Skip to main content

Foundational Trials

HELIOS

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effects of Fractionated Plasma Separation and Adsorption on Survival in Patients With Acute-on-Chronic Liver Failure
  • Acronym: HELIOS
  • Year: 2012
  • Journal published in: Gastroenterology
  • Citation: Kribben A, Gerken G, Haag S, Herget-Rosenthal S, Treichel U, Betz C, et al; HELIOS Study Group. Effects of fractionated plasma separation and adsorption on survival in patients with acute-on-chronic liver failure. Gastroenterology. 2012;142(4):782-789.

Context & Rationale

  • Background
    • Acute deterioration in chronic liver disease (then termed AOCLF/ACLF) is associated with high short-term mortality, frequently in the setting of infection, bleeding, hepatic encephalopathy, and multi-organ failure.
    • Detoxification failure includes accumulation of protein-bound toxins (e.g., bilirubin/bile acids) that are poorly cleared by conventional renal replacement techniques.
    • Artificial liver support systems (albumin dialysis / plasma separation-based adsorption) were biologically plausible and had shown improvements in biochemical and clinical surrogates in early studies, but hard outcome evidence (survival) remained uncertain.
    • Prometheus fractionated plasma separation and adsorption (FPSA) was designed to remove albumin-bound and water-soluble toxins by “regenerating” albumin in circuit, without plasma removal.
  • Research Question/Hypothesis
    • In adults with severe acute deterioration of chronic liver disease meeting predefined bilirubin and Child–Pugh thresholds, does adding Prometheus FPSA (scheduled 8–11 sessions over 3 weeks) to standard medical therapy improve survival at 28 and 90 days versus standard medical therapy alone?
  • Why This Matters
    • FPSA is resource-intensive and requires specialised staff/equipment; demonstration of survival benefit is essential before widespread adoption.
    • If effective, extracorporeal detoxification could act as a bridge to liver transplantation or to spontaneous recovery by limiting toxin-driven organ dysfunction.
    • Negative survival results would argue that biochemical detoxification alone is insufficient in an organ-failure syndrome dominated by dysregulated inflammation and infection.

