Regional Citrate Anticoagulation versus No Anticoagulation for CKRT in Patients with Liver Failure with Increased Bleeding Risk

Visual Abstract

Abstract

Background

The opinions on the efficacy and safety of no anticoagulation versus regional citrate anticoagulation for continuous KRT (CKRT) were controversial in patients with severe liver failure with a higher bleeding risk. We performed a randomized controlled trial to assess no anticoagulation versus regional citrate anticoagulation for CKRT in these patients.

Methods

Adult patients with liver failure with a higher bleeding risk who required CKRT were considered candidates. The included participants were randomized to receive regional citrate anticoagulation or no-anticoagulation CKRT. The primary end point was filter failure.

Results

Of the included participants, 44 and 45 were randomized to receive regional citrate anticoagulation and no-anticoagulation CKRT, respectively. The no-anticoagulation group had a significantly higher filter failure rate (25 [56%] versus 12 [27%], P = 0.003), which was confirmed by cumulative incidence function analysis and sensitive analysis including only the first CKRT sessions. In the cumulative incidence function analysis, the cumulative filter failure rates at 24, 48, and 72 hours of the no-anticoagulation and regional citrate anticoagulation groups were 31%, 58%, and 76% and 11%, 23%, and 35%, respectively. Participants in the regional citrate anticoagulation group had significantly higher incidences of Ca²⁺_tot/Ca²⁺_ion >2.5 (7% versus 57%, P < 0.001), hypocalcemia (51% versus 82%, P = 0.002), and severe hypocalcemia (13% versus 77%, P < 0.001). However, most (73%) of the increased Ca²⁺_tot/Ca²⁺_ion ratios were normalized after the upregulation of the calcium substitution rate. In the regional citrate anticoagulation group, there was no significant additional increase in the systemic citrate concentration after 6 hours.

Conclusions

For patients with liver failure with a higher bleeding risk who required CKRT, regional citrate anticoagulation resulted in significantly longer filter lifespan than no anticoagulation. However, regional citrate anticoagulation in patients with liver failure was associated with a significantly higher risk of hypocalcemia, severe hypocalcemia, and Ca²⁺_tot/Ca²⁺_ion >2.5.

Clinical Trial registry name and registration number

RCA for CRRT in Liver Failure and High Risk Bleeding Patients, NCT03791190.

Introduction

The incidence of AKI can be as high as 29% and 33% in patients with advanced cirrhosis and patients who underwent liver transplantation,^1–6 respectively. For severe AKI, continuous KRT (CKRT) is commonly used among patients with hemodynamic instability.³

The 2012 Kidney Disease Improving Global Outcomes (KDIGO) AKI guidelines suggested that regional citrate anticoagulation should be considered as the first-line anticoagulation for CKRT for patients with no contraindication.^3,7 Severe liver failure was considered one of the potential contraindications. In addition, a higher bleeding risk, including impaired coagulation, gastrointestinal tract bleeding, and recent surgical operation,^8–10 is commonly observed in patients with liver failure who require CKRT.^11,12 For these patients, CKRT was suggested to proceed without the use of any anticoagulants.^3,7 However, patients with liver failure were reported to be hypercoagulative as well.^13,14 The average circuit lifespan of no-anticoagulation CKRT could be as short as 7 hours.⁴

The consideration of severe liver failure as a contraindication to regional citrate anticoagulation is mainly on the basis of the theoretically reduced metabolism rate of sodium citrate.¹⁵ Our previous meta-analysis of cohort studies demonstrated that the average citrate accumulation rate (on the basis of the Ca²⁺_tot/Ca²⁺_ion ratio >2.5) of regional citrate anticoagulation for CKRT in patients with liver failure was 12%, which was significantly higher than the reports in patients without liver failure.¹⁶ However, severe consequences were seldom observed among patients with liver failure with citrate accumulation identified by the Ca²⁺_tot/Ca²⁺_ion ratio.¹⁷ In 2012, a trial by Meijers et al. found that citrate anticoagulation for molecular adsorbent recycling systems (6 hours per session) provided superior patency of the extracorporeal circuit without major adverse events.¹⁸ With the increased evidence, some researchers have suggested that regional citrate anticoagulation CKRT is probably safe in patients with liver failure as well.^17,19–21 Moreover, the current evidence is limited to cohort studies, which results in controversial opinions on the choice of anticoagulation for CKRT in these patients, especially in those with a higher bleeding risk.

