Skip to main content

Foundational Trials

ELAD

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Extracorporeal Cellular Therapy (ELAD) in Severe Alcoholic Hepatitis: A Multinational, Prospective, Controlled, Randomised Trial
  • Acronym: ELAD
  • Year: 2018
  • Journal published in: Liver Transplantation
  • Citation: Thompson J, Jones N, Al-Khafaji A, Malik S, Reich D, Munoz S, et al. Extracorporeal cellular therapy (ELAD) in severe alcoholic hepatitis: a multinational, prospective, controlled, randomised trial. Liver Transpl. 2018;24(3):380-393.

Context & Rationale

  • Background
    • Severe alcoholic hepatitis (sAH) is characterised by jaundice and coagulopathy with high short-term mortality despite supportive care.
    • Pharmacotherapy options historically used in sAH (e.g., corticosteroids; pentoxifylline) have limited, time-dependent, or inconsistent survival benefit and may be constrained by infection, bleeding risk, and contraindications.
    • Early liver transplantation can improve outcomes in selected non-responders, but applicability is limited by eligibility, organ availability, and system-level constraints.
    • Bioartificial liver support strategies aim to provide extracorporeal metabolic and immunomodulatory support to bridge patients to recovery or transplantation.
    • ELAD is a human hepatic cell–based extracorporeal system using C3A cells intended to augment liver recovery signals (including proteins implicated in the acute-phase response such as IL-1 receptor antagonist).
  • Research Question/Hypothesis
    • In hospitalised adults with sAH (bilirubin ≥8 mg/dL; Maddrey’s discriminant function ≥32; MELD 18–35), does adding 3–5 days of continuous ELAD therapy to guideline-based standard of care improve overall survival compared with standard of care alone?
  • Why This Matters
    • A positive trial would have established a scalable, non-transplant extracorporeal therapy for a high-mortality population with limited options.
    • For critical care clinicians, ELAD tested whether a high-intensity device intervention can translate biological signals into patient-centred benefit without unacceptable haemostatic and haemodynamic harms.
    • For trialists/methodologists, it provides a large contemporary example of device trials in acute liver inflammatory syndromes with complex co-interventions and substantial treatment non-completion.