Design & Methods

  • Research Question: Does Prometheus FPSA plus standard medical therapy improve 28- and 90-day survival (irrespective of liver transplant) compared with standard medical therapy alone in adults with severe acute deterioration of chronic liver disease?
  • Study Type: Prospective, randomised, controlled, parallel-group, multicentre trial conducted in 10 university hospitals across 7 European countries (June 2005–January 2008); centralised fax randomisation (1:1) stratified by centre and presence of hepatorenal syndrome; open-label follow-up to 90 days; trial registration ISRCTN50777531.
  • Population:
    • Key inclusion: age 18–70 years; severe deterioration of chronic liver disease; Child–Pugh score ≥10 sustained over 72 hours; serum bilirubin ≥5 mg/dL sustained over 72 hours; written informed consent.
    • Key exclusions (selected): pregnancy/lactation; HIV infection; intracranial bleeding/cerebrovascular infarction or haemorrhage; acute respiratory distress syndrome; circulatory shock with vasopressor therapy ≥0.5 μg·kg−1·min−1; persistent bleeding requiring ≥4 units packed red blood cells in 24 hours pre-enrolment; chronic renal failure stage V (GFR ≤15 mL/min or dialysis ≥3 months); acute necrotising pancreatitis; hepatocellular carcinoma beyond Milan criteria or extrahepatic malignancy; INR ≥3.0 or platelet count <30,000/μL; extrahepatic cholestasis; major hepatobiliary surgery within 6 months (except laparoscopic cholecystectomy); transplant recipients except those with liver transplant ≥2 years prior; other liver support/plasma exchange within 7 days; participation in another clinical trial or prior participation in this study.
    • Clinical phenotype at baseline (contextual): hepatic encephalopathy any grade 78% vs 82%; acute renal failure 55% vs 56% (Type 1 HRS 35% vs 26%; Type 2 HRS 16% vs 19%).
  • Intervention:
    • FPSA platform: Prometheus system delivering fractionated plasma separation and adsorption, added to standard medical therapy.
    • Schedule: 5 treatments in week 1, 3 treatments in week 2 (8 total), with 3 additional treatments in week 3 if after 8 treatments bilirubin remained ≥5 mg/dL and Child–Pugh score had not decreased by ≥2 points to <10.
    • Session target: intended ≥4 hours per treatment (excluding interruptions); blood sampling in FPSA arm drawn ≥2 days after last FPSA session to assess sustained effects.
  • Comparison:
    • Standard medical therapy (SMT) alone using a standardised manual aligned to contemporary guideline-based care for complications of cirrhosis.
    • Key SMT elements (examples): variceal bleeding treated with terlipressin or somatostatin plus band ligation; ascites managed with sodium restriction and large-volume paracentesis plus albumin; hepatorenal syndrome treated with vasoconstrictors plus albumin then renal replacement therapy if indicated; encephalopathy treated with lactulose/lactitol and enemas; spontaneous bacterial peritonitis treated with third-generation cephalosporins.
    • Protocol fidelity: adherence to SMT was confirmed pre-analysis by a study-independent medical expert blinded to group allocation.
  • Blinding: Open-label treatment allocation; primary outcome (mortality) objective; SMT adherence assessment performed blinded by an independent expert.
  • Statistics: Power calculation targeted a 20% absolute increase in 90-day survival (from 50% to 70%) with 80% power at a one-sided 2.5% significance level, requiring 204 patients (102 per arm) accounting for one interim analysis; confirmatory intention-to-treat analysis; multiplicity handled by hierarchical testing (day-28 survival first, day-90 survival second); Kaplan–Meier point estimates at days 28 and 90 compared using a t-test approach; overall survival compared by log-rank; prespecified subgroup analyses (HRS, aetiology, MELD severity categories).
  • Follow-Up Period: 90 days (or until death), irrespective of ICU/hospital discharge or liver transplantation.

Key Results

This trial was stopped early. Recruitment was terminated after the planned interim analysis (performed once 90 patients had completed follow-up) because the prespecified futility criterion was met; 145/204 planned patients were randomised (FPSA+SMT n=77; SMT n=68).

Outcome FPSA + SMT SMT Effect p value / 95% CI Notes
Survival probability (Day 28; ITT; primary) 66% 63% Not reported P=0.70 Kaplan–Meier point estimate; first in hierarchical primary endpoint sequence.
Survival probability (Day 90; ITT; second primary) 47% 38% Not reported P=0.35 Kaplan–Meier point estimate; formally subordinate to day-28 survival in hierarchical testing.
Overall survival (median time to death; ITT) 79 days 59 days Not reported Log-rank P=0.3872 Time-to-event comparison of survival curves.
Per-protocol survival probability (Day 28) 71% 67% Not reported Not reported PP analysis: FPSA n=55; SMT n=54; descriptive stability check.
Per-protocol survival probability (Day 90) 41% 39% Not reported Not reported PP analysis; descriptive only.
Survival probability with transplant censored (Day 28) 64% 64% Not reported Not reported Secondary survival analysis censoring at transplant date.
Survival probability with transplant censored (Day 90) 46% 38% Not reported Not reported Secondary survival analysis censoring at transplant date.
Prespecified subgroup: MELD >30 (n=48; 24 per arm) — survival probability (Day 90) 48% 9% Adjusted HR 0.47 95% CI 0.22 to 0.99; log-rank P=0.0241 Day-28 survival 57% vs 42%; subgroup signal persisted after adjustment for baseline prognostic factors.
Prespecified subgroup: Type 1 HRS (n=45) — survival probability (Day 90) 42% 6% Not reported Log-rank P=0.04 (univariate); adjusted P=0.33 Day-28 survival 62% vs 39%; attenuation after adjustment limits inference.
Serum bilirubin (change baseline to Day 28) −8 ± 11 mg/dL 0 ± 9 mg/dL Not reported P=0.0001 Biochemical “detoxification” separation achieved; sampling ≥2 days after last FPSA session.
Renal replacement therapy overall (any time during study) 21/77 (27%) 18/68 (27%) Not reported P=1.0 Renal support burden similar; new-onset acute renal failure 21% vs 24% (P=0.93).
Bleeding from any source (patients with ≥1 event; safety) 33/77 (43%) 33/68 (49%) Not reported P=0.79 No excess bleeding signal attributable to extracorporeal treatment.
Bacterial infections (patients with ≥1 event; safety) 48/77 (62%) 37/68 (54%) Not reported P=0.63 High infection burden in both groups; infections and multi-organ failure were common causes of death.
  • Despite marked biochemical separation (bilirubin change −8 ± 11 vs 0 ± 9 mg/dL; P=0.0001), there was no statistically significant improvement in survival at day 28 (66% vs 63%; P=0.70) or day 90 (47% vs 38%; P=0.35).
  • The observed absolute survival difference at day 90 (~9%) was far smaller than the prespecified 20% target used for power; early stopping for futility left the trial underpowered for modest but clinically important effects.
  • Prespecified subgroup signals (MELD >30; Type 1 HRS) were numerically large but require cautious interpretation given multiplicity and limited sample size.

