Plasma exchange in patients with acute and acute-on-chronic liver failure: A systematic review

Eunice Xiang-Xuan Tan, Min-Xian Wang, Junxiong Pang, Guan-Huei Lee

Abstract BACKGROUND Acute liver failure (ALF) and acute-on-chronic liver (ACLF) carry high short-term mortality rate, and may result from a wide variety of causes. Plasma exchange has been shown in a randomized control trial to improve survival in ALF especially in patients who did not receive a liver transplant. Other cohort studies demonstrated potential improvement in survival in patients with ACLF.AIM To assess utility of plasma exchange in liver failure and its effect on mortality in patients who do not undergo liver transplantation.METHODS Databases MEDLINE via PubMed, and EMBASE were searched and relevant publications up to 30 March, 2019 were assessed. Studies were included if they involved human participants diagnosed with liver failure who underwent plasma exchange, with or without another alternative non-bioartificial liver assist device.RESULTS Three hundred twenty four records were reviewed, of which 62 studies were found to be duplicates. Of the 262 records screened, 211 studies were excluded.Fifty-one articles were assessed for eligibility, for which 7 were excluded.Twenty-nine studies were included for ALF only, and 9 studies for ACLF only.Six studies included both ALF and ACLF patients. A total of 44 publications were included. Of the included publications, 2 were randomized controlled trials, 14 cohort studies, 12 case series, 16 case reports. All of three ALF studies which looked at survival rate or survival days reported improvement in outcome with plasma exchange. In two out of four studies where plasma exchange-based liver support systems were compared to standard medical treatment (SMT) for ACLF,a biochemical improvement was seen. Survival in the non-transplanted patients was improved in all four studies in patients with ACLF comparing plasma exchange vs SMT. Using the aforementioned studies, plasma exchange based therapy in ACLF compared to SMT improved survival in non-transplanted patients at 30 and 90-d with a pooled OR of 0.60 (95%CI 0.46-0.77, P ＜ 0.01).CONCLUSION The level of evidence for use of high volume plasma exchange in selected ALF cases is high. Plasma exchange in ACLF improves survival at 30-and 90-d in nontransplanted patients. Further well-designed randomized control trials will need to be carried out to ascertain the optimal duration and amount of plasma exchange required and assess if the use of high volume plasma exchange can be extrapolated to patients with ACLF.

Key words: Acute-on-chronic liver failure; Acute liver failure; Plasmapheresis; Plasma exchange; Liver failure

INTRODUCTION

Acute and acute-on-chronic liver (ACLF) carry high short-term mortality rate, and may result from a wide variety of causes. Regardless of its underlying etiology, liver failure at its final stages results in jaundice, hepatic encephalopathy, hepato-renal syndrome, hemodynamic instability, increased susceptibility to severe infections and finally multi-organ failure[1].

Acute liver failure has been defined as a rapid decline in hepatic function characterized by jaundice, coagulopathy, and hepatic encephalopathy in patients with no prior liver disease. There are more overlaps in terminologies for ACLF, and there are currently more than ten definitions of ACLF. The two most widely used definitions are from the Asian Pacific Association for the Study of the Liver (APASL)and the European Association for the Study of the Liver (EASL) Chronic Liver Failure consortium[2]. Besides treating the underlying etiologies and supportive therapy, liver transplantation is the only definitive therapy for those with advanced disease.However, the availability of donor organ limits the availability of the patients that can be saved.

In recent years, there is increasing interest in plasma exchange for the treatment of liver failure. Since Larsen et al[3]published the first open randomized control trial of plasma exchange in patients with acute liver failure in 2016, plasmapheresis has been added to the armamentarium. High volume plasma exchange has been included in European guidelines[1]as level I, grade 1 recommendation in management of acute liver failure. Its proposed mechanism is removal of plasma cytokines and drivers of systemic inflammatory cascade through plasma exchange. Preceding the aforementioned publication, published studies on the use of plasma exchange in the setting of liver failure were mostly retrospective case series or cohort studies. These studies differed greatly in the protocols of plasma exchange. In ACLF, the data is less clear.

The objectives of this review is to provide a summary and analysis of the current evidence for the use of plasmapheresis in patients with ACLF and acute liver failure(ALF) and its effect on mortality particularly in the non-transplanted patients. In addition, the review will summarise the current literature on volume of plasma used during exchange, the duration and frequency of plasma exchange and briefly outline other available apheresis or liver support devices used in liver failure.

MATERIALS AND METHODS

Eligibility criteria

We included studies ranging from case reports to randomized control trials that have been published till 30 March, 2019. We excluded abstracts in this review and have restricted to only studies in English. We excluded studies with insufficient information concerning our outcomes of interest and areas of comparison: e.g.,survival, the volume of plasma exchange and type of product exchanged. We included studies with only human participants diagnosed with liver failure who underwent plasma exchange, with or without other alternative liver support systems.There were no restrictions on the dose, duration, and type of plasma exchange (Table 1 for PICOS criteria). A PRISMA checklist was also used to guide the development of the systematic review.