Design & Methods

  • Research Question: In hospitalised adults with severe alcoholic hepatitis (bilirubin ≥8 mg/dL; Maddrey’s DF ≥32; MELD 18–35), does adding continuous extracorporeal cellular therapy (ELAD) to guideline-based standard of care improve overall survival versus standard of care alone?
  • Study Type: Multinational, prospective, multicentre (40 sites; United States/United Kingdom/Australia), randomised (1:1), open-label, stratified (Stratum A vs B), parallel-group, controlled superiority trial; minimum 91-day follow-up with a pre-planned extension for survival follow-up up to 5 years (VTI-208E); trial registration NCT01471028.
  • Population:
    • Setting: Hospitalised patients with sAH managed in specialist inpatient services (including higher-acuity care where needed) across participating centres.
    • Key inclusion: Adults ≥18 years; clinical sAH with jaundice (bilirubin ≥8 mg/dL) and coagulopathy (Maddrey’s DF ≥32); MELD 18–35; informed consent obtained.
    • Diagnostic strata: Stratum A (alcoholic liver disease with sAH) required biopsy-proven sAH or ≥2 of hepatomegaly, AST>ALT, leukocytosis, ascites; Stratum B included sAH with underlying chronic liver disease not due to alcohol (documented by biopsy/labs/history).
    • Key exclusions: End-stage cirrhosis features (liver size <10 cm, liver volume <750 cc, or portal vein thrombosis); MELD >35; platelets <40,000/mm3; uncontrolled bleeding; infection clinically unresponsive to antibiotics; haemodynamic instability; chronic dialysis; evidence of bilirubin reduction ≥20% in the prior 72 hours; significant concomitant disease with life expectancy <3 months.
  • Intervention:
    • ELAD + standard of care: Blood was drawn via a dual-lumen catheter into an extracorporeal pumping unit, separated using an ultrafiltrate generator cartridge, and the ultrafiltrate circulated at high flow through 4 hollow-fibre cartridges containing ~440 g of C3A cells; ultrafiltrate then passed through a 0.2 μm cell filter, recombined with cellular components, and returned to the patient.
    • Dose/timing: Continuous ELAD therapy for up to 120 hours (3–5 days); protocol permitted stopping after 72 hours if rapid response, and discontinuation for futility/deterioration/withdrawal/technical or clinical reasons.
    • Anticoagulation: Site protocols for continuous haemofiltration were used, except citrate anticoagulation was not permitted.
  • Comparison:
    • Standard of care alone: Sites were instructed to apply AASLD/EASL guideline-based management for complications of liver disease; concomitant sAH medications were permitted and documented (e.g., corticosteroids, pentoxifylline, N-acetylcysteine, antibiotics) with usual supportive care measures (nutrition, complication management).
    • No sham device: Control patients did not undergo catheter placement or simulated extracorporeal support.
  • Blinding: Open-label; primary endpoint (mortality) is objective, but non-fatal outcomes (adverse events; decisions to discontinue therapy) were potentially susceptible to performance/detection bias; follow-up health assessments were performed by an independent physician/nurse practitioner applying standard of care.
  • Statistics: A total of 200 subjects (100 per group) were required to detect a 20% absolute increase in 91-day survival (from 40% to 60%) with ≥95% power at the 5% significance level using a log-rank test; primary analysis was intention-to-treat with Kaplan–Meier/log-rank, with Cox modelling for hazard ratios; safety analyses were performed “as treated”.
  • Follow-Up Period: Minimum follow-up ≥91 days after randomisation of the last subject; scheduled assessments daily for week 1 (if not discharged), then days 7, 14, 21, 28, 63, and 91; extension protocol (VTI-208E) collected survival data up to a maximum of 5 years.

Key Results

This trial was not stopped early. Recruitment reached 203 randomised participants (planned 200); no interim efficacy stopping was reported.

Outcome ELAD + SOC SOC Effect p value / 95% CI Notes
Overall survival (primary; time-to-event; minimum follow-up ≥91 days) Death rate 47.9% Death rate 47.7% HR 1.03 95% CI 0.69 to 1.53; P=0.90 Kaplan–Meier/log-rank in the intention-to-treat population.
Survival proportion at day 28 (secondary) 76.0% 80.4% Not reported Not reported Secondary endpoint; between-group statistical comparison not reported.
Survival proportion at day 91 (secondary) 59.4% 61.7% Not reported Not reported Secondary endpoint; between-group statistical comparison not reported.
Prespecified subgroup: baseline MELD <28 (mortality at database lock) 29.4% 43.5% HR 0.58 P=0.08; 95% CI not reported Trend towards lower mortality with ELAD in lower MELD subgroup (n=120).
Prespecified subgroup: baseline MELD ≥28 (mortality at database lock) 68.9% 55.3% HR 1.50 P=0.15; 95% CI not reported Directionally worse outcomes with ELAD in higher MELD subgroup (n=83).
Prespecified subgroup: age <46.9 years (mortality at database lock) 32.6% 44.8% HR 0.63 P=0.17; 95% CI not reported Trend towards lower mortality with ELAD in younger subgroup (n=101).
Prespecified subgroup: age ≥46.9 years (mortality at database lock) 60.4% 51.0% HR 1.35 P=0.26; 95% CI not reported Directionally worse outcomes with ELAD in older subgroup (n=102).
Post-hoc combined subgroup: MELD <28 and age <46.9 years (day-91 survival) 100% 73% Not reported P=0.006 Not prespecified as a combined subgroup comparison; interpret as hypothesis-generating.
Biomarker response: bilirubin reduction >20% by day 7 (safety population) 59% (56/95) 23% (25/108) Not reported Not reported ELAD produced a consistent biochemical signal during days 2–7.
Lille response at day 7 (Lille <0.45) 74.7% 47.0% Not reported P<0.01 Higher early “response” signal did not translate into overall survival benefit.
TESAEs (as-treated safety population) 76.8% 69.4% Not reported Not reported Two deaths judged related to ELAD: DIC precipitation (baseline MELD >28) and catheter insertion complication.
TEAEs of interest (as-treated safety population) Anaemia 44%; thrombocytopenia 35%; coagulopathy 31%; hypotension 31% Anaemia 16%; thrombocytopenia 11%; coagulopathy 12%; hypotension 17% Not reported Not reported Nonserious adverse events were increased with ELAD, predominantly haematological and haemodynamic.
  • Despite clear biomarker shifts (bilirubin reduction; higher Lille response), ELAD did not improve overall survival (HR 1.03; 95% CI 0.69 to 1.53; P=0.90).
  • Prespecified median-split subgroup trends suggested possible benefit in younger and lower-MELD patients, with directionally worse outcomes in older and higher-MELD patients.
  • ELAD exposure was frequently truncated (only 45/96 completed 120 hours), and nonserious adverse events were more common with ELAD, highlighting the procedure–benefit trade-off in extracorporeal support.