Internal Validity

  • Randomisation and Allocation: Centralised fax randomisation (1:1), stratified by centre and baseline hepatorenal syndrome; allocation sequence concealment is supported by central randomisation workflow.
  • Drop out / exclusions (post-randomisation):
    • Allocated to FPSA+SMT: 5/77 (6.5%) did not receive FPSA within 4 days (death n=2 on days 1–2; transplant n=1 day 1; withdrawal of consent n=1 day 0; variceal bleeding/drop-out n=1 day 4).
    • Unknown outcome: FPSA+SMT 7/77 vs SMT 2/68 (withdrawal of consent 2 vs 1; loss to follow-up 5 vs 1).
    • Per-protocol population: FPSA+SMT 55/77 (excluded n=22) vs SMT 54/68 (excluded n=14); exclusions included short study duration (<4 days) and protocol violations (including SMT documentation violations and FPSA treatment block violations).
  • Performance/Detection Bias: Open-label delivery creates potential for co-intervention imbalance; however, primary endpoint mortality is objective and SMT adherence was independently verified in a blinded manner.
  • Protocol Adherence / Dose delivery:
    • 585 FPSA sessions were delivered to 72 treated patients (mean 8.1 sessions per patient) with mean effective duration 5.7 ± 1.3 hours.
    • 47 treatments (8%) ended prematurely (<4 hours), most commonly due to circuit clotting (39 cases among causes; more than one cause possible per treatment).
    • 38/72 treated patients (52.8%) received the maximum 11 sessions; most survivors beyond 2 weeks still met criteria for the third treatment block.
    • Anticoagulation and circuit performance: no visible clotting achieved in 489 sessions (84%); most sessions used low-dose citrate (60%), with citrate plus low-dose heparin in 31%, and heparin alone in 8%.
  • Baseline Characteristics:
    • Severity was well balanced: Child–Pugh 12 ± 1 vs 12 ± 1; MELD 28 ± 10 vs 27 ± 10; SOFA 10 ± 3 vs 10 ± 4.
    • Renal phenotype similar: renal failure 55% vs 56%; Type 1 HRS 35% vs 26%; Type 2 HRS 16% vs 19%.
    • Potential imbalance: alcoholic aetiology 48% vs 65% (P=0.07) and “other aetiologies” 22% vs 12% (P=0.07 overall across aetiology categories).
  • Heterogeneity: Broad entry criteria (bilirubin/Child–Pugh thresholds) captured a heterogeneous AOCLF population with multiple precipitants and aetiologies; such heterogeneity can dilute an intervention effect if only a subset is biologically responsive.
  • Timing:
    • Median time from precipitating event to randomisation/start: 10 days (FPSA+SMT) vs 13 days (SMT) (P=0.56).
    • Median time from hospital admission to randomisation: 7 vs 9 days (P=0.08).
  • Separation of the Variable of Interest:
    • Detoxification separation was demonstrated biochemically (bilirubin change baseline→day 28: −8 ± 11 vs 0 ± 9 mg/dL; P=0.0001).
    • Other key lab separations were limited (e.g., creatinine change −0.2 ± 1.7 vs −0.1 ± 1.7 mg/dL; P=0.60; sodium change −2.5 ± 6.5 vs 0.7 ± 6.7 mEq/L; P=0.08).
    • Organ support signals were not clearly separated (renal replacement therapy overall 27% vs 27%).
  • Outcome Assessment: Mortality is robust; adverse events were prospectively collected over 90 days and coded using MedDRA, with cause-of-death attribution allowed to include multiple adverse events as potential causes.
  • Statistical Rigor: Prespecified hierarchical primary endpoints and interim analysis rules were adhered to; early stopping for futility limits precision and power for smaller-than-expected treatment effects.

