heparin coating of the dialyser membrane and regional citrate anticoagulation (RCA).

Recently, two well-performed studies compared the efficacy of heparin-grafted membranes with saline infusion to prevent clotting of the extracorporeal circuit. The HepZero trial was a multinational study, conducted in 10 centres across Europe and Canada [6]. Two hundred and fifty-one patients were randomly allocated to treatment with either a heparin-grafted membrane or the standard-of-care, i.e. intermittent or continuous saline infusion. Completion of prescribed dialysis treatment was reached in 68.5% of patients randomized to the heparin-grafted membrane group as compared with 50.4% in standard of care. Although use of a heparin-grafted membrane was noninferior to saline infusion, the prespecified superiority endpoint was not met. A second French study found similar efficacy of heparin-grafted membranes [7]. Overall, the rate of clotting using heparin-coated membranes is considered to be too high [8].

RCA is performed by infusing citrate into the arterial line of the dialysis tubing to reduce ionized calcium concentrations to very low levels [1, 9]. Ionized calcium concentrations are restored by calcium supplementation prior to reinfusion of the blood into the patient, either by using a separate infusion pump or by using calcium-containing dialysate. Two previous studies found that RCA is superior to heparin-coated polyacrylonitrile dialysers and resulted in significantly greater instantaneous urea nitrogen clearances [10, 11]. RCA is more laborious as it requires additional actions both prior to preparation of citrate and calcium infusion pumps) and during dialysis sessions (measurements of ionized calcium). RCA is also associated with an increased risk of metabolic derangements, including hypocalcaemia and acid–base disturbances. Moreover, costs associated with RCA exceed those of conventional HD with (low molecular weight) heparin. As a result, use of RCA varies between regions and is limited in several countries including the USA.

Citrate-containing dialysate concentrates were recently introduced into clinical practice. Besides the advantages of acetate-free dialysate, this provides a modest local anticoagulant effect inside the dialyser, allowing reduced heparin dosing while maintaining extracorporeal circuit patency and optimizing dialyser clearances [12]. This effect, however, was insufficient to obviate the need for anticoagulation in all patients [13]. More recently, citrate-containing dialysate has been used in combination with a heparin-grafted membrane. In a retrospective cohort study, this combination therapy showed promising results, although no head-to-head comparisons were made [14].

Previous descriptive studies have suggested that a combination of a heparin-grafted mebrane plus citrate-containing dialysate could be an acceptable strategy for systemic heparin-free dialysis. The present study aimed to test whether this combination is indeed noninferior to RCA. We performed a comparative study of RCA versus the combination of citrate-containing dialysate plus a heparin-grafted membrane in an unselected cohort of patients treated with conventional thrice-weekly HD.

MATERIALS AND METHODS

Study design

We performed an open-label randomized crossover trial to compare the efficacy of the combination of a heparin-grafted membrane and citrate-containing dialysate (intervention) versus RCA (comparator). The primary endpoint was the difference in the failure rate to complete prescribed dialysis sessions due to clotting phenomena. Completion of dialysis was defined as the absence of complete clotting of the dialyser or venous air chamber, no change of dialyser or blood lines and no additional interventions, e.g. saline flushes to prevent clotting and no early rinse-back for impeding clotting of the circuit leading to premature termination of the dialysis session.

The comparator RCA has a high success rate with previous studies all showing completion in more than 90% of performed sessions. We therefore designed this study as a noninferiority trial with a predefined noninferiority threshold set at 10%. The power to demonstrate noninferiority depends on the actual failure rates of the experimental and control treatment. For power calculations we used data from a previous study comparing different types of RCA with a heparin-coated membrane [10]. We calculated that with 25 patients and 30 sessions in each arm (1500 sessions in total) the study would be adequately powered (>80%). The study adhered to the principles of Helsinki and was recorded in a trial repository (clinicaltrials.gov NCT02281045). Study design was reviewed and approved by the institutional medical ethics committee.