Information sources

A comprehensive search of databases and conference proceedings to identify all relevant studies up to 30 March, 2019 was performed. The following electronic databases were searched: MEDLINE via PubMed, and EMBASE. We use both text words and medical subject heading terms. The literature search strategy was adapted to suit each database.

For example, on PubMed we use the combination of the following medical subject heading terms "plasma exchange" or "plasmapheresis" and "liver failure" or "acute liver failure" or "acute on chronic liver failure”. Search was limited by “Case reports”,“Classical article”, “Clinical study”, “Clinical trial”, “Controlled clinical trial”,“Observational study”, “Randomized controlled trial”, “Review”, “Humans”,“English”, “Core clinical journals” and “MEDLINE”. The methods for data collection and analysis were based on the Cochrane Handbook of Systematic Reviews for Interventions. Where clarification of information in published data was required,corresponding authors were contacted through electronic mail for clarification.

Selection of studies, data collection and summary measures

Two review authors (Tan EXX and Lee GH) independently reviewed relevant material identified from the above search. After reading the titles and abstracts of the identified articles, full-text articles of all citations deemed to meet the inclusion criteria were sought. Duplicates were excluded. Each article was independently inspected to verify that they meet the pre-specified inclusion criteria. Study selection process is being summarized in Figure 1. Studies that were included in this systematic review are included in Tables 2-5. We created a case report form specifically for this study for systematic study review/selection and structured data extraction. Relevant study data was independently reviewed selected and extracted. Outcomes of interest such as allcause mortality, changes in liver biochemistry, and survival in non-transplanted patients were primary outcomes of interest. The volume of plasma exchange used,duration of exchange, and etiology of liver failure were also compared in conjunction with study outcomes of interest.

In addition, pooled odds ratios and its corresponding 95% confidence intervals were respectively calculated for 30- and 90-d mortality in ACLF patients using the random effects model. The data extracted for this calculation includes the number of events (deaths) for the respective time periods and total sample size in the intervention and control arms, and were extracted in duplicate by ET and MXW. The I2statistic and Cochran Q test was used to evaluate statistical heterogeneity, where heterogeneity was characterized as minimal (＜ 25%), low (25%-50%), moderate (50%-75%) or high (＞ 75%) and was significant if P ＜ 0.05. All calculations performed were 2-sided and done through Review Manager 5.3.

Table 1 Participants, interventions, comparisons, outcomes and study design criteria used to define the research question for this systematic review

RESULTS

Search results

A total of 324 records were reviewed, of which 62 duplicates studies were removed.Of the 262 records screened, 211 studies were excluded. Fifty-one articles were assessed for eligibility, for which 7 were excluded (Figure 1). Twenty-nine studies were included for ALF only, and 9 studies for ACLF only. Six studies included patients who had both ALF and ACLF. A total of 44 publications were included(Figure 1).

Plasma exchange in ALF

A total of 35 studies included patients with ALF (Table 2, Table 3, and Table 5). Of this, 24 were studies in adults and 11 in the pediatric population (Table 3 and Table 5).In the studies that included adult subjects, 4 also included patients with ACLF. Of the 24 studies in adults, there was 1 randomised controlled trial, 4 cohort studies, 9 case series and 10 case reports. Of the 11 studies with pediatric subjects, there was 1 cohort study, 4 case series and 6 case reports; and 2 of the 11 studies included patients with both ALF and ACLF.

Mortality

There is only one randomized trial to date[3]that assessed transplant-free survival comparing standard medical treatment (SMT) vs plasma exchange and SMT in patients with ALF. In the Larsen et al[3]study, high volume plasma exchange (HVP)increased survival in non-transplanted patients after three months. However, there was no significant difference in the effect of HVP in patients who received emergency liver transplantation. Of the three studies comparing plasma exchange vs SMT or alternative liver support systems, all reported an improvement in survival in patients who did not undergo liver transplant[3,4]or improvement in survival days[5].

Biochemical improvement

All studies that assessed biochemical improvement pre- and post-plasma exchange,found an improvement in biochemical parameters such as coagulopathy, bilirubin,aspartate aminotransferase (AST), alanine aminotransferase or ammonia. However,biochemical improvement did not directly relate to mortality outcome. Even in the patients who did not survive, there was also biochemical improvement post- plasma exchange[5].

Standard vs high volume plasma exchange and other liver assist devices

There is heterogeneity in the amount of plasma exchange a patient gets in ALF amongst the various studies. Two studies[3,6]used plasma exchange at least 15% ideal body weight removal at 1-2 L per hour while in Buckner et al[7]'s case series, plasma exchange with 10 L of donor plasma regardless of weight was used. Similar to Buckner et al[7], Damsgaard et al[8]in a case report used 8-9 L of plasma per session for plasma exchange for a patient with ALF from Wilson's disease, who survived without need for liver transplant. In contrast, in case series by Akdogan et al[9], only oneplasma volume was being exchanged daily till patient expired or improved. Majority of studies used approximately 2-4 L of fresh frozen plasma at each plasma exchange[4,10-18].