Therefore, we performed a randomized controlled trial to assess the efficacy and safety of regional citrate anticoagulation versus no anticoagulation for CKRT in patients with liver failure with a higher bleeding risk.

Methods

Study Design

This study was an open-label randomized controlled trial performed in Xijing Hospital (Xi'an, China), which has 3218 in-hospital beds. The study was approved by the ethics committee of our hospital (No. KY20182037-1), was conducted in accordance with the Declaration of Helsinki,²² and was registered at ClinicalTrials.gov (NCT03791190). Written consent was obtained from all of the included patients or their substitute decision makers before randomization. There was no change to the methods and outcomes after the trial commencement.

Participants

All patients with liver failure who required CKRT in our center were screened on the basis of the inclusion and exclusion criteria (Table 1). The criteria for the diagnosis of liver failure and higher bleeding risk are listed in Table 1. The requirement of CKRT was determined by the attending doctor according to the KDIGO guidelines, mainly including AKI with severe acid–base and electrolyte disorders, volume overload, and uremic complications.³

Table 1.

Criteria and definition for patient recruitment

Inclusion criteria

Age ≥18 yr

Liver failure (acute liver failure, acute-on-chronic liver failure, or decompensated liver cirrhosis)

The requirement of CKRT

Higher bleeding risk

Exclusion criteria

Disagreement of participation

Systemic anticoagulation within 24 h, including heparin, low–molecular-weight heparin, warfarin, protease inhibitor nafamostat, platelet inhibitors (prostacyclin or its analogues), and antiplatelet drugs

Pregnancy and lactation period

Shock with uncorrectable hypoxemia (PaO2 <60 mm Hg even on mechanical ventilation) or systemic hypoperfusion (mean arterial pressure <65 mm Hg) even on vasopressor

Prescribed KRT time <12 h

Definition of liver failure

Acute liver failure is diagnosed on the basis of the occurrence of acute liver injury with impaired synthetic function (INR >1.5, prothrombin activity <40%, and bilirubin >3 mg/dl) and hepatic encephalopathy in patients without cirrhosis or preexisting liver disease

For patients with acute liver failure with proved chronic liver disease, the occurrence of acute severe liver injury was considered as acute-on-chronic liver failure

Decompensated liver cirrhosis was diagnosed on the basis of the presence of dramatic complications (hepatic encephalopathy, ascites, variceal hemorrhage, Child–Pugh score ≥10, INR >1.5, prothrombin activity <40%, and bilirubin >3 mg/dl) in patients with proved cirrhosis (physical examination or evidence of cirrhosis on laboratory or radiologic testing)

Patients with one or more of the following conditions were considered as having a higher bleeding risk

Platelet <40×1000/mm3

Activated partial thromboplastin time >60 s

INR >2.0

Recent or active bleeding

Recent trauma or surgical operation

Recent stroke, intracranial arteriovenous malformations or aneurysm, or retinal hemorrhage

Other conditions that the study group confirmed to be associated with a higher bleeding risk

CKRT, continuous KRT; INR, international normalized ratio.

Randomization

Randomization was performed at a 1:1 ratio. The randomized numbers were generated by using a computer, and blocked randomization (the size of blocks=4, the two larger randomized numbers in each block were assigned to the regional citrate anticoagulation group and the two smaller randomized numbers were assigned to the no-anticoagulation group) was used. The randomization was conducted using opaque envelopes.

Procedures

CKRT was performed by using the Prismaflex with the M100 set (Prismaflex-Gambro Lundia, Lund, Sweden). Continuous venous-venous hemofiltration was used for all of the observed CKRT sessions. The replacement fluid (Chengdu Qingshanlikang Pharmaceutical Co., Ltd.) was infused prefilter and postfilter at 50%/50% with a dose of 2 L/h. The replacement fluid (Ca²⁺ 6.4 mg/dl, Mg²⁺ 1.9 mg/dl, Na⁺ 113 mEq/L, Cl⁻ 118 mEq/L, and glucose 191 mg/dl) was used with the supplement of NaHCO₃ (5%, 595.2 mmol/L) to final Na⁺ 141 mEq/L and HCO₃⁻ 35 mEq/L (4 L of replacement fluid with 250 ml of 5% NaHCO₃ for participants without acid–base disturbance). The addition of KCl (10%, 1.34 mmol/L) depended on the patient's systemic K⁺ level.