Study participants and interventions

Patients older than 18 years and treated with maintenance HD (>3 months) at the University Hospitals Leuven (Belgium) were considered eligible to be included in this study. Exclusion criteria were any haemostatic disorder favouring either bleeding or clotting or the use of anti-vitamin K therapy. Treatment with aspirin, dipyridamole or other drugs likely to interfere with platelet aggregation was registered prospectively, but was not an exclusion criterion. Vascular access was recorded for all patients. In case of tunnelled central venous catheters, trisodium citrate 30% locking solution (Citra-Lock S, Citra-Gean Ag, S.A.L.F. SpA, Bergamo, Italy) was used in between treatments.

In the active arm, patients were treated with HD using a combination of a citrate-containing dialysate and a heparin-grafted membrane. Citrate-containing dialysate was produced using Selectbag® Citrate 1/200 A concentrate (Gambro Dasco, Sondalo, Italy). We used the Evodial™ 1.6 (Gambro Industries, Meyzieu, France) with an effective membrane surface area of 1.65 m². This device is a precoated heparin-grafted membrane. In contrast to a previous design (Nephral™, Gambro-Hospal, Meyzieu, France) no separate preparation step is needed and priming of the extracorporeal circuit is identical to conventional membranes. Priming volume was fixed at 1500 mL. As residual air bubbles would lead to activation of the coagulation cascade, great care was taken to remove residual air during the priming procedure. Dialysate flow was set at 700 mL/min immediately before start of dialysis (500 mL/min during priming). This study arm has been coined the CiTED arm (CiTrate and EvoDial).

Patients in the comparator arm were prescribed HD with RCA as published previously [10, 15]. In short, a hypertonic sterile solution of trisodium citrate dihydrate (1.035 mol/L, Baxter, Lessines, Belgium) was infused into the afferent blood line at a rate of 62.1 mM/h (60 mL/h) using a separate infusion pump. The anticoagulant effect of citrate was neutralized using calcium-containing dialysate with a calcium concentration of 1.50 mmol/L. In all sessions, a polyarylethersulfone dialyser (Polyflux® 170H, Gambro Dialysatoren GmbH, Hechingen, Germany) was used, with an effective membrane surface area of 1.7 m². Throughout the manuscript this arm is denoted as RCA.

For all sessions, prescribed dialysis duration was 4 h, irrespective of individual patients’ Kt/V. Target blood flow during treatment was 300–350 mL/min. Access for dialysis was either via arteriovenous (AV) fistula or tunnelled central catheter. For all procedures dual access was required, de facto excluding patients on single-needle HD from study participation.

Dropout from the study was due to transfer to a different dialysis unit, transplantation, repetitive clotting for three consecutive sessions and medical reasons at the discretion of the treating physician. The reason for dropout was recorded in the medical records.

Biochemical analyses and secondary endpoints

Throughout the study period, routine biochemistry was performed according to local practice, including single-pool Kt/V measurements every fortnight during the midweek session. We measured systemic-ionized calcium concentrations in all sessions at the beginning, after 30 min, 2 h and at the end of the treatment.

We analysed the following secondary endpoints: differences in time to clotting, patency of the dialyser, clotting of the venous air chamber and dialytic clearance. Patency of the dialyser is difficult to interpret using visual clotting scores, as fibre transparency differs between the two dialysers used in this study. To overcome this issue we measured total cell volume (TCV) using the Renatron®II system, originally developed for reprocessing of used membranes [16]. During this process, the total volume of patent fibres is measured. Although we did not reuse membranes, the Renatron allowed us to objectively quantitate dialyser patency after treatment (prior to the regeneration process). Clotting of the venous air chamber was done using a semiquantitative score ranging from zero (minimum) to three (maximum), with zero meaning no visible clotting, one minimal clotting, two clotting with thrombi <4 cm, and three-equalling thrombotic occlusion of the venous air chamber necessitation interruption of the dialysis treatment. For dialysis efficacy, Kt/V was measured every 2 weeks during the midweek dialysis session.

