Skip to main content

Foundational Trials

Keays

Keays R, Harrison PM, Wendon JA, Forbes A, Gove C, Alexander GJM, Williams R. Intravenous acetylcysteine in paracetamol induced fulminant hepatic failure: a prospective controlled trial. BMJ. 1991;303:1026-9

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Intravenous acetylcysteine in paracetamol induced fulminant hepatic failure: a prospective controlled trial
  • Acronym: Not applicable
  • Year: 1991
  • Journal published in: BMJ
  • Citation: Keays R, Harrison PM, Wendon JA, Forbes A, Gove C, Alexander GJM, Williams R. Intravenous acetylcysteine in paracetamol induced fulminant hepatic failure: a prospective controlled trial. BMJ. 1991;303:1026-9.

Context & Rationale

  • Background
    N-acetylcysteine (NAC) was established as highly effective when given early after paracetamol overdose (via glutathione repletion), but late initiation (>~15 hours) was widely viewed as ineffective for preventing severe liver injury.
    Reluctance to administer NAC late was reinforced by animal data suggesting potential harm and by concerns about hypersensitivity reactions with intravenous NAC.
    Retrospective data from the same centre suggested improved outcomes when NAC was started late (10–36 hours after overdose) despite no measurable improvement in conventional liver function indices (notably prothrombin time), creating equipoise for a randomised trial in established fulminant hepatic failure.
  • Research Question/Hypothesis
    In patients with established fulminant hepatic failure following paracetamol overdose who had not previously received NAC, does late intravenous NAC (continued until recovery from encephalopathy or death) improve survival and reduce major complications compared with placebo infusion plus the same intensive liver care?
  • Why This Matters
    Paracetamol-induced fulminant hepatic failure carries high short-term mortality and drives urgent decisions regarding intensive organ support and liver transplantation.
    If NAC conferred benefit even after fulminant hepatic failure is established, it would represent a low-cost, widely accessible intervention with major implications for late presenters and those awaiting transplant assessment.

Design & Methods

  • Research Question: Whether late intravenous NAC improves outcomes in established paracetamol-induced fulminant hepatic failure.
  • Study Type: Prospective randomised controlled trial; single-centre specialist liver failure unit (King’s College Hospital, London); parallel-group; open-label (not blinded).
  • Population:
    • Setting: Institute of Liver Studies liver failure unit; all participants received “conventional intensive liver care”.
    • Inclusion: 50 consecutive patients aged 16–60 with fulminant hepatic failure after paracetamol overdose; had not previously received acetylcysteine.
    • Exclusions: Not fully detailed beyond age limits and prior acetylcysteine exposure.
    • Severity / prognostic strata (used descriptively): “Poor prognosis” criteria included admission arterial pH <7.30; or prothrombin time >100 s plus creatinine >300 μmol/L plus grade 3–4 coma; or rising prothrombin time on day 4 vs day 3 after overdose.
  • Intervention:
    • Intravenous acetylcysteine: 150 mg/kg in 200 mL 5% dextrose over 15 minutes; then 50 mg/kg in 500 mL 5% dextrose over 4 hours; then 100 mg/kg in 1 L 5% dextrose over 16 hours.
    • Maintenance infusion continued after the standard regimen: 150 mg/kg per 24 hours in 1 L 5% dextrose, continued until recovery from encephalopathy or death.
  • Comparison:
    • Equivalent volume of 5% dextrose without acetylcysteine (placebo infusion) plus the same conventional intensive liver care.
    • Intensive care described included pulmonary artery catheter and arterial line; colloid to maintain filling pressures; dopamine for oliguria despite wedge pressure >12 mmHg; renal replacement therapy for refractory oliguria/anuria; intubation/ventilation for worsening coma; mannitol for clinical signs of cerebral oedema; inotropes for hypotension despite colloid.
  • Blinding: Unblinded; NAC could not be blinded due to a distinctive pungent aroma.
  • Statistics: Sample size of 25 per group planned to detect a 40% absolute difference in survival with 90% power at the 5% significance level; primary comparisons by χ² test; Mann–Whitney U test for prothrombin time and encephalopathy grade; log-rank test for time to death; survivors censored at 21 days.
  • Follow-Up Period: Up to 21 days (no further deaths occurred after day 16 in-trial).

Key Results

This trial was not stopped early. Recruitment proceeded to the planned 50 patients (25 per group).

