Regional Citrate Anticoagulation Compared With No Heparin in Continuous Renal Replacement Therapy Among Patients with Acute on Chronic Liver Failure

Introduction

The global mortality reports on acute on chronic liver failure (ACLF)¹ are alarmingly high, even in India.² ACLF management is complicated by multi-organ involvement, with the kidneys being frequently affected.³ Owing to frequent hemodynamic instability and sickness in ACLF, continuous renal replacement therapy (CKRT) is the preferred modality for dialysis.⁴ However, implementation in these patients is challenging as they are more prone to bleeding, and hence anticoagulation is always a tricky decision. In practical scenarios, anticoagulation is not used in the majority of centers, resulting in increased CKRT downtime due to filter clotting and suboptimal efficacy. Regional citrate anticoagulation (RCA) is the most common anticoagulation modality in the general population, but due to potential citrate toxicity in liver failure, it is avoided.⁵ Moreover, real-world data on ACLF patients with RCA use in CKRT are sparse. Additionally, data on CKRT from emerging nations are absent.⁶ We address this knowledge gap by conducting a retrospective analysis of CKRT in ACLF patients. This report will strengthen the evidence base for the practice of RCA-CKRT in ACLF patients worldwide.

Materials and Methods

This was an ethically approved, observational, retrospective, single-center study, carried out in the Department of Nephrology and Hepatology, ILBS, Vasant Kunj, New Delhi, and patient consent was obtained.

All patients with ACLF, requiring CKRT, admitted to the ICU were eligible. The inclusion criterion was chosen regardless of the etiological ACLF diagnosis or the patient’s age. CKRT was started for following indications:⁴ uremia (with symptoms unresponsive to conservative management), hyperkalemia, and metabolic acidosis (unresponsive to medical therapy), dysnatremia, AKI stage 3, anuria, hyperammonimea⁷ (cut-off>150 mmol/L), hepatic encephalopathy irrespective of ammonia levels, hemodynamic instability (inotropes, not tolerating SLED), refractory volume overload. Patients with ACLF and co-existing chronic kidney disease on maintenance hemodialysis (CKD5D) were excluded from the study. In all patients, continuous veno-venous hemodiafiltration (CVVHDF) was the chosen CKRT modality. The first case was admitted on 10 July 2022, and the last on 26 September 2023. The follow-up was reported until discharge, referral, or death, whichever occurred first. The data were retrieved from electronic records, discharge files, and dialysis charts.

In our center, the Prismaflex machine with A69 filters, as per the patient’s weight, was used for CKRT. Blood flow (Qb), dialysate flow (Qd), and ultrafiltration rate were initially set based on the patient’s clinical profile and adjusted accordingly during the illness. The replacement fluids used were regocit⁸ (citrate=18, Sodium=140, and chloride=86 mmol/L, which produces Sodium chloride of 5.03 g/L and Sodium citrate of 5.29 g/L), biphozyl⁹ (Sodium=140 mmol/L, Potassium=4 mmol/L, Magnesium=0.75 mmol/L, chloride=122 mmol/L, Hydrogen phosphate=1 mmol/L, Hydrogen carbonate=22 mmol/L), and prismasol¹⁰ (Potassium=4 mEq/L, Calcium=2.5 mEq/L, Sodium=140 mEq/L, Chloride=113 mEq/L, bicarbonate=32 mEq/L, lactate=3 mEq/L, dextrose=100 mg/dL, osmolarity=300 mosmol/L). Generally, the CKRT dose target was approximately 30 mL/kg/hr. As all ACLF cases are highly susceptible to hypotension, we maintained a higher pre-dilution 80:20, with a target of 200 mL/hr for the pre-to-post ratio) than the post-dilution as per institutional protocol. To perform RCA CKRT, calcium chloride infusion was made (10 mL of 10% calcium chloride intravenous instilled with 40 mL normal saline in a 50 mL syringe pump with an infusion rate of 10-15 mL/hr, depending on the patient’s initial ionized calcium). The 6-hourly laboratory measurements included total calcium, ionized calcium, and venous blood gas analysis. The 24-hourly laboratory measurements included complete blood count, S. ammonia, S. urea, S. creatinine (Scr), S. sodium, S. potassium, S. magnesium, and S. phosphorus. Citrate toxicity was measured indirectly by dividing total S. calcium with ionized calcium, which was maintained in the 2 to 2.4 mmol/L range by adjusting the infusion rate accordingly.