Statistical analyses

Baseline demographics are reported as mean [standard deviation (SD)] or median (minimum, maximum), dependent on distribution analysis. For analysis of the primary endpoint, we used two different statistics. In a first basic approach we performed ordinary table analysis using χ²-statistics. This, however, assumes independency of individual sessions. In a second approach, to correct for repeated measures, a binary logistic linear mixed model was constructed to model the effect of intervention on the risk of clotting using a random intercept, random slopes and a first-order autoregressive covariance structure. The three hierarchical levels of fixed effects were treatment arm, within-patient treatment period and dialysis session. For secondary endpoints analyses, we used (nonparametric) ANOVA as appropriate. Time to clotting was analysed according to Kaplan–Meier. Power calculations were performed using PASS 2008 V08.0.15 software (supported by Gambro). Statistical analyses were performed using SAS version 9.3 and SPSS (version 23).

RESULTS

Baseline demographics and patient flow

We performed an open-label randomized crossover trial to compare the efficacy of the combination of citrate-containing dialysate plus a heparin-grafted membrane (CiTED arm) versus RCA. Twenty-five patients providing informed consent were randomized to start the study. Baseline demographics are depicted in Table 1 and dialysis prescription in Table 2. Twelve patients were randomized to start in the CiTED arm and 13 patients in the RCA arm (see Figure 1). During the first study period, one patient randomized to the CiTED arm was transferred to another dialysis unit. A second patient randomized to the RCA arm received a kidney transplant. At time of crossover, therefore, 11 patients were switched from CiTED to RCA and 12 patients were switched from RCA to the CiTED arm. In the second study period, one patient in the RCA arm received a kidney transplant. One patient developed repetitive clotting both when in the CiTED arm and the RCA arm and left the study preterm. Three more patients in the CiTED arm developed repetitive clotting and stopped this arm preterm. None of the patients was withdrawn from the study for medical reasons.

FIGURE 1

Patient flow during the study. At the beginning of the study, 25 patients were randomized to either a combination of citrate-containing dialysate plus a heparin-grafted membrane (CiTED arm) or regional citrate anticoagulation (RCA arm). All patients were crossed over after 30 scheduled session in each arm. The number of scheduled, performed and analysed sessions is reported. n, number of patients at each time point.

Table 1

Study population

Variable	Overall	CiTED arm	RCA arm	P
Age  (years)	71.3  (12.8)	–	–
Gender, male/female  (%)	16/9  (64/36)	–	–
Dialysis vintage  (months)	24  (3–49)	–	–
Cause of ESRD, n  (%)		–	–
Diabetes	5   (20)
Vascular	7  (28)
Glomerulonephritis	5  (20)
Tubulointerstitial	3  (12)
ADPKD	3  (12)
Unknown	2  (8)
BMI  (kg/m2)	22.8  (3.6)	–	–
Haemoglobin  (g/dL)	10.3  (1.0)	10.2  (1.1)	10.4  (1.0)	0.3
C-reactive protein  (mg/dL)	3.7  (0.4–147.9)	3.5  (0.4–136)	3.7  (0.4–147.9)	0.8
Albumin  (g/L)	40.0  (3.3)	40.2  (3.8)	39.9  (2.8)	0.6
Magnesium (mmol/L)	0.82  (0.15)	0.81  (0.13)	0.84  (0.16)	0.6
Anticoagulation, n  (%)	20  (80)	20  (80)	20  (80)	1.0
Acetylsalicylic acid  (%)	80	74	78
Clopidogrel  (%)	16	17	13
Vitamin K antagonists  (%)	0	0	0
NOAC  (%)	0	0	0
Sertraline  (%)	4	4	4

Relevant demographic, biochemical and drug therapy data. For drug therapy, the percentage of prescribed drugs at any time point  (overall) and per study period is given. Descriptive data are given as mean  (SD) or median  (minimum, maximum) as appropriate.

ESRD, end-stage renal disease; ADPKD, autosomal dominant polycystic kidney disease; BMI, body mass index; NOAC, non-vitamin-K oral anticoagulants.