Outcome IV acetylcysteine Control (5% dextrose) Effect p value / 95% CI Notes
Survival (in-trial) 12/25 (48%) 5/25 (20%) Difference in proportions surviving p=0.037; 95% CI for difference 3% to 53% Two participants randomised to NAC underwent orthotopic liver transplantation; analysis assumed they would have died.
Cerebral oedema (clinical signs) 10/25 (40%) 17/25 (68%) Difference in incidence p=0.047; 95% CI for difference 2% to 54% Diagnosis based on typical clinical signs (systemic hypertension and dilated pupils) and treated with mannitol, hyperventilation, thiopentone.
Hypotension requiring inotropic support 12/25 (48%) 20/25 (80%) Difference in incidence p=0.018; 95% CI for difference 7% to 57% Adrenaline or noradrenaline used to maintain mean arterial pressure.
Renal support (RRT) 10/25 (40%) 15/25 (60%) Difference in incidence Not significant; 95% CI for difference 7% higher to 47% lower RRT by intermittent haemodialysis or continuous arteriovenous haemofiltration.
Time to death (non-survivors) Not reported Not reported Log-rank test χ²=2.4 (df=1); p not reported Trial report states time to death was similar between groups.
Prothrombin time at day 4 (control time 15 s) 28 s (16–125) 46 s (20–340) No significant difference p not reported Values are medians (ranges); no significant differences on any day 1–4.
Encephalopathy grade at day 4 2 (0–4) 2 (0–4) No significant difference p not reported Values are medians (ranges); no significant differences on any day 1–4.
Adverse reactions (hypersensitivity) 0/25 Not applicable Not reported Not reported No adverse reactions to acetylcysteine were encountered; specifically no hypersensitivity reactions reported.
  • Survival was higher with NAC: 48% (12/25) vs 20% (5/25) with a reported 95% CI for the difference in survival of 3% to 53%.
  • Major complications commonly driving early death in fulminant hepatic failure were less frequent with NAC: cerebral oedema 40% vs 68%; hypotension requiring inotropes 48% vs 80%.
  • Despite survival and haemodynamic differences, conventional liver function indices tracked similarly (prothrombin time and encephalopathy grade), supporting a hypothesis of extra-hepatic physiological benefit rather than accelerated hepatocyte recovery.

Internal Validity

  • Randomisation and allocation: Allocation by opening one of 50 identical sealed envelopes on admission to the liver failure unit; the report does not specify sequential numbering or opaque safeguards beyond “identical sealed envelopes”.
  • Baseline comparability: Groups were similar for age (median 33 vs 34 years), sex (12/13 vs 9/16 M/F), time from overdose to presentation (median 28 vs 32 hours), and time from overdose to admission to the liver failure unit (median 53 vs 56 hours).
  • Severity balance: Similar numbers met at least one “poor prognosis” criterion at baseline (17 vs 18), including comparable day-4 prothrombin time rise criterion (13 vs 13).
  • Blinding and performance bias: Open-label by necessity (NAC smell); co-interventions were protocolised at a high level (described intensive liver care), but unblinded care could plausibly influence thresholds for inotrope initiation and some complication ascertainment.
  • Detection bias: Primary endpoint (survival) is objective; however “cerebral oedema” was defined clinically (systemic hypertension and dilated pupils) rather than via intracranial pressure monitoring or imaging, increasing susceptibility to ascertainment bias.
  • Protocol adherence / separation: Treatment separation was clear (NAC infusion vs dextrose placebo); NAC dosing followed standard early-overdose regimen then continued at 150 mg/kg/24 h until encephalopathy recovery or death; controls received an equivalent volume of dextrose.
  • Post-randomisation events: Two women in the NAC arm underwent liver transplantation; for statistical analysis the investigators assumed both would have died (a conservative assumption that is not an observed outcome and may affect interpretability).
  • Follow-up completeness: Short-term follow-up was essentially complete within the 21-day censoring window; no further deaths occurred after day 16 in the trial.
  • Timing of intervention: NAC was initiated on admission to the specialist unit at a late stage (median ~53–56 hours after overdose), aligning with the intended question (late rescue therapy in established fulminant hepatic failure).
  • Dose adequacy: Dose was “standard regimen” plus ongoing maintenance; the study did not compare alternative maintenance doses or durations, so optimality cannot be inferred.
  • Statistical rigour: A priori power calculation targeted a very large absolute survival difference (40%); small sample size yields wide confidence intervals and limits precision for secondary endpoints; multiple secondary comparisons were reported without adjustment.

Conclusion on Internal Validity: Moderate: randomisation and baseline balance support causal inference for survival, but open-label conduct, clinically defined secondary outcomes, small sample size, and the analytic handling of transplantation (counted as death) limit certainty and precision.

External Validity

  • Population representativeness: Consecutive patients with established fulminant hepatic failure after paracetamol overdose, aged 16–60, and not previously treated with NAC; this is most applicable to late presenters and to systems where early NAC is not universally delivered before tertiary referral.
  • Care environment: Specialist liver failure unit with invasive haemodynamic monitoring and transplant capability; outcomes and complication profiles may differ in centres without these resources.
  • Modern relevance: Current practice typically administers NAC early in suspected overdose; therefore, this trial’s key niche is the subgroup presenting late with established liver failure (or those with uncertain timing who were not treated prior to transfer).
  • Applicability across systems: The biological plausibility of NAC’s haemodynamic/microcirculatory effects supports generalisability, but the degree of benefit may vary with contemporary critical care practices and transplant thresholds.