Internal Validity

  • Randomisation and allocation:
    • Block randomisation (block size 4) with 1:1 allocation; stratified schedules for Stratum A and B; allocation concealment via computerised eCRF and sponsor hotline code assignment.
    • Uneven group sizes (96 vs 107) occurred because some sites did not complete full randomisation blocks.
  • Dropout/exclusions after randomisation:
    • At 91 days: 1 lost to follow-up in each arm; 2 withdrew consent in the control arm; outcome known in 95 ELAD vs 104 control subjects.
    • Extension follow-up (as of July 2015): outcome known in 94 ELAD vs 102 control; 1 ELAD and 3 control lost to follow-up; 1 ELAD and 2 control withdrew consent.
  • Performance/detection bias:
    • Open-label design without sham extracorporeal support increases risk of differential co-interventions and reporting of non-fatal outcomes.
    • Primary endpoint (mortality) is objective and less susceptible to adjudication bias than subjective outcomes.
  • Protocol adherence and treatment exposure:
    • Of 96 randomised to ELAD: 45 completed 120 hours; 37 completed 72–120 hours; 12 received <72 hours; 2 did not initiate due to instability.
    • Fifty ELAD subjects did not complete 120 hours, commonly due to adverse events (n=18) or investigator judgement of instability (n=9); 7 discontinuations lacked documented reasons.
    • Safety population differed from ITT: 2 randomised to ELAD did not receive ELAD; 1 control subject inadvertently received ELAD (as-treated: 95 ELAD; 108 control).
  • Baseline characteristics and illness severity:
    • Mean MELD 27.3; mean bilirubin 25.1 mg/dL; vasopressor therapy required in 38 subjects with even distribution between arms.
    • Ventilator support at baseline: 8 ELAD vs 3 control; acute dialysis at baseline: 2 ELAD vs 0 control.
    • Control arm had a higher proportion of patients aged <35 years.
  • Timing:
    • Time between last alcohol intake, hospitalisation, and randomisation was comparable between groups.
  • Dose:
    • Protocol-specified maximum ELAD exposure was 120 hours, but only 46.9% (45/96) received the full intended duration.
    • Early termination was frequent and plausibly related to the haemodynamic/haemostatic burden of extracorporeal therapy in advanced liver failure.
  • Separation of the variable of interest:
    • Biochemical response separation was present: bilirubin reduction >20% by day 7 occurred in 59% (56/95) with ELAD vs 23% (25/108) with SOC.
    • Lille responders at day 7: 74.7% with ELAD vs 47% with SOC (P<0.01).
    • Alkaline phosphatase: 188 ± 110 to 152 ± 81 U/L with ELAD vs 173 ± 83 to 178 ± 100 U/L with SOC over 7 days.
  • Adjunctive therapies and co-interventions:
    • Prednisone standard dose (>25 mg for 7 days): 34.4% (33/96) in ELAD vs 42.1% (45/107) in control.
    • Systemic antibiotics for infection at baseline: 60.4% (58/96) ELAD vs 54.2% (58/107) control.
    • Pentoxifylline at baseline: 34.4% (33/96) ELAD vs 31.8% (34/107) control.
  • Outcome assessment:
    • Overall survival assessed via Kaplan–Meier/log-rank in ITT; extended survival follow-up collected via the extension protocol.
    • Adverse events were collected and serious events reviewed by an independent DSMB; ELAD-relatedness was investigator-assessed.
  • Statistical rigor:
    • Primary analysis followed a prespecified statistical plan; ITT used for efficacy and as-treated for safety.
    • Multiple subgroup and biomarker analyses were performed, with subgroup results reported as trends alongside P values.