Table 2

Dialysis prescription

Variable	CiTED arm	RCA arm
Access type
AVF/AVG/CVC (n)	19/1/5	19/1/5
Time on AVF (%)	86	82
Time on AVG (%)	5	5
Time on CVC (%)	9	13
Blood volume (L/session)	77.65 (6.44)	77.73 (5.70)
Dialysate
Temperature (°C)	36 (35.5–36.5)	36 (35.5–36.5)
Na+ (mmol/L) (range)	136–140	136–140
K+ (mmol/L) (range)	1–3	1–3
${\text{HCO}}_3^{\text{−}}$ (mmol/L)	35	25
Ca2+ (mmol/L)	1.5	1.5
Mg2+ (mmol/L)	0.5	0.5
Citrate3− (mmol/L)	1	0
Dialyser
Material	Acrylonitrile/ sodium methallylsulfonate/ polyethyleneimine	Polyarylethersulfone/ polyvinylpyrrolidone/polyamide
Heparin grafted	Yes	No
Membrane surface area (m2)	1.65	1.7
Inner diameter (µm)	210	215
Effective length (mm)	280	270
Kurea (mL/min) (QB300, QD500, QUF0)	250	270
UF coefficient (mL/h/mmHg)	50	70
Sieving coefficient inulin	0.96	1.0
Sieving coefficient albumin	<0.01	<0.01

Characteristics of the dialysis prescription in the CiTED arm, consisting of a citrate-containing dialysate and a heparin-grafted membrane, and the RCA arm. For access type, distribution at the start of each study period is given. Time on AVF, AVG and CVC are calculated based on number of sessions per total analysed sessions. Descriptive data are given as mean (SD) or median (minimum, maximum) as appropriate.

AVF, arteriovenous fistula; AVG, arteriovenous graft; CVC, central venous graft; UF, ultrafiltration.

Overall, 1500 sessions were scheduled, 750 sessions in the CiTED arm and 750 sessions in the RCA arm. Total dropout amounted to 14.1% and 1289 HD study sessions were delivered. Protocol violations were observed in five additional sessions (0.4%), three in the RCA arm (citrate infusion pump not started or started too late) and two in the CiTED arm (heparin-grafted membrane was not used). In total, 1284 prescribed sessions were performed according to study protocol, of which 636 were in the CiTED arm and 648 in the RCA arm (P > 0.2).

Completion of study sessions

The primary endpoint of this study was completion of prescribed study sessions, defined as the absence of complete clotting of the dialyser or venous air chamber, no change of dialyser or blood lines, and no additional interventions, e.g. saline flushes to prevent clotting and no early rinse-back for impeding clotting of the circuit leading to premature termination of the dialysis session. Overall, the number of sessions not completed due to clotting phenomena was low in both arms and amounted to 5.9%. In the CiTED arm, 36 out of 636 sessions were not completed due to clotting, whereas 94.3% (97.5% confidence interval 91.9–96.2%) of sessions were completed. In the RCA arm, 40 out of 648 sessions were not completed, whereas 93.8% (97.5% confidence interval 91.3–95.8%) of sessions were completed.

When assuming each session as a separate event, the noninferiority endpoint was met (P < 0.0001). To compensate for repeated measures, we used mixed modelling taking into account treatment periods. The mean estimated difference in the odds of clotting between the intervention and control arms was 0.4% (±2.8%; P = 0.779), less than the prespecified noninferiority margin of 10%.

Secondary efficacy endpoints

We analysed the impact of both strategies on dialysis efficacy using two different approaches, i.e. delivered treatment time and delivered dialysis dose. First, we looked at time to clotting, as treatment downtime affects dialysis efficacy. Early clotting (<2 h) of the extracorporeal circuit did not occur, while most (>80%) clotting events occurred in the fourth hour of dialysis sessions (Figure 2). We analysed time to clotting according to the Kaplan-Meier method, under the assumption that each session was a separate event. Time to clotting was not significantly different between study arms (P = 0.67).

FIGURE 2

Kaplan–Meier curve of time to clotting. Kaplan–Meier curve of time to clotting in the CiTED arm and RCA arm. Time to clotting was not significantly different (P = 0.67). CiTED, the combination of citrate-containing dialysate plus a heparin-grafted membrane; RCA, regional citrate anticoagulation.

Second, we measured small solute clearance, measured as single-pool Kt/V_urea. We did not observe a significant difference in Kt/V_urea, although a trend toward higher Kt/V_urea was found in the RCA arm (P = 0.07). In both arms, however, single-pool Kt/V_urea values were well above the minimum requirements for dialysis. In the CiTED arm, mean Kt/V amounted to 1.71 (SD 0.25), whereas in the RCA arm mean Kt/V was 1.79 (SD 0.39).