In the regional citrate anticoagulation group, the primary blood flow rates were 150–180 ml/min with a sodium citrate (4% trisodium citrate solution, 136 mmol/L, Chengdu Qingshanlikang Pharmaceutical Co., Ltd.) rate of 200 ml/h. The supplement of 5% NaHCO₃ was reduced according to the dose of citrate (8 ml of trisodium citrate to 5 ml of NaHCO₃). On the basis of the postfilter Ca²⁺_ion level (target postfilter Ca²⁺_ion level 1–1.4 mg/dl, the regional citrate anticoagulation group), the blood flow rate or citrate rate was adjusted (Supplemental Table 1).²³ The calcium substitution rate (10% calcium gluconate, 233 mmol/L, Sichuan Meidakang Huakang Pharmaceutical Co., Ltd.) was adjusted according to the systemic Ca²⁺_ion concentration (Supplemental Table 2).²³ If the Ca²⁺_tot/Ca²⁺_ion ratio continuously increased after the adjustment of the calcium substation infusion rate, the citrate rate and blood flow rate were reduced 30 ml/h and 20 ml/min at a time, respectively. In the no-anticoagulation group, CKRTs were conducted without the use of any anticoagulant with a blood flow rate of 200 ml/min. CKRT was discontinued when it was no longer required or no longer consistent with the goal of care.³ The detailed protocol is presented in Supplemental File 1 study protocol.

Outcomes

The primary outcome was filter failure rate within 72 hours. Filter failure was defined as the termination of CKRT because of a transmembrane pressure >300 mm Hg, a visible blood clot that obstructed blood flow through the machine, or the blood pump was unable to rotate because of blood clot obstruction. The filter lifespan was also recorded, which was defined from the beginning to the termination of the observed CKRT session because of any of the following reasons: filter failure, upper time limitation of manufacturer's recommendation (72 hours), and termination of CKRT due to no-clotting events (achievement of treatment goal, surgery or other operations required patient transport, CKRT technical problems, and death).^24,25

Secondary outcomes included bleeding complications, citrate accumulation, hypercalcemia (systemic Ca²⁺_tot >10.0 mg/dl, a posteriori definition), hypocalcemia (systemic Ca²⁺_ion <4.0 mg/dl and Ca²⁺_ion <3.6 mg/dl was severe hypocalcemia, a posteriori definition), severe alkalosis, metabolic acidosis, and the serum Ca²⁺_tot/Ca²⁺_ion ratio, serum Ca²⁺_tot and Ca²⁺_ion, aspartate aminotransferase (AST), alanine aminotransferase (ALT), and bilirubin. The systemic electrocytes and acid–base were measured at 2, 6, 14, and then every 8 hours by the arterial blood gas test during the observed CKRT session. Liver function was tested every 24 hours during the observed CKRT session. Bleeding complications were defined as major bleeding with blood transfusion and/or reoperation requirements and new onset of intracranial bleeding without traumatic events. Citrate accumulation was clinically diagnosed when systemic ionized hypocalcemia (serum Ca²⁺_ion <4.0 mg/dl) was accompanied by at least two additional criteria: (1) a persistent systemic ionized hypocalcemia despite adequate calcium replacement (upregulation of calcium substitution rate for 4–6 hours); (2) a concomitant increase in Ca²⁺_tot concentration (serum Ca²⁺_tot >10.0 mg/dl) and, thus, an increase in the Ca²⁺_tot/Ca²⁺_ion ratio (serum Ca²⁺_tot/Ca²⁺_ion ratio >2.5); and (3) relevant metabolic acidosis (pH <7.2 and/or BE <−5 mEq/L) without or (4) with an increase in the anion gap (serum anion gap >11 mEq/L).^26,27 When the evidence of citrate accumulation was sufficient, regional citrate anticoagulation was stopped and CKRT was performed without the use of any anticoagulant. The patient systemic blood samples were collected, and the serum citrate concentration was tested by using a commercially available kit (R-BIOPHARM Enzymatic BioAnalysis, Cat. No. 10139076035).

Statistical Analysis

According to previous reports, the 72-hour filter failure proportions were approximately 80% and 50% for no-anticoagulation CKRT and regional citrate anticoagulation CKRT in patients with liver failure with a higher bleeding risk, respectively.^16,17,20 The required sample size was calculated with 80% power at two-sided α=0.05. Considering the potential lost before CKRT initiation in critically ill patients, 10% was added to the calculated sample size. Therefore, 45 filters were planned for each group. No interim analysis was planned for this study.