Conclusion on External Validity: Moderate: findings are most generalisable to late-presenting paracetamol toxicity with established fulminant hepatic failure in centres capable of advanced organ support; generalisability to early-treated overdose populations is limited.

Strengths & Limitations

  • Strengths: Prospective randomised design; consecutive enrolment; clinically meaningful primary endpoint (survival); pragmatic intervention using a readily deployable drug with a clearly specified infusion strategy continued until recovery/death; studied a highly selected high-mortality phenotype where effect sizes could be clinically large.
  • Limitations: Small single-centre trial; unblinded intervention; clinically adjudicated complications (notably cerebral oedema); limited detail on allocation concealment and eligibility exclusions; multiple secondary outcomes with borderline p values; transplantation handled via assumption (counted as death) rather than observed counterfactual.

Interpretation & Why It Matters

  • Clinical signal
    Late intravenous NAC in established paracetamol-induced fulminant hepatic failure was associated with higher survival (48% vs 20%) and fewer major complications (cerebral oedema 40% vs 68%; inotrope-requiring hypotension 48% vs 80%).
    The absence of an obvious between-group difference in prothrombin time trajectory suggests any benefit may be mediated through systemic/organ-support effects (haemodynamics, microcirculation, oxidative stress) rather than rapid restoration of synthetic function.
  • Practical implication
    For late presenters with established encephalopathy and coagulopathy after paracetamol overdose, this trial supports continuing NAC as a disease-modifying adjunct alongside high-intensity supportive care, rather than restricting NAC to the “early antidote” window.
  • Methodological implication
    The trial illustrates a common acute care challenge: interventions with plausible benefit late in disease may show outcome improvement through extra-target effects, while conventional disease biomarkers (here prothrombin time) remain unchanged.

Controversies & Subsequent Evidence

  • Evidence base concerns for this specific clinical niche persist: the Keays trial is small, single-centre, and unblinded, with wide confidence intervals around the survival effect; replication at scale in paracetamol-induced fulminant hepatic failure has been limited (partly because early NAC has become near-universal).
  • Modern hepatology guidance and position papers generally endorse NAC in paracetamol-associated acute liver failure (and commonly in late presenters), which aligns with the trial’s core premise that NAC may remain beneficial after onset of severe liver injury.12
  • Randomised evidence in non-paracetamol acute liver failure supports broader biological plausibility for NAC as a late adjunct in acute liver failure syndromes (improved transplant-free survival in early coma grades), reinforcing the concept that NAC’s benefit may extend beyond antidotal glutathione repletion.3
  • Systematic reviews of NAC in paracetamol poisoning and acute liver failure highlight heterogeneity of populations, timing, and outcomes, and underscore that high-grade randomised outcome evidence in established fulminant hepatic failure remains limited despite long-standing widespread use.4

Summary

  • Single-centre randomised controlled trial of late intravenous NAC in established paracetamol-induced fulminant hepatic failure (n=50; 25 per group).
  • Survival was higher with NAC: 48% (12/25) vs 20% (5/25); p=0.037; 95% CI for difference 3% to 53%.
  • Major complications were less frequent with NAC: cerebral oedema 40% vs 68% (p=0.047) and inotrope-requiring hypotension 48% vs 80% (p=0.018).
  • Liver synthetic function trajectory (prothrombin time) and encephalopathy grades were similar across groups during the first 4 days.
  • No adverse or hypersensitivity reactions to NAC were reported in this critically ill cohort.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Where DOI landing pages were not readily available for older foundational studies, PubMed search landing pages are provided to support traceability without displaying DOI text.

Overall Takeaway

In a small but rigorous-for-era randomised trial in established paracetamol-induced fulminant hepatic failure, late intravenous acetylcysteine was associated with substantially higher survival and fewer life-threatening complications, despite similar trajectories in conventional liver function measures. The trial remains “landmark” because it challenged the then-dominant view that NAC is only an early antidote, supporting its role as a late adjunct in acute liver failure physiology and informing modern guideline-endorsed practice.

Overall Summary

  • Late IV acetylcysteine in established paracetamol fulminant hepatic failure improved survival (48% vs 20%) and reduced cerebral oedema and inotrope-requiring hypotension.
  • Benefit occurred without measurable acceleration of prothrombin time recovery, suggesting extra-hepatic physiological mechanisms.
  • Evidence remains limited by small size and open-label conduct, but later guidelines broadly align with continued NAC use in acute liver failure syndromes.

Bibliography