Conclusion on Internal Validity: Overall, internal validity is moderate: randomisation with allocation concealment and an objective mortality endpoint are strengths, but the open-label design and substantial truncation/non-initiation of ELAD exposure meaningfully limit separation and complicate causal attribution for both efficacy and harms.

External Validity

  • Population representativeness:
    • Participants reflect hospitalised sAH with very high bilirubin (mean 25.1 mg/dL) and MELD in a clinically severe range (18–35; mean 27.3) across multiple countries and centres.
    • Generalisation is constrained by key exclusions (MELD >35; chronic dialysis; severe thrombocytopenia; haemodynamic instability; uncontrolled infection; end-stage cirrhosis by imaging; rapidly improving bilirubin).
    • Inability to provide informed consent was a common screening failure (39/171 exclusions), potentially selecting against more encephalopathic or unstable ICU patients.
  • Applicability:
    • ELAD requires specialised equipment, trained staff, continuous extracorporeal circuit management, and reliable vascular access, limiting feasibility to high-resource settings.
    • Because ELAD does not provide toxin removal via dialysis/filtration and provides no renal support, applicability is limited in populations with prominent kidney failure or multi-organ failure patterns.
    • The enrolled population included some high-acuity features (vasopressors in 38 subjects; ventilation in 11 subjects; acute dialysis in 2 subjects), but many ICU-level extremes were excluded, narrowing the “critical care” envelope.

Conclusion on External Validity: Generalisability is limited-to-moderate: results are most applicable to hospitalised sAH patients meeting the trial’s physiological and imaging constraints in specialised centres; they are less applicable to the sickest ICU cohorts with profound coagulopathy, uncontrolled infection, or established renal failure.

Strengths & Limitations

  • Strengths:
    • Large contemporary multicentre, multinational randomised controlled evaluation of a complex device intervention in sAH.
    • Objective primary endpoint (mortality) with minimum 91-day follow-up and a pre-planned extension for longer-term survival ascertainment.
    • Detailed reporting of treatment exposure, discontinuation reasons, and safety signals specific to extracorporeal therapies.
    • Demonstrated mechanistic/biological separation (bilirubin reduction, Lille response, IL1Ra signal) enabling refined hypothesis generation.
  • Limitations:
    • Open-label design without sham procedure introduces risk of differential management and subjective outcome reporting (particularly adverse events and discontinuation decisions).
    • Substantial non-completion of the intended 120-hour ELAD “dose” (only 45/96 completed), with discontinuations often related to adverse events/instability.
    • Selection constraints (e.g., consent capacity; exclusion of rapidly improving bilirubin and end-stage cirrhosis features) may limit applicability to the broader ICU sAH population.
    • Safety profile includes increased haematological and haemodynamic adverse events and two deaths judged related to ELAD treatment.

