Citrate anticoagulation versus systemic heparinisation in continuous venovenous hemofiltration in critically ill patients with acute kidney injury: a multi-center randomized clinical trial

This was a multi-centre, randomized, non-blinded, prospective clinical trial with parallel group design performed in 10 ICUs in the Netherlands. The study was carried out in accordance with the Declaration of Helsinki, and was approved by the institutional review boards at each of the participating study centres (see Acknowledgements). The study was registered at clinicaltrials.gov number NCT00209378. Informed consent was obtained from all study participants or their next of kin. We use the acronym CASH for citrate anticoagulation versus systemic heparinisation.

Adults admitted to the ICUs of the participating centres, and who required CVVH, were randomly assigned by sealed opaque envelopes with concealed treatment allocation inside, stratified by centre, and drawn by an independent individual not involved in the trial, to receive heparin or citrate for CVVH in predilution mode in a single-blinded fashion. All randomized patients were enroled. CVVH was started for AKI and uncontrolled uraemia, diuretic-resistant volume overload, respiratory distress, multiorgan failure, or any combination of these features, at the discretion of the treating physician and consulting nephrologist. Exclusion criteria were the presence of an increased bleeding risk (defined as a platelet count below 40 × 10⁹/L, an activated partial thromboplastin time (aPTT) longer than 60 seconds, a prothrombin time-international normalised ratio (PT-INR) greater than 2.0 or recent major bleeding), age below 18 or over 80 years, the need for therapeutic systemic anticoagulation (heparin or coumarins) or a known HIT. The administration of activated protein C or plasma exchange therapy were also considered exclusion criteria, as was chronic dependence on renal replacement therapy prior to admission to the ICU. The presence of liver disease of any kind was not an exclusion criterion.

The participating centres used the locally available haemofiltration machine, venous catheter and haemofilter. In patients receiving heparin, commercially prepared bicarbonate-buffered haemofiltration replacement solution (HF32bic, Dirinco, Rosmalen, the Netherlands) was used in most patients; lactate-based (BH504, Dirinco) haemofiltration replacement solution was used in six patients. For citrate anticoagulation, we adapted the protocol conceived by Pallson and Niles [6], and used a custom-made calcium-free trisodium citrate replacement fluid (HFCitPre, Dirinco), acting as anticoagulation and buffer in predilution mode, as described before [17],[18]. For composition of the replacement fluids used, see Additional file 1. Patients anticoagulated with heparin received a heparin bolus of 5,000 IU in a pre-filter fashion at the arterial pole prior to the start of CVVH and a separate heparin pump (20,000 IU/48 mL) was set at 2.0 mL/h and adjusted targeting a systemic aPTT of 50 seconds. Patients anticoagulated with regional citrate had a separate intravenous calcium drip, with calcium glubionate (Calcium Sandoz, containing 0.225 mmol/mL calcium, Novartis Consumer Health, Breda, The Netherlands). Calcium administration was adapted to concentrations of systemic ionised calcium by a specially designed algorithm, targeting systemic ionised calcium levels of 1.0 to 1.35 mmol/L [19]; for calcium pump settings see Additional file 1. Calcium levels in the extracorporeal circuit were not measured routinely, in order to keep the treatment protocol simple. Blood flow was initially set at 180 mL/minute in both groups. Predilution replacement flow rates varied between 2,000 and 4,000 mL/h according to local guidelines. The rate of infusion of replacement solution was coupled to the blood flow, aiming at stable citrate concentrations in the extracorporeal circuit. The replacement solution ran through a warming coil set at 39°C.

The pH, anion gap, ionised and total calcium and their ratio were monitored in patients on citrate, at least four times daily. The first measurement was done 1 h after initiation of CVVH. Routine daily laboratory measurements included acid-base balance, electrolytes, haemoglobin, and white blood cell and platelet counts. Also, levels of magnesium and sodium were monitored to detect citrate-related derangements. In patients on heparin the aPTT was determined every 6 to 8 h. An electrocardiogram was made if hypocalcaemia (ionised calcium below 0.9 mmol/L) occurred. Citrate accumulation was suspected if the patient fulfilled one or more of the following criteria: a ratio of total calcium to ionised calcium greater than 2.5, clinical signs of hypocalcaemia (tetanic symptoms or prolonged QT interval not due to medication), or progressive acidosis (pH <7.20) with an increased anion gap (>13 mmol/L) in the presumed absence of anions other than citrate. If there were signs of citrate accumulation, CVVH was continued with heparin. Patients treated with heparin, who were clinically suspected of having developed a HIT, continued CVVH with citrate. Also, in case of a clinically suspected bleeding episode, heparin was discontinued and patients continued CVVH with regional citrate anticoagulation. The crossover between study arms was documented. Filters were routinely replaced after 72 hours of use, apart from incidental cases when still functioning well (n = 2). CVVH was continued at the discretion of the treating physician. Patients were otherwise treated by board-certified intensivists, according to international and national guidelines.

Primary endpoints were patient mortality at 28 and 90 days after initiation of CVVH and renal outcome. For renal outcome, independence of renal replacement therapy 28 days after initiation of CVVH in surviving patients was studied. Secondary endpoints were safety and efficacy. Safety was defined as the absence of any adverse event necessitating discontinuation of study anticoagulant, such as bleeding or HIT in the heparin group and citrate accumulation in the citrate group. For efficacy, the survival time of the first filter, the number of filters used and off-time within the first 72 hours of therapy were studied. Additionally, a cost analysis for the first 72 hours of prescribed CVVH was performed. Follow up until day 90 was achieved in 134 of the 139 (96%) patients.

