Skip to main content

Foundational Trials

REVERSE

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Terlipressin Plus Albumin Is More Effective Than Albumin Alone in Improving Renal Function in Patients With Cirrhosis and Hepatorenal Syndrome Type 1
  • Acronym: REVERSE
  • Year: 2016
  • Journal published in: Gastroenterology
  • Citation: Boyer TD, Sanyal AJ, Wong F, Frederick RT, Lake JR, O'Leary JG, et al. Terlipressin plus albumin is more effective than albumin alone in improving renal function in patients with cirrhosis and hepatorenal syndrome type 1. Gastroenterology. 2016;150(7):1579-1589.

Context & Rationale

  • Background
    • Hepatorenal syndrome type 1 (HRS-1; contemporaneous definition) represents rapidly progressive functional renal failure in advanced cirrhosis and ascites, with very high short-term mortality and frequent need for transplantation as definitive therapy.
    • Pathophysiology is dominated by marked splanchnic vasodilatation and effective arterial hypovolaemia, with renal vasoconstriction; therefore vasoconstrictor therapy plus albumin volume expansion is biologically plausible.
    • Terlipressin (a vasopressin analogue) was widely used outside North America; prior RCTs and meta-analyses suggested improved renal function and HRS reversal, but the evidence base was heterogeneous (definitions, endpoints, and safety reporting).
    • At the time, regulatory approval in the US/Canada was lacking; a large, rigorously blinded trial using a stringent “confirmed reversal” endpoint was needed.
  • Research Question/Hypothesis
    • Does terlipressin (with protocolised dose escalation) plus albumin, compared with placebo plus albumin, increase the proportion of patients achieving confirmed HRS reversal and improve clinically meaningful outcomes in HRS-1?
    • Hypothesis: terlipressin would increase confirmed HRS reversal beyond albumin alone without unacceptable harms.
  • Why This Matters
    • HRS is a frequent ICU-adjacent problem in decompensated cirrhosis, where reversibility of renal dysfunction can determine candidacy for transplant and need for renal replacement therapy (RRT).
    • Demonstrating efficacy (and quantifying harms) in a multicentre North American setting would directly inform clinical practice, trial design, and regulatory decisions.
    • Use of a stringent “confirmed reversal” endpoint aimed to distinguish durable renal recovery from transient creatinine fluctuations in an unstable population.

Design & Methods

  • Research Question: In adults with cirrhosis, ascites, and HRS-1, does terlipressin (with albumin) increase confirmed HRS reversal compared with placebo (with albumin)?
  • Study Type: Randomised, multicentre (50 US and 2 Canadian centres), double-blind, placebo-controlled, phase 3 trial; October 2010 to February 2013; design published before completion 1
  • Population:
    • Inclusion: Adults (≥18 years) with cirrhosis and ascites; HRS-1 by 2007 International Club of Ascites criteria (doubling of serum creatinine (SCr) to ≥2.5 mg/dL within 2 weeks) with no sustained improvement (<20% decrease in SCr) despite diuretic withdrawal and albumin.
    • Key exclusions: SCr >7 mg/dL; mean arterial pressure (MAP) <70 mmHg with hypoperfusion; sepsis; untreated infection; recent (<48 h) use of vasoactive agents (octreotide, midodrine, vasopressin, dopamine, or other vasoconstrictors); intrinsic renal disease/structural kidney injury.
    • Baseline severity (illustrative): MELD 33.5 ± 5.8 (terlipressin) vs 32.6 ± 5.4 (placebo); baseline SCr 3.6 ± 0.9 vs 3.7 ± 1.0 mg/dL; CLIF-SOFA 11.3 ± 2.2 vs 11.3 ± 2.0.
  • Intervention:
    • Terlipressin: 1 mg IV bolus over 2 minutes every 6 hours (4 mg/day).
    • Escalation: Increased to 2 mg IV every 6 hours if SCr decreased by ≤30% by day 4 (protocolised non-response approach).
    • Albumin co-therapy: Recommended 20–40 g/day; continued at investigator discretion (not fully standardised).
    • Duration: Continued until confirmed reversal, treatment failure, liver transplantation, RRT, or up to 14 days.
  • Comparison:
    • Placebo: Matching IV bolus every 6 hours.
    • Albumin co-therapy: Recommended 20–40 g/day (same approach as intervention arm).
    • Rescue/co-interventions: RRT and transplantation permitted as clinically indicated; criteria for non-response and discontinuation were protocolised.
  • Blinding: Double-blind (participants, clinicians, and assessors); matching placebo dosing schedule; objective creatinine-based primary endpoint reduced risk of detection bias.
  • Statistics: Power calculation targeted an absolute increase in confirmed HRS reversal from 34% (placebo) to 48% (terlipressin) with 80% power at two-sided α=0.05, requiring 180 patients; enrolment continued until ≥30 confirmed reversal events accrued, resulting in 196 randomised; primary analysis by intention-to-treat (stratified CMH approach as specified in trial reports) 1
  • Follow-Up Period: Up to 90 days post-randomisation (including 30-, 60-, and 90-day follow-up).