Descriptive statistics are presented as the mean±SD for continuous variables and as relative frequencies for categorical variables. Repeated measurements were recorded as changes occurring during the study period. Comparisons between the groups were calculated by using the t test and the Mann–Whitney rank test for normally distributed and nonnormally distributed continuous variables, respectively. For categorical variables, the chi-squared test or Fisher exact test was used to assess the difference between groups. Filter survival was assessed by using cumulative incidence functions and Gray test when considering death as the competing factor.²⁸ The Fine–Gray subdistribution hazard models were used to adjust for the factors of sex, age, Model for End-Stage Liver Disease score, Sequential Organ Failure Assessment score, Acute Physiology and Chronic Health Evaluation II score, international normalized ratio, and platelet count.²⁸ A two-sided P value < 0.05 was considered statistically significant (SPSS 21.0 and R 4.3.1).

Results

Participant Characteristics

Between September 2018 and September 2021, 168 patients with liver failure with a higher bleeding risk were screened, 90 of whom were included (Figure 1). In addition, 45 patients were randomized to the no-anticoagulation and regional citrate anticoagulation groups. The study was stopped after the inclusion of the planned number of participants. One participant randomized to the regional citrate anticoagulation group was excluded from the final analysis because the participant died within 30 minutes after randomization and did not receive CKRT. Therefore, 45 and 44 participants in the no-anticoagulation and regional citrate anticoagulation groups were analyzed, respectively.

Figure 1.

Patient inclusion flow chart. CKRT, continuous KRT.

Of the included participants, 38 (84%) and 38 (86%) had acute liver failure in the no-anticoagulation and regional citrate anticoagulation groups, respectively. The two groups were comparable in baseline characteristics (Table 2).

Table 2.

Baseline characteristics of participants in a randomized controlled trial of regional citrate anticoagulation versus no-anticoagulation for continuous KRT in patients with liver failure

Variables	No-Anticoagulation Group	Regional Citrate Anticoagulation Group
	n=45	n=44
Age, yr, mean (SD)	55 (17)	56 (16)
Sex, No. (%)
Male	31 (69)	35 (80)
Female	14 (31)	9 (20)
Liver failurea, No. (%)
Acute liver failure	38 (84)	38 (86)
Acute-on-chronic liver failure	6 (13)	5 (11)
Decompensated cirrhosis	1 (2)	1 (2)
Comorbidities, No. (%)
Hypertension	11 (24)	13 (30)
Coronary heart disease	3 (7)	2 (5)
Cardiac arrhythmia	3 (7)	4 (9)
Congestive heart failure	4 (9)	8 (18)
CKD (eGFR <60 ml/min per 1.73 m2)	0 (0)	1 (2)
Diabetes	5 (11)	5 (11)
Reason for a higher bleeding risk, No. (%)
Recent or active bleeding	21 (47)	18 (41)
Recent trauma or surgical operation	8 (18)	10 (23)
Impaired coagulation	28 (62)	28 (64)
Reason for hospital admission, No. (%)
Cardiovascular surgery	14 (31)	15 (34)
Sepsis	12 (27)	9 (21)
General surgery	3 (7)	2 (5)
Trauma	2 (4)	3 (7)
Pancreatitis	0 (0)	2 (5)
Drug/toxic liver failure	2 (4)	4 (9)
Decompensated cirrhosisb	5 (11)	5 (11)
Other	7 (16)	4 (9)
Clinical measures at randomization, No. (%)
Mechanical ventilation	26 (58)	20 (46)
Vasopressors	15 (33)	15 (34)
SOFA score, mean (SD)	12 (4)	12 (3)
APACHE II, mean (SD)	17 (6)	17 (7)
MELD score, mean (SD)	29 (9)	32 (9)
Total bilirubin, mg/dl, mean (SD)	11.0 (8.8)	12.1 (10.5)
ALT, IU/L, mean (SD)	817 (1483)	676 (1349)
AST, IU/L, mean (SD)	1306 (2906)	690 (1721)
Albumin, g/dl, mean (SD)	3.0 (0.6)	3.1 (0.5)
INR, mean (SD)	2.2 (1.5)	2.2 (1.4)
Platelet, 1000/mm3, mean (SD)	54.2 (59.5)	53.8 (39.3)
Creatinine, mg/dl, mean (SD)	3.3 (2.0)	3.8 (2.4)
Lactic acid, mg/dl, mean (SD)	25.2 (11.7)	24.3 (16.2)
Anion gap, mEq/L, mean (SD)	12.3 (6.4)	11.5 (6.6)
AKI stage, 0/1/2/3, No. (%)	5 (11) / 6 (13) / 8 (18) / 26 (58)	5 (11) / 3 (7) / 10 (23) / 26 (59)
Indication for CKRT, No. (%)
AKI stage 2/3	34 (76)	36 (82)
Fluid overload	16 (36)	14 (32)
Electrolyte and acid–base abnormity	21 (47)	28 (63)
Other indicationsc	6 (13)	9 (21)
Time from randomization to CKRT, min, mean (SD)	52.6 (21.3)	57.3 (19.1)
Participants who had CKRT before inclusion, n (%)	18 (40)	14 (32)
Anticoagulation of the last CKRT session before inclusion, no-anticoagulation/regional citrate anticoagulation/heparin, No. (%)	7 (39) / 10 (56) / 1 (6)	8 (57) / 6 (43) / 0 (0)
Filter used time of the last CKRT session before inclusion, mean (SD)	46.4 (21.2)	39.8 (18.7)
Time from the previous regional citrate CKRT to the inclusion, h, median (IQR)	36.3 (26.3–62.5)	34.3 (24.0–70.8)