We measured occlusive clotting of individual dialyser capillaries by determining TCV after dialysis. As TCV of the different dialysers was not identical, we calculated loss of TCV relative to baseline dialyser-specific TCV as a measure of loss of dialyser fibre volume due to clotting. Due to temporary breakdown of the Renatron®II device, measured TCV was available for a subset of sessions [CiTED 449, RCA 426, total 875 (68.1% of all) sessions]. We observed no difference in TCV (mean 22.7 versus 26.6% loss of TCV in CiTED versus RCA, P = 0.6).

We also estimated the degree of clotting in the venous air chamber, based on visual inspection and scores ranging from no visible clotting (0), minimal clot formation (1), semiocclusive clot (2) to complete clotting of the air chamber and dialyser, without possibility to continue dialysis (3). Previous studies have shown validity of this semiquantitative scoring system [10, 15]. In both arms, clotting scores followed a similar trend and increased over time during dialysis (Figure 3). We observed significantly higher venous air chamber clotting scores over time in the RCA arm, although numerical differences were small.

FIGURE 3

Secondary efficacy and safety endpoints. (A) Venous air chamber clotting scores over time. (B) Ionized calcium concentrations over time. CiTED, the combination of citrate-containing dialysate plus a heparin-grafted membrane; RCA, regional citrate anticoagulation.

Secondary safety endpoints

As systemic hypocalcaemia is a well-known risk associated with the use of citrate, we measured serum ionized calcium in the arterial line of the dialysis tubing. In both arms, mean ionized calcium concentrations were well within the acceptable range (Figure 3). However, systemic-ionized calcium levels during treatment were significantly lower in the RCA arm (Figure 3). Clinically relevant hypocalcaemia was noted only in patients in the RCA arm: Ca²⁺<0.8 mmol/L in one session, Ca²⁺ between 0.8 and 0.89 mmol/L in 23 sessions. When looking at hypercalcaemia, Ca²⁺>1.35 mmol/L was measured during one session in the RCA arm. During the study period, we did not observe any clinical endpoints due to low or high serum calcium.

DISCUSSION

In this noninferiority crossover trial we compared two different approaches of systemic heparin-free dialysis. We showed that a combination of citrate-containing dialysate plus a heparin-grafted membrane is noninferior to conventional RCA.

Adequate anticoagulation of the extracorporeal circuit is a key requisite for successful HD. Heparin and its low-molecular weight derivatives (LMWH) are the standard of care in most dialysis units, as this provides adequate and titratable anticoagulation. On the flip side, heparin/LMWH results in systemic anticoagulation and may result in increased bleeding risk, dependent on patient comorbidities and concomitant drug therapy. Apart from increased bleeding risk, a subset of patients develops heparin-induced thrombocytopenia. Moreover, heparin and to a lesser degree LMWH have unwanted metabolic side effects. For these reasons, in a subset of patients, it is necessary to perform systemic heparin-free dialysis.

International guidelines, however, do not provide unequivocal guidance on what is the preferred modality in these patients [17]. Some advocate using (intermittent) saline infusion or the use of a heparin-grafted membrane. At least two independent studies have demonstrated that these strategies are inferior to RCA [10, 11]. Despite the proven efficacy of RCA, most centres consider this procedure more cumbersome and prone to error. Moreover, RCA imposes a risk of inducing metabolic derangements, including hypo- and hypercalcaemia as well as metabolic acidosis and metabolic alkalosis [10, 15].

More recently, citrate-containing dialysates were introduced. The primary aim was to replace acetate by citrate. Interestingly, low-dose influx of citrate into the blood compartment also was shown to reduce the need for heparinization during dialysis, presumably due to chelation of some calcium in the blood compartment resulting in reduced ionized calcium concentrations. Although in itself insufficient to allow heparin-free dialysis we speculated that combining this effect with the use of a heparin-grafted membrane would reduce clotting.