Figure 1 Summary of study selection process.

While there are stark differences in amount of plasma exchanged in these aforementioned studies, in the study of the use of plasma exchange in the treatment of acute liver failure, there is only one open randomized controlled trial[3]favoring the use of high volume plasma exchange over standard medical treatment. There are no head to head studies comparing high volume to standard volume plasma exchange.Although most cohort studies or case series that used 2-4 L of plasma and or additional fluid for plasma exchange saw positive results whereby there were reported improvement in biochemical parameters such as bilirubin and coagulopathy[4,5,9,11,12,14-16,18,19], and some of which also reported increased transplant-free survival days[5], at present evidence favor large volume plasma exchange for treatment of ALF[3].

There is currently no clear evidence to support the use of other assist devices in addition to plasma exchange in management of ALF. Several studies[4,10,20]included alternative assist devices to plasma exchange and made comparisons of its efficacy in the treatment of acute liver failure. For example, comparing plasma exchange vs plasma exchange + continuous venovenous hemodiafiltration (CVVHDF), Nakae et al[10]showed that the latter resulted in a decrease in inflammatory mediators and an increase in citrate compared to the former group. Another study, also by Nakae et al[10], reported use of plasmadiafiltration, a blood purification therapy where plasma exchange is performed using a selective membrane plasma separator while the dialysate flows outside the hollow fibers for management of ALF. In that study[21], lessplasma was used per cycle: 1200 mL fresh frozen plasma (FFP) and 50 mL of 25%albumin per session. However, patients had an average of 8.3 cycles of plasmadiafiltration, which is higher compared to other studies (Table 2). Transplantfree survival rate was 38.1%, 54.5% in ALF, and 20% in fulminant hepatitis; there was no control arm. Pediatric studies were evaluated separately, and in the included pediatric studies[19,22-27], the amount of plasma exchange per session ranged from 1-4 plasma volumes per exchange.

Table 2 Studies included for study of plasmapheresis in acute liver failure in adults

?

?

PE: Plasma exchange; SMT: Standard medical treatment; ALF: Acute liver failure; HLH: Hemophagocytic lymphohistiocytosis; FFP: Fresh frozen plasma;INR: International normalised ratio; ALT: Alanine aminotransferase; CVVHDF: Continuous venous-venous hemodiafiltration; CHDF: Continuous hemodiafiltration; HBV: Hepatitis B virus; HSV: Herpes simplex virus; LDH: Lactate dehydrogenase; DILI: Drug induced liver injury; AFLP: Acute fatty liver of pregnancy; DILI: Drug induced liver injury; TPE: Therapeutic plasma exchange.

Table 3 Studies included in for study of plasmapheresis in acute liver failure in pediatric cohort

CPFA: Continuous plasma filtration adsorption; HBO: Hyperbaric oxygen; PE: Plasma exchange; SMT: Standard medical treatment; ALF: Acute liver failure; HLH: Hemophagocytic lymphohistiocytosis; FFP: Fresh frozen plasma; INR: International normalised ratio; ALT: Alanine aminotransferase;CVVHDF: Continuous venous-venous hemodiafiltration; CHDF: Continuous hemodiafiltration; HBV: Hepatitis B virus; HSV: Herpes simplex virus; LDH:Lactate dehydrogenase; DILI: Drug induced liver injury; AFLP: Acute fatty liver of pregnancy; DILI: Drug induced liver injury; TPE: Therapeutic plasma exchange; AFP: Alphafetoprotein.

FFP vs albumin

Most studies used 100% FFP for plasma exchange with the exception of few studies[13,19,21,28-31], where plasma substitutes or albumin were used in conjunction with plasma. For example, in a case series by Liu et al[31], Seven liters of fluid was used for plasma exchange, but the first 4.7 L was composed of fresh frozen plasma, while the rest comprised of plasma consisting of 25% human albumin, 0.9% saline and Ringer's solution.

There are by far no studies that use pure albumin as replacement fluid for plasma exchange in ALF. However, Collins et al[32]has described in their case report, a patient with fulminant hepatitis from Wilson's disease who underwent single-pass albumin dialysis (SPAD) with improvements in bilirubin. Although of note, the same patient underwent plasma exchange after stopping SPAD in view of serum copper rebound.

Three-day therapy versus intermittent or response guided

As aforementioned, only few[3,6,18,33]studies used a strict consecutive daily or everyother day 3-d therapy plasma exchange regime as in the open-RCT by Larsen et al[3].Instead, most studies continued plasma exchange till patient dies, or improves clinically, or receives a liver transplant at a range of intervals from every other day to intermittent (as and when necessary). Buckner et al[7]reported an interesting finding in their case series where a patient with halothane toxicity and acute liver failure received plasma exchange with 5-10 L of plasma almost daily for 37 d before she roused from coma. Few studies did not include detailed information on the frequency of plasma exchange[5,29,34.