ALT, alanine aminotransferase; APACHE II, Acute Physiology and Chronic Health Evaluation II; AST, aspartate aminotransferase; CKRT, continuous KRT; INR, international normalized ratio; IQR, interquartile range; MELD, Model for End-Stage Liver Disease; SOFA, Sequential Organ Failure Assessment.

^aThe final diagnosis after the in-hospital examination.

^bThe primary diagnosis of the participants for hospitalization.

^cOther indications included toxicosis, severe hyperpyrexia, and rhabdomyolysis.

Primary Outcome

Participants randomized to the no-anticoagulation group had a significantly higher filter failure rate (56% versus 27%, P = 0.003, Table 3). The result was confirmed by the cumulative incidence function (Figure 2) and the sensitivity analyses when including only the first filters (Table 3). In the cumulative incidence function analysis, the cumulative filter failure rates at 24, 48, and 72 hours of the no-anticoagulation and regional citrate anticoagulation groups were 31%, 58%, and 76% and 11%, 23%, and 35%, respectively. The multivariable Fine–Gray hazard model also demonstrated that the no-anticoagulation group had a significantly higher risk of filter failure (Supplemental Table 3).

Table 3.

Clinical outcomes

Variables	No-Anticoagulation Group	Regional Citrate Anticoagulation Group	Difference between Groups (95% CI)	P Value
Primary outcome, n (%)
Filter failure	25 (56)	12 (27)	—	0.007
Filter failurea	16 (57)	9 (30)	—	0.04
Secondary outcomes, n (%)
Other reasons for circuit termination
Completed 72 h of filter useb	1 (2)	14 (32)	—
Termination of CKRTc	12 (27)	13 (30)	—
Patient death	6 (13)	4 (9)	—
Transportd	1 (2)	1 (2)	—
Actual filter duration, h, mean (SD)	28 (19)	44 (23)	16 (7 to 25)	0.001
Bleeding, n (%)e	2 (4)	0 (0)	—	0.49
RBC transfusion, n (%)	6 (13)	1 (2)	—	0.11
RBC transfusion volume, ml, median (IQR)	400 (291–1811)	900 (900–900)	—	0.86
Plasma transfusion, n (%)	13 (29)	8 (18)	—	0.23
Plasma transfusion volume, ml, median (IQR)	400 (330–930)	410 (238–910)	—	0.97
Other outcomes
Vasopressor initiation after randomization, n (%)f	2 (4)	2 (5)	—	1.00
Increased vasopressor dose after randomization, n (%)g	3 (7)	2 (5)	—	1.00
ICU time, d, median (IQR)	9 (5–18)	10 (7–20)	—	0.52
In-hospital time, d, median (IQR)	16 (9–23)	14 (9–21)	—	1.00
28-d mortality, n (%)	22 (49)	23 (52)	—	0.75

CI, confidence interval; CKRT, continuous KRT; ICU, intensive care unit; IQR, interquartile range; RBC, red blood cell.