Key Results

This trial was not stopped early. Recruitment continued to achieve the pre-specified minimum number of confirmed reversal events, with 196 patients randomised.

Outcome Terlipressin (+ albumin) Placebo (+ albumin) Effect p value / 95% CI Notes
Confirmed HRS reversal (CHRSR) 19/97 (19.6%) 13/99 (13.1%) Not reported P=0.22; 95% CI not reported Primary endpoint; required 2 SCr values ≤1.5 mg/dL ≥40 hours apart while on treatment (within 24 hours of last dose) and no intervening RRT or liver transplant.
HRS reversal (≥1 SCr ≤1.5 mg/dL) 23/97 (23.7%) 15/99 (15.2%) Not reported P=0.13; 95% CI not reported Secondary/clinical efficacy signal; less stringent than CHRSR.
Change in SCr from baseline to end of treatment −1.1 mg/dL −0.6 mg/dL Not reported P<0.001; 95% CI not reported Renal function improved more with terlipressin despite negative primary endpoint.
Estimated creatinine clearance at end of treatment (Cockcroft–Gault) LS mean (SEM) 23.0 (1.49) mL/min LS mean (SEM) 14.4 (1.36) mL/min Not reported P<0.0001; 95% CI not reported Supports physiological effect on renal perfusion/function.
Change in MAP from baseline to end of treatment +5.3 mmHg −0.5 mmHg Not reported P<0.01; 95% CI not reported Haemodynamic response aligned with vasoconstrictor mechanism.
90-day survival (Kaplan–Meier probability) 0.577 0.537 Not reported Log-rank P=0.61; 95% CI not reported No evidence of survival benefit.
90-day transplant-free survival (Kaplan–Meier probability) 0.308 0.248 Not reported Log-rank P=0.56; 95% CI not reported Median transplant-free survival: 23.8 vs 20.7 days.
Treatment withdrawal owing to adverse events (safety population) 19/93 (20.4%) 6/95 (6.3%) Not reported Not reported Higher discontinuation burden with terlipressin.
Treatment-related serious adverse events (safety population) 14/93 (15.1%) 2/95 (2.1%) Not reported Not reported Safety signal relevant to ICU monitoring/selection.
Ischaemic adverse events leading to discontinuation (safety population) 7/93 (7.5%) 1/95 (1.1%) Not reported Not reported None were fatal; all resolved after discontinuation per trial report.
    • The primary endpoint (confirmed HRS reversal) was not met (19.6% vs 13.1%; P=0.22), despite consistent physiological signals (greater improvements in SCr, creatinine clearance, and MAP).
    • No improvement was demonstrated in survival or transplant-free survival through 90 days (survival probability 0.577 vs 0.537; P=0.61).
    • Terlipressin increased treatment-related adverse events and discontinuations (treatment-related AEs 66.7% vs 40.0%; treatment withdrawal owing to AEs 20.4% vs 6.3%).