Table 4 Studies included for study of plasmapheresis in acute-on-chronic liver failure

?

?

?

PE: Plasma exchange; SMT: Standard medical treatment; ALF: Acute liver failure; HLH: Hemophagocytic lymphohistiocytosis; FFP: Fresh frozen plasma;INR: International normalised ratio; ALT: Alanine aminotransferase; CVVHDF: Continuous venous-venous hemodiafiltration; CHDF: Continuous hemodiafiltration; HBV: Hepatitis B virus; HSV: Herpes simplex virus; LDH: Lactate dehydrogenase; DILI: Drug induced liver injury; AFLP: Acute fatty liver of pregnancy; DILI: Drug induced liver injury; TPE: Therapeutic plasma exchange; DPMAS: Double plasma molecular adsorption system; AOCLF:Acute on chronic liver failure; PTA: Prothrombin activity; PBA: Plasma bilirubin adsorption; HE: Hepatic encephalopathy; DPMAS: Double plasma adsorption system; ALSS: Artificial liver support system; NBAL: Non-bioartifical liver support; SALF: Subacute liver failure; PP: Plasma perfusion; HRS:Hepatorenal syndrome; PT: Prothrombin time; UCMSCs: Umbilical Cord-Derived Mesenchymal Stem Cell Transplantation.

Etiology specific outcome

Not all studies assessed included etiology of liver failure and its effect on transplantsurvival. In larger cohort studies, predominant causes of ALF include paracetamol,followed by unknown cause. As already been discussed in previous literature,paracetamol-induced ALF has improved prognosis compared to injury from other causes, for example, viral hepatitis causes. From the earlier studies[6,14]where in two case series survival in ALF patients receiving plasma exchange was 50%-55%, the subgroup of patients who had paracetamol-induced ALF had survival ranging from 83%-100%. However, these are from case series and the level of evidence is not strong.In a Chinese cohort[4]study comparing the efficacy of plasma exchange +hemoperfusion + CVVHDF to plasma exchange + CVVHDF and hemoperfusion +CVVHDF, treatment of the 61 patients using the artificial liver support system yielded a combined survival rate of 62.3% (38/61). When subdivided into viral versus nonviral groups, the viral group survival rate was 35.0% (7/20) while the non-viral group survival rate was 75.6% (31/41).

Table 5 Studies included for use of plasmapheresis in acute liver failure and acute-on-chronic liver failure in adults

?

PE: Plasma exchange; SMT: Standard medical treatment; ALF: Acute liver failure; HLH: Hemophagocytic lymphohistiocytosis; FFP: Fresh frozen plasma;INR: International normalised ratio; ALT: Alanine aminotransferase; CVVHDF: Continuous venous-venous hemodiafiltration; CHDF: Continuous hemodiafiltration; HBV: Hepatitis B virus; HSV: Herpes simplex virus; LDH: Lactate dehydrogenase; DILI: Drug induced liver injury; AFLP: Acute fatty liver of pregnancy; DILI: Drug induced liver injury; TPE: Therapeutic plasma exchange; BA: Biliary atresia; IEM: Inborn errors of metabolism.

The use of plasma exchange in patients with fulminant liver failure from Wilson's disease has been reported in case reports and series with encouraging outcomes[8,19,23-25,30,32,35]. EASL guidelines for Wilson's disease[36]recommends that patients with acute liver failure due to Wilsons disease should be treated with liver transplantation when revised King's score is 11 or higher (Grade II-2 B1 Class I, Level B). In one case report[8], a patient who met criteria for requiring liver transplant improved with plasma exchange alone, thereby averting a high-risk liver transplant.In addition, for patients who were subsequently transplanted, plasma exchange temporarily stabilized patients before liver transplant thereby allowing time to source for potential donors for liver transplant[24,25]. Nevertheless, most of the studies that included Wilson's disease are small case series, and the level of evidence remains weak.

Plasma exchange in ACLF

While there is strong evidence to use plasma exchange in ALF to improve survival[3],there has yet to be robust evidence to use plasma exchange in ACLF. Existing studies are mostly cohort studies done in Asia on patients with ACLF of predominantly hepatitis B viral etiology. Prognosis of patients with ACLF is extremely poor with mortality rates ranging from 30%-70%[2]in the absence of timely liver transplant. As there were very few studies in ACLF that only used plasma exchange solely, this review included all studies that used plasma exchange based-liver support systems in the management of ACLF.

A total of 15 studies of patients with ACLF were included (Tables 4, 5), of which 6 studies included patients with both ACLF and ALF (Table 5). 2 of the studies, which included both ACLF and ALF, were in pediatric patients. The rest of the 13 studies included adult patients only. Of the 13, 1 is a randomised controlled trial, 10 are cohort studies, 2 are case series.