In the current study, we show that the anticlotting effect of the combination of citrate-containing dialysate and a heparin-grafted membrane is comparable to that of conventional RCA. The number of completed sessions in the regional citrate arm is comparable to previous studies. In fact, nominally more sessions could be completed using the CiTED protocol. The results of the primary endpoint are supported by secondary endpoints showing that time to complete clotting is comparable between arms (P = 0.7), as is the loss of TCV (P = 0.6). A surprising finding is that clotting scores of the venous air chamber are significantly lower in the CiTED arm, although numerical differences are small. We do not have a clear explanation for this finding.

The use of a heparin-grafted membrane resulted in slightly lower Kt/V_urea (mean 1.71) compared with the RCA arm (spKt/V_urea mean 1.79). This probably reflects the small difference in membrane surface area (1.65 versus 1.7 m² for Evodial versus Polyflux 170H, respectively). In both arms, dialysis dose was well above target levels. As the HEMO trial [18] showed no benefit of further increments of Kt/V_urea, we believe the observed small difference in Kt/V_urea does not translate into clinically meaningful outcome differences.

A major benefit of the CiTED protocol is the ease of use. There is no need for separate infusion pumps for citrate of calcium. Moreover, systemic-ionized calcium concentrations are stable and stay within the normal range in both arms. Despite the fact that our unit has longstanding experience performing RCA, in the current study hypocalcaemia was recorded in a small number of sessions. Although this did not translate into clinical side effects and no cardiac arrhythmias were reported, it should be acknowledged that these were not systematically monitored. Based on the current study we would recommend to continue monitoring of ionized calcium concentrations during RCA, whereas no monitoring of ionized calcium is required during the CiTED protocol.

This study has several limitations. First, it should be noted that both anticoagulation strategies to date are more expensive than conventional dialysis using unfractionated or LMWHs. Actual costs may differ between countries, dependent on the cost of personnel, equipment, consumables and biochemical analyses. Cost per treatment for either strategy, taking into account dialyser, concentrates, additional ionized calcium measurements and additional nursing time, exceeds that of conventional dialysis by ∼50%. While we performed this study in stable maintenance dialysis patients, it is likely that such anticoagulation strategies will be prescribed for selected patients only. It is unclear whether these findings can be extrapolated to the setting of acute dialysis. Second, as these were stable maintenance dialysis patients, a significant fraction was taking antiplatelet drugs during the study period. This may have affected the results of this study. Due to the crossover design, however, the effect was balanced between the study arms. Finally, we performed RCA with a calcium-containing dialysate. Previous studies in our unit have shown that there is a small increase in the degree of clotting as compared with RCA with a zero-calcium dialysate. The main advantage of using calcium-containing dialysate is that ionized calcium concentrations are more stable and the risk of severe hypocalcaemia is reduced. As RCA with a calcium-containing dialysate is the standard of care in our unit we chose to use this as comparator for our study. However, the confidence range of the actual success rate in the CiTED arm is <10% lower than maximum (100%) success rate, thus suggesting that noninferiority would have been demonstrated even with the most efficient RCA protocol.

In conclusion, combining citrate-containing dialysate with a heparin-grafted membrane is an efficient and easy to use alternative to RCA during dialysis. This combination might be a valid therapeutic regimen in patients at increased risk of bleeding.

ACKNOWLEDGEMENTS

On-site support by A. De Winter, R. Bruyneel, J. Andries and D. Desmyter was greatly appreciated. Part of this work has been presented at the XLVIIth ESAO congress, 2–5 September 2015, Leuven, Belgium and at the ASN renal week, 3–8 November 2015, San Diego, USA.

FUNDING

This is an investigator-driven study, supported by a study grant from Gambro. Gambro (including all subsidiaries) is now part of Baxter International Inc. The company had no part in study design, patient inclusion, data collection, statistical analysis and writing of the report.

CONFLICT OF INTEREST STATEMENT

B.M. reports having received consultancy and speaker fees from Bellco and Gambro, unrelated to this trial.

AUTHORS’ CONTRIBUTIONS

Research idea and study design: B.M., D.K., P.E.; data acquisition: C.M.; data analysis/interpretation: B.M., C.M.; statistical analysis: C.M., T.V., R.P.; supervision or mentorship: D.K., P.E. Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved. B.M. takes responsibility that this study has been reported honestly, accurately and transparently; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.