Introduction

Acute‑on‑chronic liver failure (ACLF) is the most severe complication of liver cirrhosis, with a 28‑day mortality rate of 30% or higher, depending on the number of dysfunctional organ systems.¹ It is characterized by acute decompensation of liver cirrhosis, complicated by failure of various organ systems (liver, kidneys, brain, coagulation, circulation, and lungs) and severe systematic inflammation.^2,3 The PREDICT study⁴ identified different precipitants for ACLF, that is, bacterial infection, severe alcoholic hepatitis, hemorrhagic shock, and drug‑induced toxic encephalopathy. Currently, management of patients with ACLF consists in identification and treatment of precipitating factors as well as organ failure therapy. No specific treatment can be offered to this group of patients, and liver transplantation (LT) remains the only curative and potentially life‑saving therapeutic option. Recent data confirmed that patients with ACLF grades 1 and 2 should be listed for LT; however, in the case of ACLF grade 3 patients, there are still uncertainties regarding timing and selection for LT.^5-7 On the other hand, in terms of eligibility, patients with ACLF may present with contraindications for LT, that is, uncontrolled bacterial and fungal infections. In this case, bridging strategies, for example, extracorporeal liver support (ECLS) systems such as albumin dialysis, are required. The most recent network meta‑analysis comparing and ranking different liver support systems and standard medical treatment in patients with ACLF included 16 trials, and assessed the Molecular Absorbent Recirculating System, Fractionated Plasma Separation, Adsorption and Dialysis (Prometheus), extracorporeal cellular therapy, total plasma exchange, and the BioLogic‑DT sorbent suspension dialyzer.⁸ Although albumin dialysis can improve the severity of hepatic encephalopathy, there is no evidence that it improves survival of patients with ACLF.⁸ Moreover, the quality of evidence was moderate to very low. The recent European Association for the Study of the Liver Clinical Practice Guidelines on ACLF do not recommend routine use of artificial or bioartificial ECLS or plasma exchange in ACLF outside investigative trials.⁹ This leaves single‑pass albumin dialysis (SPAD) as the last remaining option; however, it lacks a standardized management algorithm. We aimed to analyze retrospective data from a single liver transplant center to evaluate the effectiveness of SPAD performed in patients with ACLF of alcoholic origin, with particular focus on laboratory and clinical parameters associated with liver failure.

Patients and methods

We performed a retrospective single‑center data analysis of all ACLF patients treated with SPAD between 2019 and 2022. In our center, SPAD is indicated as a therapeutic option only in the patients who simultaneously present symptoms of hepatic encephalopathy grade II/III and progressive acute kidney injury. Usually, these are individuals with ACLF of alcoholic origin, for whom LT is not considered, mainly due to active alcoholism.

Clinical data, including age and sex, as well as laboratory parameters, disease severity scores, and the need for organ support before and after treatment were recorded.

For SPAD, a multiFiltratePRO hemodialysis device (Fresenius SE & Co. KGaA, Bad Homburg v. d. Höhe, Germany) with a high‑flux polysulfone membrane was used. The procedure consisted of 3 sessions per individual, performed according to the center’s protocol as follows: the blood flow rate was maintained at 100 to 120 ml/min and calcium‑free dialysis fluid enriched with human serum albumin to obtain a 4% albumin concentration was pumped at 1400 ml/h. The SPAD procedure lasted for 8 hours and was followed by continuous veno‑venous hemodialysis for up to 64 hours, with no change of the dialyzer; the blood flow rate was reduced to 60 to 100 ml/min and albumin‑free dialysis fluid was pumped at a flow rate of 1000 to 1500 ml/h. The albumin solution did not recirculate and was disposed of after passing through the hemodialysis filter. The system is depicted in Supplementary material, Figure S1.

Vascular access was obtained through a double- or triple‑lumen hemodialysis catheter placed in the femoral or jugular vein.

Results

Between 2019 and 2022, a total of 20 patients with ACLF of alcoholic origin underwent SPAD in our center. Their median (IQR) age was 41.5 (37–51) years, and a majority (n = 16 [80%]) were men. At the start of the SPAD procedure, the patients had a median (IQR) Model for End‑stage Liver Disease (MELD) score of 34 (28.6–40.4), MELD‑Na score of 34 (25.8–40.8), MELDNa score of 35 (29.3–40.4), MELD 3.0 score of 35 (30.6–43.3), Chronic Liver Failure Consortium (CLIF‑C) Organ Failure score of 13 (11–14.5), and CLIF‑C ACLF score of 57 (50.7–63.2). Most patients (65%) were classified as ACLF grade 3. Group characteristics are presented in Supplementary material, Table S1. Only 1 patient (5%) was transplanted and 3 individuals (15%) survived. The remaining patients (85%) died during hospitalization. The median (IQR) length of hospitalization in the study group was 26 (13.5–37.5) days. We observed a significant difference in 30‑day mortality between the patients with MELD scores below 30 (43%) and those with scores above 30 points (92%) (P = 0.045). The Kaplan–Meier analysis of 30‑day mortality is presented in Supplementary material, Figure S2.