Mortality:Plasma exchange with or without the use of other liver support systems improves survival in non-transplanted in patients with ACLF. An open-label randomized control study by Qin et al[37]recruited 234 patients with HBV-related ACLF not suitable for liver transplant and randomized patients to SMT vs plasma exchange centered ALSS plus SMT. In this study, survival rates in plasma exchangebased ALSS were significantly higher: 60% vs 47% in the control group. Other retrospective cohort studies[38-40]also favored plasma exchange (and or plasma exchange-based non-bioartificial liver support system) to SMT in patients with ACLF from hepatitis B infection. For example, In Yue-Meng et al[39], patients with ACLF who had plasma exchange+SMT had increased rate of survival compared to patients who had SMT only: 4-wk mortality was 82% vs 63%, P = 0.001; 12-wk mortality 86% vs 71%, P = 0.001).

Similarly, Mao et al[38]reported increased 30-day survival in patients with HBV related ACLF where survival rates were 41.9% and 25.2% for plasma exchange and medical therapy respectively (P ＜ 0.05). In the same study[38], time from the initial diagnosis to initiation of plasma exchange was found to be longer when in nonsurvivors compared to survivors, although this was not statistically significant.

Four of the studies[37-40]that reported plasma exchange based liver support system compared to standard medical treatment were analysed for pooled mortality at 30 and 90 d and data was presented in a forest plot (Figure 2). Other studies that had no comparative standard medical treatment arms were excluded from this analysis.Using available published in the aforementioned studies, plasma exchange was superior to SMT for survival in patients with ACLF, at 30-d (OR: 0.38, 95%CI: 0.38-0.88, P = 0.01, I2= 44%; 3 studies) and at 90-d (OR: 0.54, 95%CI: 0.35-0.84, P ≤ 0.01, I2=0; 2 studies). The pooled mortality at 30 and 90 d is significantly reduced in patients with ACLF who underwent plasma exchange (or plasma exchange based ALSS) vs SMT (OR: 0.60, 95%CI: 0.46-0.77, P ＜ 0.01, I2= 15%; 4 studies).

Several studies compared use of plasma exchange only vs plasma exchange together with other liver support system, such as the use of double plasma molecular adsorption system (DPMAS). For example, Yao et al[41]reported significantly higher 28-d survival rate in specifically intermediate-advanced stage patients with ACLF who underwent DPMAS + plasma exchange compared to plasma exchange alone(57.4% vs 41.7%, P = 0.043).

Biochemical improvement:Most studies reporting the effect of plasma exchangebased therapy on patients with ACLF which assessed biochemical improvement preand post-plasma exchange found an improvement in biochemical parameters such as coagulopathy, bilirubin, aspartate aminotransferase, alanine aminotransferase, or ammonia[12,39-44], though this was not always associated with clinical improvement.Zhou et al[45]reported a predictive model using baseline age, MELD score, number of complications and type of ALSS to predict survival after ALSS in patients with ACLF.In the same study, authors report that plasma bilirubin adsorption (PBA) + plasma exchange compared to plasma exchange only had better 90-day survival 70.3% vs 58.3%; although there was no mention if PBA + plasma exchange significantly decreased levels of bilirubin compared to plasma exchange alone. However when DPMAS vs plasma exchange was compared in other studies[41,44], increased clearance of bilirubin (as seen in the plasma exchange arm in Wan et al[44], and DPMAS + plasma exchange arm in Yao et al[41]) was not associated with respective significant improvements in survival. Of note, in the study by Zhou et al[45], baseline characteristics of the group of patients who underwent PBA+ plasma exchange vs plasma exchange was unavailable, essential information that could have influenced survival outcomes.

Figure 2 Forest plot for 30- and 90- d mortality in acute-on-chronic liver patients undergoing plasma exchange-based interventions or standard medical treatment. PE: Plasma exchange; SMT: Standard medical treatment.

Standard vs high volume and other liver assist devices:Several studies used solely plasma exchange in management of ACLF. For the studies that included plasma exchange in management of ACLF, a range of 2000-4500 mL of plasma exchange per session[12,37,38,41,43-45]was adopted. There were no studies in ACLF group that used high volume plasma exchange. However, there were more studies in ACLF compared to ALF that use other liver assist devices in conjunction with plasma exchange - for example, DPMAS, PBA, hemofiltration, hemodiafiltration, plasma diafiltration and its combinations. As there is no head to head trial and most studies are retrospective, it is not possible to draw any conclusion as to whether one modality was superior to another.

FFP vs albumin:Most studies used FFP for plasma exchange or plasma exchangebased ALSS[12,37-39,41-45]. In addition, Mao et al[38]used additional albumin during plasma exchange. Albumin dialysis was not compared with plasma exchange in this review.