Conclusion on Internal Validity: Moderate. Randomisation and objective primary outcome support credibility, and the intervention achieved biochemical separation; however, open-label delivery, early termination for futility, post-randomisation exclusions in per-protocol analyses, and outcome missingness (unknown outcome 7 vs 2) reduce confidence in detecting modest but clinically meaningful survival effects.

External Validity

  • Population Representativeness: Hospitalised, high-severity decompensated cirrhosis patients in tertiary European university hospitals; high prevalence of encephalopathy (≈80%) and renal failure (≈55%) approximates an ICU-relevant cirrhosis phenotype.
  • Important Exclusions: Patients with shock requiring high-dose vasopressors, ARDS, intracranial bleeding/cerebrovascular events, severe coagulopathy (INR ≥3 or platelets <30,000/μL), active major bleeding, and advanced malignancy were excluded; these are common in real-world ACLF/ICU practice and limit applicability to the sickest cohort.
  • Applicability: Findings translate best to specialist centres with access to Prometheus technology and experience managing complex cirrhosis complications; generalisability to resource-limited settings is low due to infrastructure requirements.
  • Clinical workflow relevance: Median delays from precipitating event and admission to enrolment reflect referral and stabilisation pathways typical of tertiary practice, but may not generalise to earlier intervention models.

Conclusion on External Validity: Moderate. The cohort resembles many tertiary ICU cirrhosis admissions but excludes haemodynamically and respiratory unstable patients; the findings are most applicable to selected AOCLF/ACLF patients stable enough for extracorporeal therapy in high-resource centres.

Strengths & Limitations

  • Strengths:
    • Largest randomised therapeutic trial of Prometheus FPSA in AOCLF at the time, conducted across multiple European academic centres.
    • Central stratified randomisation (centre; hepatorenal syndrome) and objective primary endpoint (mortality) limit key biases.
    • SMT was protocolised and adherence was confirmed by an independent blinded reviewer, supporting fairness of the comparator.
    • Clear intervention separation at a mechanistic level (substantial bilirubin reduction sustained beyond the last treatment session).
  • Limitations:
    • Stopped early for futility with 145/204 planned enrolment, limiting power and precision for modest survival benefits.
    • Open-label design creates risk of performance bias and co-intervention imbalance, despite protocolisation.
    • Entry criteria likely captured a heterogeneous syndrome (including end-stage decompensation) potentially less amenable to “bridge” detoxification therapy.
    • High burden of complications and organ failure with limited evidence of separation in organ support endpoints (e.g., renal replacement therapy identical at 27%).
    • Industry support (Fresenius Medical Care) and an author employed by the manufacturer introduce potential perception of conflict, though hard outcomes and multicentre conduct mitigate some concerns.