Data were abstracted from the electronic hospital records. Clinical and laboratory data were obtained. The CKRT details included Qb, Qd, replacement fluid flow, dose, replacement fluid (biphozyl, regocit, and prismasol), number of filters used, and filter duration and clotting (unplanned CKRT stoppage attributed to thrombosis of filter-provided venous access dysfunction was excluded). The data for dialysis catheter placement site, oliguria-anuria, and frequency of adjustment of the calcium rate in RCA-CKRT was missing.

The outcome was measured as time taken to filter the clot in both groups. Filter clotting was reported by the technician and confirmed by the resident doctor. The secondary outcome was change in blood pH, ammonia, and lactate in both groups, following CKRT discontinuation.

ACLF¹¹ was defined by the Asian Pacific Society of Liver Disease¹² criteria. Drug-induced liver injury (DILI)¹³ was diagnosed based on the updated criteria. Severe acute hepatitis (SAH)¹⁴ was defined as per the consensus. Non-alcoholic steatohepatitis (NASH) was diagnosed as per the recent criteria.¹⁵ CLIF C-ACLF score¹⁶ is a scoring system that prognosticates organ failure in ACLF; the formula is 10 × [0.33 × CLIF Organ Failure Score + 0.04 × age, years + 0.63 × log_e(WBC count, 10⁹ cells/L) - 2], where the CLIF-¹⁷ organ failure score is calculated by bilirubin, INR, MAP, Scr, hepatic encephalopathy status, and oxygenation status. The MELD score¹⁸ assesses risk of mortality in liver failure; the formula is 0.957 × Logₑ (creatinine mg/dL) + 0.378 × Logₑ (bilirubin mg/dL) + 1.120 x Logₑ (INR) + 0.6431. MELD-Na¹⁹ was derived with the addition of serum sodium, and was obtained from the formula: MELD-Na = MELD + 1.32 × (137-Na) log_e [0.033 × MELD × (137-Na)]. Hyperlactatemia²⁰ was defined as serum lactate > 2 mmol/L. Citrate toxicity during CKRT procedure was defined as the ratio of ionized to total calcium.

Results

The detailed baseline characteristics of ACLF categorized by groups of CKRT have been described in Table 1. The median age of the overall cohort was 41 (34-49.5) years. Only 10.6% population comprised older patients, >60 years of age. There was a male predominance (88%) in the study composition. The RCA-CKRT group had a similar age and sex distribution compared with NH-CKRT. The etiology of ACLF (n=75) cases was SAH-alcoholic liver disease (ALD) (n=42), which was the most frequent, followed by DILI-NASH (n=10), sepsis-NASH (n=5), and others (n=18). The median CLF OFS was 14 (13-15), and the median CLIF-C ACLF score was 49.95 (44.58-54.775). The cohort had a very high MELD and Na-MELD score of 35 (27-41.6). CVVHDF with pre-dilution higher than post-dilution was used for dialysis to avoid hypotension in all the patients. The mean substitution fluid, blood flow, and dialysate flow rates were 650 (600-775) mL/hr, 150 (100-150) mL/hr, and 650 (600-775) mL/hr, respectively. At baseline, 33% required noradrenaline, and 33% required noradrenaline with vasopressin for maintaining MAP >65 mmHg. The median INR of patients was 2.21 (1.66-3.01). There was no difference in baseline blood parameters between the two groups, as shown in Table 1. The baseline ammonia, pH, and lactate for the whole cohort were 205 (158.6.45-258) mmol/L, 7.37 (7.30-7.40), and 1.9 (1.3-4.05) mg/dL, respectively. There was no baseline difference between these values and the two groups.

The indication for CKRT initiation included hepatic encephalopathy (n=41), hyperammonemia (n=59), non-refractory extra-vascular fluid accumulation (n=25), uremia (n=17), persistent oligo-anuria (n=18), dysnatremia (n=4), hyperkalemia (n=3), and medical refractory acidosis (n=3).

The length of stay from hospital admission to last follow-up or death was 11 (6.5-18) days for the whole cohort. This duration was similar for both groups [12 (7-17) vs. 10 (7-17); p-value = 1]. During the study, 25 patients died, of which 9 (42%) and 16 (46%) died in the NH-CKRT and RCA-CKRT groups, respectively.