The SPAD procedure resulted in a significant decrease in total bilirubin, serum albumin, and serum creatinine levels, with a median (IQR) reduction of bilirubin of 4.19 (1.37–8.23) mg/dl, median (IQR) reduction of albumin of 0.2 (0.05–0.45) g/dl, and median (IQR) reduction of creatinine of 0.54 (0.12–1.17) mg/dl (Table 1). Serum sodium concentration did not change significantly during the treatment.

White blood cell count and international normalized ratio remained unchanged during therapy, while platelet count dropped significantly, with a median decrease of 29.5 (9.5–62) G/l (Table 1).

Median MELD and MELDNa scores significantly decreased by a median (IQR) of 5.26 (0.86–7.31) and 3.56 (0.21–6.28) points, respectively (Table 1). The CLIF‑C ACLF score showed no significant changes during the treatment.

Discussion

Impaired detoxifying function of the liver in ACLF results in accumulation of toxic substances, both hydrophilic (which could be eliminated through conventional dialysis) and hydrophobic. This leads to organ dysfunction, clinically observed as worsening hepatic encephalopathy and kidney injury.^10,11 ECLS systems were developed to enable removal of hydrophobic toxins accumulated due to liver failure.¹² In our study, a significant reduction in the total bilirubin level after the SPAD procedure was observed. This can be considered a determinant of successful albumin‑bound toxin detoxification in clinical practice, as previously shown in the literature.^13-15 Clearance of albumin‑bound toxins depends on the dialysate flow.¹⁶ In our protocol, SPAD duration was set at 8 hours with a 1400 ml/h flow, and resulted in a median (IQR) reduction of total bilirubin level of 4.19 (1.37–8.23) mg/dl, which corresponds to a median reduction of 17.9% of its initial serum concentration. In comparison, Sommerfeld et al¹⁴ showed a median reduction of total bilirubin concentration during SPAD of 9.7% in a SPAD time cycle of 7 hours and a flow rate of 700 ml/h.

In our investigation, a significant reduction of serum creatinine level was observed. This represents effective detoxification of water‑soluble substances. Sponholz et al¹⁷ claimed that SPAD was not effective in achieving a significant decrease in the concentration of creatinine and even induced a median increase of 5.04% in creatinine level. The observed difference may also be associated with a lower albumin dialysate flow rate of 700 ml/h in their study. Increasing the dialysis flow rate can lead to a higher detoxification capacity, as shown in our study, but it is accompanied by higher treatment costs. Similar findings on the key role of dialysate flow rate in SPAD were also described by Piechota et al.¹⁸

Interestingly, in our evaluation, serum albumin level significantly decreased during SPAD, which was not recorded in previous studies. We propose that this decrease is a result of poor functional status inherent in malnutrition and sarcopenia in patients with ACLF of alcoholic origin.

ACLF patients are at a high risk of hemorrhage, especially in the setting of bacterial infections, which can be a precipitant of variceal bleeding.¹⁹ Incidence of bleeding complications during ECLS varies between 9% and 40% according to the literature, mostly due to gastrointestinal hemorrhage or diffuse bleeding or from indwelling catheters.^14,20,21 We observed a significant decrease in platelet count during SPAD treatment. A similar platelet count reduction was reported in previous studies on ECLS systems.^14,15,22 Therefore, patients treated with ECLS, including SPAD, require close monitoring of clinical and laboratory markers of bleeding.

With the significant reduction of median serum creatinine and total bilirubin levels, SPAD also affects median MELD and MELDNa scores, both of which significantly decreased in our study. Interestingly, a more sophisticated score, CLIF‑C ACLF, remained unchanged during the procedure and reflected poor prognosis of ACLF patients in this study. This discrepancy most likely results from the fact that, unlike the MELD score, the CLIF‑C ACLF score includes not only laboratory parameters, but also clinical findings. These data coincide with existing evidence that multiorgan failure scores are more effective than traditional liver function metrics in predicting mortality.²³ Notably, 85% of the patients died during hospitalization, most of them within the first 10 days of observation. This shows the need for early listing for transplantation of these patients (in the absence of contraindications), preferably in the first 7 days.²⁴ The observed significant difference in 30‑day mortality between patients with MELD scores below 30 (43%) and those with scores above 30 points (92%) may be helpful in determining who can benefit from SPAD in terms of eligibility for transplantation.

This study has several limitations, which include its monocentric and retrospective design with a small participant cohort reflecting the rarity of ACLF as well as the lack of comparable data. SPAD is not a standardized method, neither in terms of the procedure protocol nor in the selection of the treated patients. Therefore, we cannot exclude selection bias. Our study may be viewed as part of the discussion on evaluation of the SPAD procedure.

Our findings contribute to the discourse on SPAD effectiveness, suggesting that while albumin dialysis devices may not significantly improve survival in ACLF, their technical limitations hinder the removal of all toxins, particularly those that are not albumin‑bound or are too large to filter out. Currently, only early LT has been shown to change the natural course of ACLF²⁴; no ECLS system tested to date has achieved this goal.^14,15,25 These devices do not restore the critical synthetic and metabolic functions of the liver, which are essential for homeostasis and immunity. Limitations may arise from both technical factors, such as the membranes and filters used, and from gaps in our understanding of ACLF pathophysiology. Therefore, SPAD should not replace timely transplantation evaluations of ACLF patients.