Three-day therapy vs intermittent or response guided:All studies included for review extended plasma exchange beyond three days wherever relevant based on clinical necessity[12,37-39,41-45]. In addition, most studies do not use daily plasma exchange,and instead, this was performed 2-3 times per week and were response guided, where plasma exchange often was continued till clinical improvement, transplant, or death.

Etiology specific outcome:Of the 13 included studies for plasma exchange in ACLF in adult patients, all were being conducted in Asia where hepatitis B is endemic. Thus,the majority of the patients assessed have HBV related ACLF. In comparison to nonviral causes, ACLF in the presence of viral causes tends to have a poorer survival rate.For example, Cheng et al[12]reported a 24% survival in hepatitis B related ACLF vs 67%in alcohol-related ACLF in their retrospective cohort study of 45 ACLF and 10 ALF patients. Furthermore, where there were more than two causes for chronic liver injury e.g., HCV and alcohol, HBV and alcohol or in autoimmune hepatitis, mortality was high at 100%. However, this will need to be interpreted with caution as degree fibrosis or severity of cirrhosis of each patient was not available in the published study.

DISCUSSION

Acute liver failure and acute on chronic liver failure carry a high risk of mortality in patients in the absence of a liver transplant, a scarce resource. Liver assist devices,some of which are plasma exchange-based, have been used in patients with ALF or ACLF majority of which reported showing some benefit compared to standard medical treatment. However, there remains an unmet need for good quality prospective trials to be done, to ascertain the ideal volume, type, and duration of plasma exchange in management of ALF. Additional randomized controlled studies are also required to further shed light on the utility for plasma exchange or plasma exchange-based liver support systems for ACLF.

Firstly, while there has been good evidence for the use of high volume plasma exchange in acute liver failure to improve survival, due to the paucity of good-quality studies, at present it is unknown if a lower volume or a longer (or shorter) duration(i.e., beyond the first three consecutive days) of plasma exchange will achieve equal or improved survival in ALF. This is important since donor plasma is a finite resource,and HVP is not without side effects. For example as in the above case series by Freeman et al[13]where patients with ALF were treated with standard volume plasma exchange, overall survival was 55% which on surface appears comparable to HVP in Larsen et al[3]. Nevertheless, little conclusion can be drawn as case series often seem to show benefit while well-designed clinical randomized controlled trials may fail to fulfil the hopes of initial reports. In addition, as the baseline characteristics of patients could be different from these various studies, no conclusive verdict can be made until further high-quality studies are being carried out comparing high volume plasma exchange vs standard volume plasma exchange. There is also insufficient evidence to suggest if plasma exchange with albumin or a combination of albumin and fresh frozen plasma will be non-inferior to high volume plasma exchange in management of ALF.

Furthermore, whether further doses of plasma exchange will benefit a patient with ALF beyond the third exchange will be an important question to answer with future well-designed randomized controlled trials. This is especially important in donor scarce countries, where it may take more than a few days to work up a suitable liver donor for living donor liver transplant. In Buckner et al[7], it was reported that one patient who had ALF from halothane toxicity had frequent high volume plasma exchange until she awoke from a coma 37 d later. On the other hand, Chien et al[22]reported in a case series of 23 pediatric patients that plasma exchange for more than six times probably offers the little benefit with regard to patient survival in the absence of a timely liver transplant. Mechanistically, Kondrup et al[6]has reported that on theoretical assumptions, three courses of plasma exchange on consecutive days would reduce the concentration of “toxins” distributed in extracellular water to 18%of initial concentration at the end of last exchange and an additional course will only theoretically decrease this only to 16%. While this could stand true for patients whose cause of liver failure is from drugs such as paracetamol, the same may not be the case for patients who have liver failure from other causes like autoimmune or viral.

Secondly, there remains insufficient evidence to extrapolate the findings from Larsen et al[3]to recommend plasma exchange in patients with ACLF. There is no study to our best knowledge that used consecutive three days of high volume plasma exchange for the management of ACLF. Qin et al[37]has reported the use of plasma exchange based ALSS compared to SMT in the only prospective controlled trial to date in patient with HBV-ACLF - however this has not made it to standard practice.Of note, the definition of ACLF in this study follows the Chinese definition hence does not mandate the need for cirrhosis or multi-organ failure. Thus this study population is heterogenous, ranging from patients with no cirrhosis (52%) to patients requiring intensive care and renal replacement therapy. The timing of initiation and type of antiviral was also not standardized which adds limitations and potential bias to the final results. Nevertheless, it is worthwhile to note that the baseline characteristics of both treatment groups were similar and ALSS was found to have improved 90-day and 5-year mortality compared to SMT (60% vs 47%, P = 0.016; 43%vs 31%, P = 0.013). These results are promising and thus further randomised controlled trials should be done, ideally with stratification of patients according to different etiologies and grades of ACLF (or acute decompensation in cirrhotics) and assess if there are any differences in response to plasma exchange.