^aSensitivity analysis included only the first filter of continuous KRT.

^bThe upper filter use time limitation of manufacturer's recommendation was 72 hours.

^cAchievement of treatment goal.

^dTransport for outward examination or interventional procedure.

^eMajor bleeding required red blood cell transfusion and/or operation/reoperation.

^fNewly prescribed vasopressors during the observed continuous KRT session.

^gIncreased vasopressor dose during the observed continuous KRT session.

Figure 2.

Cumulative incidence function analysis of filter failure.

Secondary Outcomes

During CKRT, only two participants in the no-anticoagulation group had major bleeding (P = 0.49). pH, base excess, and mean arterial pressure were not significantly different at almost all time points during CKRT except at 30 hours (Figure 3). The two groups were not significantly different in the change in total bilirubin, ALT, and AST; newly prescribed vasopressors and increased vasopressors dose during the observed CKRT session; intensive care unit time; and in-hospital time (Table 3 and Supplemental Table 4).

Figure 3.

Serum acid–base, calcium, and citrate concentration and BP during CKRT. (A) Serum pH, (B) base excess, (C) Ca²⁺_ion, (D) Ca²⁺_tot/Ca²⁺_ion, (E) mean arterial pressure, and (F) citrate level during CKRT. #Differences between groups are statistically significant (P < 0.05). *The change in the systemic citrate level was statistically significant (P < 0.05) compared with that at 2 hours.

We observed that the regional citrate anticoagulation group had significantly higher episodes of Ca²⁺_tot/Ca²⁺_ion >2.5 (Figure 3D, Table 4, and Supplemental Table 5), hypocalcemia (51% versus 82%, P = 0.002), and severe hypocalcemia (Figure 3C, Table 4, and Supplemental Table 5), compared with the participants in the no-anticoagulation group. During CKRT, more participants in the regional citrate group received calcium substitution rate upregulation (49% versus 82%, P = <0.001) with an increased dose (median, 5.0 [interquartile range, 5.0–7.5] ml/h versus 7.5 [interquartile range, 5.0–10.0] ml/h, P = 0.023). Except one participant with hypocalcemia who did not have the test of systemic Ca²⁺_ion after the occurrence of hypocalcemia because of filter failure, all hypocalcemia episodes of other participants were improved. Of the 25 participants with Ca²⁺_tot/Ca²⁺_ion >2.5 in the regional citrate anticoagulation group, 22 participants had the test of serum Ca²⁺_tot/Ca²⁺_ion after the occurrence of Ca²⁺_tot/Ca²⁺_ion >2.5. All of these 22 participants had a reduced Ca²⁺_tot/Ca²⁺_ion ratio, and 16 (73%) of them had a normalized Ca²⁺_tot/Ca²⁺_ion ratio after the upregulation of the calcium substitution rates. The citrate and blood flow rates were reduced in four participants (9%). The two groups were not significantly different in severe alkalosis and metabolic acidosis (Table 4 and Supplemental Table 5). No clinical sequela of hypocalcemia (including muscle spasm, arrhythmias, and numbness) and no newly diagnosed hyponatremia or hypernatremia were observed in either group.

Table 4.

Participants who had at least one adverse event episode during the observed continuous KRT sessions

Variables, n (%)	No-Anticoagulation Group (n=45)	Regional Citrate Anticoagulation Group (n=44)
Ca2+tot/Ca2+ion >2.5	3 (7)	25 (57)
Ca2+tot/Ca2+ion >2.5a	3 (11)	20 (67)
Hypercalcemiab	1 (2)	3 (7)
Hypocalcemiac	23 (51)	36 (82)
Hypocalcemiaa,c	15 (54)	25 (83)
Severe hypocalcemiad	6 (13)	34 (77)
Severe hypocalcemiaa,d	6 (21)	24 (80)
Severe alkalosise	4 (9)	4 (9)
Metabolic acidosisf	1 (2)	4 (9)

^aSensitivity analysis included only the first filter of continuous KRT.

^bSystemic total calcium level >10.0 mg/dl.

^cSystemic ionized calcium level <4.0 mg/dl.

^dSystemic ionized calcium level <3.6 mg/dl.

^epH >7.50 and bicarbonate concentration >30 mEq/L.

^fpH <7.20 and bicarbonate concentration <20 mEq/L.

In the regional citrate anticoagulation group, the systemic serum citrate concentration increased quickly during the first 2–6 hours. There was no significant additional increase in systemic citrate concentration after 6 hours (Figure 3F).