Overall, 38/158 filters used registered filter clotting. The mean number of filters (2.56 ± 18 vs. 1.5 ± 1.1) (p-value = 0.018) was higher for NH-CKRT than RCA-CKRT. The CKRT duration was 42 (24-53) hours. The RCA-CKRT group also recorded a higher time to filter clotting than NH-CKRT [54 (36-67.5) vs. 47 (28-54)] hours. The time to event analysis was performed by adjusted Cox regression analysis where the CLIF score was used as a covariate [Figure 1]. The filter clotting was more reported in NH-CKRT compared with RCA-CKRT (adjusted HR = 4.5 (1.54-13) (p-test=0.006). Additionally, the Cox analysis showed that most CKRT sessions > 24 hours long belonged to the RCA-CKRT group. The total mortality reported in the study was 25 of 75 (33.3%). Table 2 shows the laboratory parameter changes during the course. There was no difference in the mean change in ammonia,pH,and lactate between the two groups before and after CKRT. Also, there was no difference in pre- and post-hemoglobin decline between the two groups (1 ± 0.4 vs. 0.9 ± 0.45 g/dL; p-value = 0.1).

Close

Figure 1:

An adjusted (CLIF ACLF) filter clotting difference between the two groups. ACLF: Acute on chronic liver failure, CLIF: Chronic liver failure consortium

Export to PPT

Of 75 cases, 13 had acidosis (pH <7.2) in the cohort. NH-CKRT and RCA-CKRT had 18 (29%) and 3 (23%) patients with acidosis at the completion of CKRT, respectively. This difference was statistically non-significant (p-value = 0.476). No episode of hypocalcemia was detected, as calcium monitoring was strictly done throughout CKRT.

Discussion

We report one of the largest Indian data sets comparing RCA use with NH in CKRT for liver patients. Our profile comprised ACLF cases, with very high MELD and CLIF-ACLF scores. High mortality was reported, and there was no difference between the groups. The primary outcome of filter clotting was lesser in RCA-CKRT compared with NH, and this difference was statistically significant. There were no reports of severe hypocalcemia in the RCA-CKRT group. We did not find any additional episodes of metabolic acidosis in the RCA group compared with the NH-CKRT group. In this study, the majority had hyperlactatemia, but the lactate levels pre- and post-CKRT did not differ significantly during RCA

The high mortality in the cohort could be explained by the sicker profile of the patients, which is supported by their high baseline MELD and CLF-SOFA scores.²¹ This is also concordant with previous literature published on patients with ACLF.²²

We effectively used RCA, as suggested by the studies supporting the rationale behind its use, showing prolonged filter life and life span, even in patients with liver dysfunction.^23,24

The feasibility and safety of citrate in ACLF patients were also confirmed by the lack of hypocalcemia/metabolic acidosis episodes.

There have been limited studies in this area. However, the recently conducted RCT^25,26 in China showed similarly improved circuit life in RCA but at the cost of a high risk of severe hypocalcemia and an additional incidence of metabolic acidosis. These were absent in our report, possibly corresponding to a lower citrate dose. There was no increase in lactate post-CKRT, which is similar to a previous study²⁷ and highlights that no absolute pre-CKRT lactate value, but increasing trends, might be an early marker of citrate toxicity.²⁸ As per the literature search for citrate protocol, we found that the starting dosage of citrate and subsequent titration protocols varied across centers and clinical scenarios. The published literature has higher citrate doses, ranging from 2.5 to 4.5 mmol/L. There is a lack of data on RCA usage in ACLF patients, especially in Indian settings. This makes the absence of increased metabolic complications with low-dose RCA in our study encouraging. Another observation from our report worth mentioning is the fact that even at a lower dose, citrate provided a better filter life span compared with NH. This finding is clinically important because frequent CKRT kit discontinuation for any reason compromises the already poor hemodynamic and renal functions. Additionally, the cost associated with filter clotting is a matter of concern in resource-restricted settings where affordability always remains an important issue. Nevertheless, work is still in progress, and further studies are being planned to make RCA-CKRT more useful in ACLF cases.²⁹ This was a retrospective study. The RCA group had a relatively small sample size. The filter clotting detection could create an observer bias; however, the clot episode was always counter-checked by resident doctors. CKRT downtime and hypotension episodes during CKRT were not captured. Long-term follow-up of patients was not possible, leading to attrition bias; however, this factor hardly affected the primary outcome of filter lifespan. The eGFR was not calculated on the basis of cystatin C as the data were not captured. Serum citrate levels were not checked as per our hospital protocol. But, we used regular calcium checking to indirectly assess the toxicity, which is being used worldwide as a standard practice in RCA-CKRT. There was no data collection for the frequency of changing the RCA settings depending on the ionized calcium. To overcome the limitations, we have planned to conduct an RCT with different CKRT dosing, which will estimate the effect of RCA in ACLF patients.

This is the first study comparing RCA vs. NH in the ACLF group in Indian patients. The low-dose citrate in the RCA protocol during CVVHDF CKRT in these patients is effective and safe without any added burden of metabolic complications. Future RCTs with different dosing and titration techniques would be helpful in this research area.