Of the 13 studies on use of plasma exchange based therapy in ACLF in adult patients, 4 studies compared plasma exchange based therapy with SMT. Of the four studies, 3 studies only included patients in HBV-associated ACLF. In the one study that included patients with other etiology of ACLF, HBV-related ACLF was predominant-91.24% of the ACLF study population. Using the aforementioned studies, plasma exchange based therapy in ACLF compared to SMT improved survival at 30- and 90-d with a pooled OR of 0.60 (95%CI: 0.46-0.77, P ＜ 0.01).However, there are limitations in that the number of studies used to generate the pooled OR estimate for mortality comparing plasma exchange vs SMT are small due to the limited number of studies reported in literature, hence the risk for bias of each study was not assessed. In addition, the definition of ACLF in these studies do not require the diagnosis of cirrhosis and or more than one organ failure. Moreover, the etiology of ACLF in these groups of patients is HBV related, and thus these findings cannot be extrapolated to ACLF caused by other etiologies.

Thirdly, it will be essential to find out if there is an objective measure of a point of no return whereby plasma exchange or plasma exchange -based ALSS will be futile,whether in ALF or ACLF. For example, Nakae et al[21]reported that while overall survival with plasma diafiltration was 54.5% in their study, there were no survivors in the MELD ＞ 40 groups when plasma diafiltration was used in ALF. Chen et al[42]also reported a trend where less patients with late-stage ACLF who underwent plasma exchange had transplant-free survival: 80.8% patients in the early stage, 75.8%patients in the middle stage and 37.4% patients in the end-stage survived for one month after diagnosis. Mechanistically, plasma exchange acts by removing plasma cytokines and adhesion molecules, which are the drivers of the systemic inflammatory cascade from the circulation and it is possible that late in the disease course of organ failure that the utility of plasma exchange would exponentially decrease.

Finally, other than using plasma exchange for management of ALF of ACLF, in our literature search there have been reports of other extracorporeal liver support devices could potentially improve outcomes in ALF or ACLF either alone or when used in conjunction with plasma exchange. Examples include molecular adsorbents recirculating system (MARS), Fractionated Plasma Separation, Adsorption and Dialysis device, SPAD. Several of our included studies used modalities like plasma exchange + DPMAS, plasma exchange + CVVHDF, or plasma diafiltration with varying results. Of the above liver devices, MARS has been studied most widely both in ALF and ACLF and its benefits remain modest. For example, a large randomized control trial did not demonstrate benefit in 6-mo survival in patients who had ALF and underwent MARS, although a major limitation of this study was that 75% of enrolled patients received a transplant within 24 h of enrollment, thereby potentially limiting the findings of the study[46]. This will remain to be a major limitation in future studies, especially in areas where donor livers can be quickly obtained. The emerging liver support devices, such as that of bioartificial liver support device which include a bioreactor that contains hepatocytes that can replace the function of the failing liver,e.g., the Extracorporeal liver assist device has entered human clinical trials. Other artificial liver support systems that combine detoxification with techniques to attenuate liver injury include the Hepa Wash, Li-Artificial Liver Support, and the University College London-Liver Dialysis Device, which has shown efficacy in animal experiments. More studies will need to be done to allow clinical use of these devices in liver failure.

In summary, the state of art therapy for acute liver failure should include plasma exchange based on the high-quality evidence, especially in patients who do not have a donor liver in sight. Current guidelines by EASL support use of HVP in ALF. While there is an open RCT supporting for plasma exchange-centered ALSS in the management of ACLF[37], and our pooled estimates from four studies favor the use of plasma exchange in ACLF as it is associated with a decreased 30 and 90 d mortality,this has not yet made it to standard practice.

The major limitation of our study is the low number of well-designed high-quality evidence available, as most studies are case series or cohort studies. This is possibly because liver failure is not common, and patients are usually critically ill. Secondly, as most of these studies are cohort studies or case series, we are unable to assess for publication bias via a funnel plot. However, publishing bias may exist as published studies are mostly positive studies and negative studies may not be reported. Lastly,the studies that have been included in plasma exchange in ACLF are over-represented by Asian patients, and the definitions of ACLF in each of these studies vary.Furthermore, it is imperative to take into account that the definition of ACLF in most Asian studies do not mandate the need for patients to be cirrhotic and have more than one organ failure, thereby potentially selecting a different group of patients from that seen in the Western ACLF literature. Whether the benefits of HVP in ALF can be extrapolated to patients with ACLF remains uncertain. Due to limited RCTs for ALF,only a subgroup meta-analysis (Figure 2) was explored comparing mortality among non-transplanted ACLF patients who underwent plasma exchange based therapy vs SMT.

There are many unanswered questions in this field: For example, the optimal type,duration, frequency, volume and time to plasma exchange one should use for ALF,and ACLF, if at all. Future randomized control trials studying the use of plasma exchange and or other liver assist devices in liver failure should aim to answer these questions. In addition, future studies should also include the study of biomarkers that can predict the success of therapy. This might further shed light on the optimal duration, volume of plasma exchange and or alternative liver support therapy since there is vast heterogeneity in patients with ALF and ACLF, and perhaps different groups of patients will require different regimes.