Discussion

To the best of our knowledge, this study is the first randomized controlled trial to assess the efficacy and safety of regional citrate anticoagulation versus no anticoagulation for CKRT in patients with liver failure with a higher bleeding risk.³ We showed that regional citrate anticoagulation could significantly prolong the filter lifespan of CKRT in patients with liver failure with a higher bleeding risk, compared with no anticoagulation. In addition, the two groups were not significantly different in the incidences of bleeding, red blood cell transfusion, and change of liver function. None of the participants in the regional citrate anticoagulation group were diagnosed with clinical citrate accumulation. However, the incidences of hypocalcemia and Ca²⁺_tot/Ca²⁺_ion >2.5 were significantly higher in the regional citrate anticoagulation group, most of which were alleviated after the adjustment of the calcium substitution rate.

Patients with liver failure are at higher bleeding risk.^29,30 According to the 2012 KDIGO recommendations, regional citrate anticoagulation and systemic anticoagulation (including heparin) are not currently recommended and left the use of no anticoagulation for CKRT for liver failure patients with higher bleeding risk.³ Despite a higher bleeding risk, CKRT anticoagulation can result in an insufficient filter lifespan.^4,31 Most likely, the coexistence of hypercoagulation in patients with liver failure contributed to the short filter lifespan for no-anticoagulation CKRT.^4,13,14,32 In our present trial, the accumulated 24-hour filter failure rate from the cumulative incidence function analysis was 31% in the no-anticoagulation group, which means that one third of participants required CKRT filter replacement within 24 hours if they accepted no-anticoagulation CKRT. Our previous systematic review and present study showed that patients with liver failure who had regional citrate anticoagulation CKRT were associated with an average filter lifespan >50 hours.¹⁶ Compared with no anticoagulation, regional citrate anticoagulation could prolong the filter lifespan in patients with liver failure with a higher bleeding risk.

For patients with a higher bleeding risk, previous studies indicate that regional citrate anticoagulation is safe as well. The average proportion of bleeding episodes was reported to be 5% for regional citrate anticoagulation CKRT in patients with liver failure.¹⁶ In our present randomized controlled trial, both groups had very low incidences of major bleeding. In addition, the two groups were comparable in red blood cell transfusion and plasma transfusion. Therefore, regional citrate anticoagulation most likely did not increase bleeding risk, compared with no anticoagulation.

During regional citrate anticoagulation CKRT, the filter removes approximately 30%–70% of the administered citrate.^33,34 The remaining citrate enters the systemic circulation and is mainly metabolized by the liver.^15,35 Theoretically, the metabolization of citrate increased the load of liver cells, which potentially aggravated liver function. However, in this study, there was no significant difference in the change in total bilirubin, AST, and ALT between the two groups.

It was reported that the citrate clearance rate was reduced by approximately 50% in patients with severe liver failure,^15,35 which theoretically led to more citrate accumulation.^15,35 On the basis of the Ca²⁺_tot/Ca²⁺_ion ratio, the average citrate accumulation rate was very low (1%–2%) in patients without liver failure.^24,25 However, in patients with liver failure, the incidence of Ca²⁺_tot/Ca²⁺_ion >2.5 could be as high as 5.6%–48.6%.^{17,19,20,36–38} In our present trial, the incidences of Ca²⁺_tot/Ca²⁺_ion ratio >2.5 were very high, which indicated the potential high risk of citrate accumulation on the basis of the Ca²⁺_tot/Ca²⁺_ion ratio.

However, we observed that none of the participants with Ca²⁺_tot/Ca²⁺_ion >2.5 had hypercalcemia and most (73%) of the increased Ca²⁺_tot/Ca²⁺_ion ratios were normalized after the upregulation of the calcium substitution rate. Meijers et al. found that citrate anticoagulation for a molecular adsorbent recycling system (6 hours per session) resulted in reduced serum Ca²⁺_ion and an increased Ca²⁺_tot/Ca²⁺_ion ratio >2.5, which could be normalized by CaCl₂ infusion and patient metabolism.¹⁸ No citrate accumulation was diagnosed during the observed CKRT sessions in our present study,²⁶ which should be restrictedly interpreted because of the limited verification of the used citrate accumulation diagnosis criteria. In addition, the incidence of severe hypocalcemia was significantly higher in the regional citrate anticoagulation group in our present study. Although no clinical consequence of hypocalcemia was observed, more attention should be paid on the systemic Ca²⁺_ion monitoring and calcium substitution rate adjustment. Szamosfalvi et al. provided a continuous venovenous hemodiafiltration-regional citrate anticoagulation Shock citrate protocol, including a low blood flow rate, a low citrate rate, a high dialysate flow rate, port-filter replacement flow, and a higher Ca²⁺ supplementation rate, which resulted in more stable systemic Ca²⁺ levels and a low risk of a high Ca²⁺_tot/Ca²⁺_ion ratio.³⁹ In our further clinical work and clinical research, the upregulation of calcium substitution and the reduction in the citrate dose at the beginning of CKRT would be helpful for the hypocalcemia prevention.