Severity of illness and organ failure were scored using the acute physiology and chronic health evaluation II (APACHE II) and the sequential organ failure assessment (SOFA) on days 0 and 3. All data were prospectively collected by the local researcher in the patient data management system of the participating centre. A research nurse, not involved in patient care, was responsible for the coordination of the collection of case record forms and accuracy of the database. The etiology of AKI, categorised as ischaemic, due to sepsis or other (including contrast nephropathy and rabdomyolysis), was assessed by the treating physician. The costs during the first 72 hours of prescribed CVVH were calculated using the following parameters: cost of replacement solution used (during first 72 hours of therapy minus off-time), cost of heparin administered in the heparin group, cost of calcium administered in the citrate group, costs of total numbers of filter sets used and the half-hourly wage of nursing staff per filter change.

We hypothesized an absolute mortality reduction of 15% after 28 days for patients on citrate, as compared to that of patients on heparin, with an estimated mortality rate of 65%. The power of the study was set to be 90% at 5% significance, so that the sample size was targeted at 180 patients per group. Values are presented as the median and range, because most data were non-normally distributed (Kolmogorov-Smirnov test, P <0.05). We compared groups using the Mann-Whitney U-test, the chi square (X²) test, or the Fisher exact test, where appropriate. Patient and filter survival data were analysed using Kaplan-Meier plots and log rank testing. Data were analysed on an intention-to-treat basis. We also performed a per-protocol analysis, excluding patients who switched from the assigned anticoagulation. Backward stepwise multiple logistic regression was used to evaluate the contribution of age, disease severity, prescribed dose and anticoagulation regimen on mortality at 28 days. The Hosmer-Lemeshow test was used to verify adequate calibration, as indicated by a P-value >0.05. Additionally, univariate logistic regression was performed generating odds ratios for mortality at 28 days in the following subgroups: age (higher versus lower than the median), APACHE II scores (higher versus lower than the median), prescribed dose (higher or lower than the recommended 20 mL/kg/h), sepsis versus non-sepsis and circulatory or respiratory failure at ICU admission. A two-sided P-value <0.05 was considered statistically significant; exact P-values are given unless <0.001.

Mortality rates at 28 and 90 days did not differ between groups: 22/66 citrate patients (33%) died versus 25/72 patients (35%) in the heparin group at 28 days, and 27/65 citrate patients (42%) died versus 29/69 patients (42%) in the heparin group at 90 days (P = 1.00 for both). Indeed, survival curves for 28- and 90-day outcomes were similar among anticoagulation groups (Figure 1). Also, a per-protocol analysis did not show any difference in mortality at 28 and 90 days: 18/61 citrate patients (30%) died versus 15/48 patients (31%) in the heparin group at 28 days and 23/60 citrate patients (38%) died versus 18/47 patients (38%) in the heparin group at 90 days (P = 1.00 for both). Multiple logistic regression showed higher age and high APACHE II scores at admission to be independent predictors of mortality at 28 days (P = 0.05 for both), whereas prescribed dose and anticoagulation regimen were not. In univariate logistic regression, there was no benefit from citrate anticoagulation in terms of mortality reduction at 28 days in any of the defined subgroups (see Additional file 4). Concerning renal outcome, there was no difference between the groups in dialysis independence among surviving patients 28 days after start of CVVH, with 29/43 patients (67%) in the citrate group versus 33/47 patients (70%) in the heparin group (P = 0.82).

Adverse events necessitating discontinuation of the primary assigned anticoagulant occurred with heparin more frequently than with citrate (Table 2 and Additional file 3). In one patient, citrate was discontinued and anticoagulation withheld. Citrate accumulation was suspected in 5 of the 66 patients (8%) randomized to citrate and proven in 4 patients (6%); in 2 patients there was a persistently elevated anion gap, attributed to citrate accumulation, and in 2 patients the calcium ratio exceeded 2.5 after 60 and 72 hours of therapy. Finally, one patient was erroneously ascribed to citrate accumulation. Of all measured ionised calcium levels within 72 hours in the citrate group, 61/508 (12%) were <0.9 mmol/L and none of the measurements exceeded 1.35 mmol/L. Potential treatment related derangements, such as metabolic alkalosis, hypernatremia or hypomagnesemia did not occur more often in the citrate group than in the heparin group (Table 2). Clinically suspected HIT was reported in 6 of 73 patients (8%) on heparin. There was a trend for fewer bleeding episodes in the citrate group (n = 5 versus n = 10 in the heparin group, P = 0.08), however, this did not result in a difference between groups in patients needing >2 erythrocyte concentrates (2 versus 4 patients in the heparin group, P = 0.68).

Efficacy parameters suggested superiority for citrate (Table 2). Survival times of the first filter were superior for citrate, with survival curves shown in Figure 2. An analysis excluding patients in whom the filter ran for more than 72 hours did not differ from the presented results. Furthermore, the off-time within 72 hours was less with citrate (1 (0 to 12) h versus 3 (0 to 31) h for heparin, P = 0.002), as were the number of filters used (1 (1 to 5) versus 2 (1 to 9) for heparin, P = 0.002). In per-protocol analysis the total duration of CVVH was longer in the citrate group, at 117 hours versus 70 hours for heparin (P = 0.04). There was a higher incidence of circuit disconnection due to clotting of the circuit in the heparin group (51% versus 24% in the citrate group) and more elective filter changes in the citrate group (30% versus 9% in the heparin group, P = 0.01). The total costs during the first 72 hours of prescribed CVVH were lower in citrate-based CVVH (P <0.001, Table 2.)