Plasma exchange in ACLF: a real 28-day signal that does not survive to 90 days
A 194-patient randomised trial found that a strategy of five planned sessions of standard-volume therapeutic plasma exchange reduced 28-day mortality by 19.6 percentage points. The difference was no longer evident at 90 days, 42% did not complete all five sessions, and the cohort did not include patients with active infection, haemodynamic instability or respiratory organ failure.
20-second summary
The study at a glance
Bottom lineStandard-volume plasma exchange improved 28-day survival in a selected ACLF cohort, but the advantage was not sustained at 90 days.
- Population
- Adults aged 18–60 with EASL-CLIF ACLF grades 1–3b; patients with active infection, haemodynamic instability or respiratory organ failure were excluded.
- Effect
- Mortality was 44.3% with plasma exchange versus 63.9% with standard care at 28 days; by 90 days it was 66.0% versus 69.1%.
- Biggest limitation
- Single-centre, selected population with no transplantation; 42% of participants allocated to plasma exchange did not complete all five sessions.
- Practice implication
- This supports further evaluation as short-term organ support. It does not establish routine use or applicability to patients with shock, active infection or respiratory failure.
Why this study matters
Therapeutic plasma exchange has a persuasive rationale in ACLF: remove ammonia and circulating inflammatory mediators, replace albumin and coagulation factors, and potentially interrupt the progression of organ failure. The difficulty is that previous studies have been small and heterogeneous, with positive short-term findings that have not established a durable survival benefit.
This investigator-initiated randomised trial matters because it tests a standardised TPE strategy using all-cause mortality rather than relying mainly on biochemical improvement. It also studies a predominantly alcohol-related population defined using EASL-CLIF criteria, broadening an evidence base previously dominated by hepatitis B-related disease.
The key question is whether an early physiological effect becomes meaningful and sustained survival.
Study at a glance
- Design: Single-centre, open-label randomised controlled trial
- Participants: 194 adults with EASL-CLIF ACLF grades 1-3b
- Intervention: Five planned daily sessions of standard-volume TPE plus standard medical therapy
- Comparator: Standard medical therapy alone
- Primary outcome: All-cause mortality at 28 days
- 28-day mortality: 44.3% versus 63.9%; absolute risk reduction 19.6% (95% CI 5.8-33.3), p=0.006
- 90-day mortality: 66.0% versus 69.1%, p=0.646
How the study was conducted
Patients were randomised 1:1 with concealed allocation. Treatment could not be masked, although the statistical analysis was performed independently with allocation concealed from the analyst. The primary analysis was intention-to-treat.
Participants were aged 18-60 years and were randomised once active sepsis had been excluded. The study did not include patients with haemodynamic instability or respiratory organ failure. Alcohol was the commonest precipitant and underlying cause of liver disease, and grade 2 ACLF was the largest subgroup.
TPE was planned on five consecutive days using standard-volume exchange with fresh frozen plasma and albumin replacement. Standard therapy included treatment of the precipitating cause and management of organ failures.
The original sample-size calculation was revised during recruitment after publication of an external meta-analysis. The authors state that no interim analysis was prespecified.
- Study design
- Single-centre, open-label randomised controlled trial with concealed allocation and an intention-to-treat primary analysis.
- Population
- Adults aged 18-60 years with EASL-CLIF ACLF grades 1-3b. Active infection, haemodynamic instability and respiratory organ failure were not represented.
- Intervention or exposure
- Five planned consecutive daily sessions of standard-volume TPE using fresh frozen plasma and 5% albumin replacement.
- Comparator
- Standard medical therapy alone, including treatment of the precipitating cause and organ support.
What the study found
The primary result was substantial. At 28 days, mortality was 44.3% with TPE and 63.9% with standard therapy, an absolute difference of 19.6 percentage points. The early separation was not maintained: by day 90, mortality was 66.0% and 69.1%, respectively.
TPE was associated with greater ammonia reduction and faster resolution of hepatic and coagulation failure at day 7. These findings support a biological effect, although changes in bilirubin and INR are difficult to separate from the infusion of plasma and albumin during the procedure.
Treatment delivery was challenging. Only 56 of 97 participants allocated to TPE completed all five sessions, and haemodynamic instability was the commonest reason for stopping early. Safety outcomes, analysed as treated, showed more early infection, worsening oxygenation and catheter-related complications in the TPE group, although the study was not powered to define these risks precisely.
No participant underwent liver transplantation.
Interpreting the findings
Features that support interpretation
- Concealed random allocation and an intention-to-treat primary analysis
- A hard primary endpoint with absolute effects and confidence intervals reported
- A standardised TPE strategy that can be reproduced in future studies
- Contemporary EASL-CLIF definitions and a predominantly alcohol-related cohort
- Investigator-initiated research with no commercial funding or declared conflicts
Context and remaining uncertainty
The most important limitation is applicability. The participants were severely unwell, but the study excluded several common high-acuity phenotypes, particularly active infection, haemodynamic instability and respiratory organ failure. Its findings should therefore not be assumed to apply to patients with shock, uncontrolled sepsis or significant hypoxaemia.
The single-centre setting, age limit of 60 years and predominantly male, alcohol-related population further narrow generalisability. Five sessions were selected empirically, yet 42% did not complete the full course. Comparisons based on treatment completion are difficult to interpret because patients had to survive and remain stable enough to continue therapy.
The safety analysis was conducted as treated rather than by randomised allocation, and the physiological secondary outcomes involved multiple comparisons. These results are useful for understanding the signal but are less secure than the primary mortality comparison.
Finally, the absence of transplantation means the study cannot determine whether the early survival interval could function as a bridge to transplant. It does show that, without transplantation in this setting, the mortality advantage was not sustained at 90 days.
Relevance to liver critical care
This trial provides credible evidence that TPE can produce a meaningful short-term effect in a selected ACLF population. The mortality result is supported by improvements in ammonia and early organ-failure resolution, making it difficult to dismiss as a purely statistical finding.
The effect was temporary, however, and the intervention was difficult to complete. The study therefore fits more comfortably with TPE as short-term organ support than as established disease-modifying therapy.
One possible interpretation is that TPE buys time for recovery or for a definitive intervention. That remains a hypothesis: no patient in this trial underwent transplantation, and evidence in patients with shock, active infection or respiratory failure is still lacking.
Study details and effect estimates are drawn from the published article. This independent summary and commentary does not reproduce the article's figures or tables; readers should consult the original publication for the complete methods, results and safety data.
Questions that remain
- Can an early survival advantage translate into successful transplantation where access is available?
- Which patients are most likely to benefit from TPE?
- Would a response-adaptive number of sessions be preferable to a fixed course of five?
- What is the balance of benefit and harm in shock, infection and respiratory failure?
- Why did the early mortality difference largely disappear by day 90?
Full reference and source
Swaroop S, Biswas S, Aggarwal A, Coshic P, Kumar S, Agarwal S, et al. Therapeutic plasma exchange improves short-term survival in patients with acute-on-chronic liver failure: a randomized controlled trial. Hepatology. 2026;00:000-000. doi:10.1097/HEP.0000000000001755.
Community
Responses
Constructive professional discussion, alternative interpretations and relevant additional evidence are welcome. Responses are moderated before publication.
No published comments yet. You can start the discussion.