Interpretation & Why It Matters

  • Clinical meaning
    • ELAD did not improve survival in an unselected sAH population meeting trial criteria, and its adverse event profile was non-trivial; routine clinical use is not supported by this trial.
    • The discordance between biomarker improvements (bilirubin; Lille response; IL1Ra signal) and neutral survival underscores that biochemical surrogate improvement is insufficient to infer patient-centred benefit in extracorporeal organ support.
    • Signals suggesting differential effect by age and MELD highlight the central importance of patient selection and harm–benefit balance for invasive device strategies in acute liver inflammatory syndromes.

Controversies & Subsequent Evidence

  • Neutral primary endpoint despite biochemical signals:
    • Contemporaneous editorial critique emphasised that improved early surrogate signals (bilirubin dynamics; Lille response) did not translate into improved 91-day survival, limiting the clinical case for ELAD as implemented.1
  • Subgroup signal versus multiplicity and mechanistic plausibility:
    • The editorial highlighted opposing directional trends across prespecified median-split subgroups (suggested benefit in younger/lower MELD; worse outcomes in older/higher MELD), raising the possibility that extracorporeal therapy harms can offset benefit as physiological reserve decreases.1
  • Broader extracorporeal liver support evidence base:
    • A systematic review/meta-analysis of artificial and bioartificial liver support systems in acute liver failure populations reported heterogeneous trials with uncertain survival benefit, reinforcing that extracorporeal liver support remains investigational and highly context-dependent.2
  • Guideline position after ELAD:
    • Modern guidance for alcohol-associated liver disease prioritises optimal supportive care, complication management, infection evaluation, and selective corticosteroid use; it does not recommend bioartificial liver support devices outside clinical trials.34

Summary

  • Multinational open-label randomised trial (40 sites) enrolled 203 hospitalised adults with sAH (bilirubin ≥8 mg/dL; Maddrey’s DF ≥32; MELD 18–35) to ELAD + SOC (n=96) versus SOC (n=107).
  • No survival benefit was demonstrated: overall survival HR 1.03; 95% CI 0.69 to 1.53; P=0.90, with day-91 survival proportions 59.4% (ELAD) vs 61.7% (SOC).
  • Prespecified subgroup trends suggested possible benefit in younger/lower-MELD patients and worse outcomes in older/higher-MELD patients; a post-hoc combined subgroup (age <46.9 and MELD <28) showed day-91 survival 100% vs 73% (P=0.006).
  • ELAD produced clear biochemical separation (bilirubin reduction >20% by day 7: 59% vs 23; Lille responders 74.7% vs 47%) but increased nonserious adverse events (anaemia, thrombocytopenia, coagulopathy, hypotension) and two deaths judged related to ELAD.
  • The trial illustrates how invasive extracorporeal support can generate plausible mechanistic signals yet fail to improve patient-centred outcomes when treatment exposure is incomplete and procedure-related harms are non-negligible.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Where a volume/issue/page range was not available in the accessible source material, it is listed as “Not reported” to avoid introducing unverifiable bibliographic detail.

Overall Takeaway

ELAD is a landmark negative device trial in severe alcoholic hepatitis: despite mechanistic plausibility and demonstrable biochemical effects, a short course of extracorporeal cellular therapy did not improve survival compared with standard care. Its prespecified subgroup trends and procedure-related harms sharpen modern thinking about extracorporeal liver support as a patient-selection problem in which physiologic reserve, coagulopathy, and renal dysfunction may dominate the risk–benefit balance.

Overall Summary

  • ELAD + SOC did not improve overall survival compared with SOC alone in severe alcoholic hepatitis.
  • ELAD produced consistent biochemical response signals but increased haematological/haemodynamic adverse events and included two device-related deaths.
  • Subgroup trends (younger/lower MELD) remain hypothesis-generating and emphasise the centrality of patient selection for high-intensity extracorporeal therapies.

Bibliography