Interpretation & Why It Matters

  • Detoxification ≠ survival
    Bilirubin fell substantially with FPSA (−8 ± 11 mg/dL) yet survival was not improved (day 28: 66% vs 63%; day 90: 47% vs 38%), indicating that biochemical clearance alone was insufficient to change hard outcomes in an unselected AOCLF cohort.
  • Signals in severe subgroups
    Prespecified analyses suggested benefit in the sickest patients (MELD >30: 90-day survival 48% vs 9%; adjusted HR 0.47; 95% CI 0.22 to 0.99), but these findings are hypothesis-generating rather than practice-changing.
  • Practical implication
    HELIOS supports FPSA feasibility and relative safety in selected patients but does not support routine use for survival benefit; use, if considered, is best framed as bridge/adjunct in highly selected patients and within protocols that target reversible phenotypes and earlier delivery.

Controversies & Subsequent Evidence

  • Syndrome definition and enrolment heterogeneity: The accompanying editorial argued that inclusion criteria risked enrolling end-stage decompensated cirrhosis rather than a reversible ACLF phenotype, reducing the likelihood of demonstrating benefit; subsequent prospective European cohort work (CANONIC) formalised ACLF definitions and highlighted organ failure/inflammation as central features. 13
  • Timing of intervention: Delay from precipitating event to initiation (median 10–13 days) was highlighted as a plausible explanation for lack of survival effect in tertiary-centre studies of extracorporeal liver support. 1
  • Device “completeness” and pathobiology: The editorial emphasised that Prometheus and MARS are detoxification systems without capacity to remove/replace dysfunctional albumin, and have limited cytokine clearance; this mismatch with ACLF biology (infection-driven inflammation and organ failure) may explain failure to translate biochemical improvement into survival benefit. 1
  • Concordant RCT evidence with other extracorporeal systems: The multicentre RELIEF trial evaluating MARS in a similar AOCLF population also failed to show an overall survival benefit, reinforcing HELIOS’s neutral result for routine extracorporeal detoxification as a mortality-modifying therapy. 2
  • Evidence synthesis: A recent systematic review/meta-analysis including MARS, Prometheus and related liver dialysis approaches summarises persistent heterogeneity in trial populations and endpoints, with biochemical improvements not reliably translating into consistent mortality reductions across studies. 4
  • Guideline integration: Contemporary European guidance on ACLF incorporates the extracorporeal liver support trial literature (including HELIOS-era devices) within its broader framework emphasising organ support, infection management, and transplant pathways. 5

Summary

  • HELIOS randomised 145 patients with severe acute deterioration of chronic liver disease (bilirubin ≥5 mg/dL; Child–Pugh ≥10) to Prometheus FPSA+SMT (n=77) vs SMT alone (n=68) with 90-day follow-up.
  • The trial stopped early for futility after interim analysis; no significant survival benefit was demonstrated at day 28 (66% vs 63%; P=0.70) or day 90 (47% vs 38%; P=0.35).
  • FPSA achieved substantial detoxification (bilirubin change −8 ± 11 vs 0 ± 9 mg/dL; P=0.0001) without clear downstream organ-support separation.
  • Safety was broadly comparable, with similar proportions experiencing bleeding (43% vs 49%) and bacterial infection (62% vs 54%).
  • Prespecified subgroup signals (MELD >30; Type 1 HRS) suggest potential benefit in severe reversible phenotypes but remain hypothesis-generating.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Terminology: HELIOS used “acute-on-chronic liver failure (AOCLF)” based on bilirubin and Child–Pugh thresholds; later frameworks (e.g., EASL-CLIF) define ACLF primarily by organ failure and short-term mortality risk, which may affect interpretation of “reversible” phenotypes and trial generalisability.

Overall Takeaway

HELIOS demonstrated that Prometheus FPSA can be delivered safely and achieves sustained biochemical detoxification in selected patients with severe acute decompensation of chronic liver disease. However, it did not improve 28- or 90-day survival in the overall cohort and was stopped early for futility, strongly suggesting that extracorporeal detoxification alone is not a reliable mortality-modifying strategy in unselected AOCLF/ACLF presentations.

Overall Summary

  • Prometheus FPSA reduced bilirubin substantially but did not improve 28- or 90-day survival in the ITT population.
  • Trial stopped early for futility; underpowered for modest effects.
  • Subgroup signals (MELD >30; Type 1 HRS) are hypothesis-generating and require dedicated confirmation.

Bibliography