ARTICLE HIGHLIGHTS

Research background

Liver failure portends a high mortality without successful liver transplantation. High volume plasma exchange has been included in European guidelines as level I, grade 1 recommendation in management of acute liver failure possibly by removal of plasma cytokines and drivers of systemic inflammatory cascade through plasma exchange. In recent years, there is increasing interest in plasma exchange for the treatment of liver failure, as there is proven improvement in survival in those who do not undergo a liver transplant. Prior to this study, there were several other cohort studies reporting the benefits of plasma exchange in acute liver failure (ALF),however the volume and duration of plasma exchange varies. The evidence for use of high volume plasma (HVP) in acute-on-chronic liver (ACLF) is less robust, but the use of plasmapheresis (not high volume) has been reported in literature.

Research motivation

While there is good evidence to use plasmapheresis in management of acute liver failure especially when there is no liver donor in sight, the optimal volume and duration of plasma exchange is unclear. Donor plasma is a finite resource, and HVP is not without side effects such as hypocalcemia requiring rapid calcium replacement. Several cohort studies showed benefit in standard volume plasmapheresis in management of ALF however no head to head comparisons have been done. Furthermore use of plasma exchange in ACLF, while has been reported to improve survival in literature, has not been widely accepted as standard treatment due to lack of high level evidence.

Research objectives

This study aims to summarize and analyze the current literature for use of plasmapheresis in patients with ACLF and ALF and its effect on mortality particularly in the non-transplanted patients. In addition, the review will summarise the current literature on volume of plasma used during exchange, the duration and frequency of plasma exchange in both ALF and ACLF. It is our hope that this review will serve as a valuable resource by analyzing available literature as well as illustrate the knowledge gaps and unmet needs for future researchers in this field.

Research methods

This systematic review uses guidance from the PRISMA checklist. Databases MEDLINE via PubMed, and EMBASE were searched and relevant publications up to 30 March, 2019 were assessed. Forty-four studies were shortlisted and included in Tables 2-5. In addition, pooled odds ratios and its corresponding 95% confidence intervals were respectively calculated for 30-and 90-d mortality in ACLF patients using the random effects model. We were unable to do this for ALF group due to paucity of studies and lack of critical information from eligible studies.

Research results

There is good evidence for use of high volume plasma exchange in ALF though the optimal duration and volume of plasma exchange at present is uncertain. While high quality randomized control trials are lacking, the use of plasma exchange in ACLF can be considered. Survival in non-transplanted patients was improved in all four studies in patients with ACLF comparing plasma exchange vs standard medical therapy (SMT). Using the aforementioned studies, plasma exchange based therapy in ACLF compared to SMT improved survival in non-transplanted patients at 30 and 90-d with a pooled OR of 0.60 (95%CI: 0.46-0.77, P ＜ 0.01). There remains insufficient evidence to extrapolate the findings which recommend plasma exchange in patients with ACLF. Whether an individualized plasma exchange regime for each patient with liver failure can be personalised based on biomarkers remains unknown. More head to head trials will need to be done.

Research conclusions

While there has been good evidence for the use of high volume plasma exchange in acute liver failure to improve survival, due to the paucity of good-quality studies, at present it is unknown if a lower volume or a longer (or shorter) duration (i.e., beyond the first three consecutive days)of plasma exchange will achieve equal or improved survival in ALF. In patients with ACLF,plasma exchange based therapy compared to SMT improves survival at 30 and 90-d in nontransplanted patients. The duration of plasma exchange therapy used in most studies in ACLF was clinical response driven and often intermittent; and not with high volume plasmapheresis.The etiology of ACLF was also mostly HBV related; the definitions of ACLF used are varied and not requiring the diagnosis of more than one organ failure and diagnosis of cirrhosis. At present,there is insufficient evidence to extrapolate the use of high volume plasmapheresis to patients with ACLF.

Research perspectives

There are unanswered questions in use of plasma exchange in liver failure: For example, the optimal type, duration, frequency, volume and time to plasma exchange one should use for ALF,and ACLF, if at all. Future randomized control trials studying the use of plasma exchange and or other liver assist devices in liver failure should aim to answer these questions. It is also essential to find out if there is an objective measure of a point of no return whereby plasma exchange or plasma exchange-based ALSS will be futile, whether in ALF or ACLF. Subsequent clinical trials in ACLF or ALF should include study of biomarkers that can predict the success of therapy. This might further shed light on the optimal duration, volume of plasma exchange and or alternative liver support therapy since there is vast heterogeneity in patients with ALF and ACLF and one regime may not fit all. Finally, the definition of ACLF in each study needs to be clearly stated, in order to allow clinicians to assess applicability of study results to their patients.