This study has several limitations. First, given the differences in anticoagulation techniques, blinding was not possible. However, most of our outcomes were objective and less affected by the open-label design, which guaranteed the reliability of our conclusions. Second, the trial was performed in a single center. Technical factors could affect filter lifespan and were heterogeneous among centers.⁴⁰ Further multicenter trials are required to validate our findings. In addition, the present trial was not designed for the evaluation of patient mortality and was underpowered to detect small differences in mortality rates between regional citrate anticoagulation and no-anticoagulation CKRT in patients with liver failure. Moreover, some of the included participants received citrate before randomization, which may bias the conclusion. We have confirmed our findings by a sensitivity analysis with the inclusion of only the first CKRT session. Finally, the observed event rates were significantly lower than the expected rates for sample size calculation, which increased the risk of negative results. Studies with larger sample sizes are needed to evaluate the the relationship of CKRT anticoagulation with patient survival.

For patients with liver failure with a higher bleeding risk who required CKRT, regional citrate anticoagulation could significantly prolong the filter lifespan, compared with no anticoagulation. However, regional citrate anticoagulation in patients with liver failure was associated with a significantly higher risk of hypocalcemia, severe hypocalcemia, and Ca²⁺_tot/Ca²⁺_ion >2.5, which required intensive monitoring and timely adjustment of the calcium substitution rate. A reduced citrate dose and increased citrate substitution rate at the beginning of CKRT may be helpful to reduce the risk of hypocalcemia and increased Ca²⁺_tot/Ca²⁺_ion ratio in clinical work and further clinical trials.

Supplementary Material

cjasn-19-151-s001.pdf ^{(622.6KB, pdf)}

Disclosures

All authors have nothing to disclose.

Funding

None.

Author Contributions

Conceptualization: Ming Bai, Shiren Sun.

Data curation: Ming Bai, Jing Jiao, Rui Jing, Yajuan Li, Yangping Li, Yi Liu, Feng Ma, Shiren Sun, Xiujuan Tian, Lei Wei, Yan Yu, Yuan Yue, Lijuan Zhao, Meilan Zhou.

Formal analysis: Ying Liang, Lijuan Zhao.

Investigation: Ming Bai, Jing Jiao, Rui Jing, Yajuan Li, Yangping Li, Yi Liu, Feng Ma, Shiren Sun, Xiujuan Tian, Lei Wei, Yan Yu, Yuan Yue, Lijuan Zhao, Meilan Zhou.

Methodology: Ming Bai, Ying Liang, Shiren Sun, Meilan Zhou.

Supervision: Ming Bai, Shiren Sun.

Writing – original draft: Ming Bai, Yan Yu, Lijuan Zhao.

Writing – review & editing: Ming Bai, Shiren Sun.

Data Sharing Statement

Individual participant data that underline the results reported in this article (text, tables, figures, and appendices) after deidentification, study protocol, and statistical analysis plan will be shared with researchers who provide a methodologically sound proposal to achieve aims in the approved proposal. Data sharing will be available in a period beginning 3 months and ending 3 years following article publication. Proposals should be directed to mingbai1983@126.com or sunshiren@medmail.com.cn.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/CJN/B829.

Supplemental Table 1. Adjustment of trisodium citrate infusion rate.

Supplemental Table 2. Adjustment calcium substitution infusion rate.

Supplemental Table 3. Multivariate analysis of the risk factors of filter failure.

Supplemental Table 4. The change of liver function during the observed CKRT session.

Supplemental Table 5. Patient adverse events at each time point during the observed CKRT sessions.

Supplemental File 1. Study Protocol.

CONSORT Checklist.