Internal Validity

    • Randomisation and allocation: Central randomisation with stratification (qualifying SCr <3.6 vs ≥3.6 mg/dL; alcoholic hepatitis yes/no); concealment supported by identical placebo dosing schedule.
    • Drop out/exclusions: 196 randomised; 4/97 (terlipressin) and 4/99 (placebo) did not receive study drug; loss to follow-up at 90 days was minimal (0 vs 1), with withdrawal of consent 1 vs 1.
    • Performance/detection bias: Double-blind design; primary endpoint based on objective laboratory values; however, clinician-driven outcomes (RRT initiation, timing of transplant) can still introduce variability.
    • Protocol adherence: Mean duration of study drug 5.6 ± 3.9 days (terlipressin) vs 6.2 ± 4.1 days (placebo); dose escalation to 2 mg q6h occurred in 23/93 (24.7%) vs 23/95 (24.2%); dose interruptions 17.2% vs 21.1%.
    • Baseline characteristics: Broadly balanced for illness severity (MELD 33.5 ± 5.8 vs 32.6 ± 5.4; baseline SCr 3.6 ± 0.9 vs 3.7 ± 1.0 mg/dL; MAP 76.2 ± 11.5 vs 75.3 ± 11.4 mmHg); notable imbalance in sex (female 46.4% vs 32.3%; P=0.04).
    • Heterogeneity: Multicentre design across 52 sites with mixed cirrhosis aetiologies increases clinical heterogeneity; stratification addressed two key prognostic variables but cannot fully mitigate site-level differences in supportive care and thresholds for RRT.
    • Timing: Eligibility required failure to improve after diuretic withdrawal/albumin and meeting HRS-1 criteria (SCr ≥2.5 mg/dL), which likely selected for later-stage renal dysfunction where reversibility is reduced.
    • Dose: Bolus terlipressin regimen (4–8 mg/day) aligns with common clinical use at the time; continuous infusion strategies (used in later practice) were not evaluated.
    • Separation of the variable of interest: Haemodynamic separation was evident (MAP change +5.3 vs −0.5 mmHg; P<0.01); albumin exposure was not fully standardised and differed numerically (concomitant albumin use 87.6% vs 86.9%; mean total albumin 185.9 g vs 246.2 g; mean duration 3.9 vs 5.1 days).
    • Outcome assessment: CHRSR was stringent (two qualifying creatinine values ≥40 hours apart) and clinically defensible, but reduced the observed event rate versus assumptions used in the power calculation.
    • Statistical rigour: Primary analysis was intention-to-treat; the observed CHRSR rate was substantially lower than assumed, increasing the risk of a false-negative result for the primary endpoint despite achieving the pre-specified minimum event count.

Conclusion on Internal Validity: Moderate to strong internal validity for measuring the physiological effect of terlipressin (blinding, objective endpoints, good follow-up), but the trial’s ability to detect a clinically meaningful effect on confirmed reversal was limited by much lower-than-expected event rates and non-standardised albumin exposure.

External Validity

    • Population representativeness: The enrolled cohort reflects very advanced cirrhosis with severe renal dysfunction (baseline SCr ~3.6–3.7 mg/dL; MELD ~33), managed at largely transplant-capable North American centres.
    • Important exclusions: Patients with MAP <70 mmHg with hypoperfusion, sepsis/untreated infection, recent vasoactive exposure, or suspected intrinsic renal disease were excluded, limiting applicability to common ICU phenotypes (shock, sepsis-associated AKI, mixed aetiology renal injury).
    • Generalisability to contemporary HRS definitions: HRS-1 criteria used a high creatinine threshold (≥2.5 mg/dL); contemporary practice increasingly targets earlier HRS-AKI, where responsiveness and harm profile may differ.
    • Co-interventions: Albumin dosing was recommended rather than tightly protocolised; real-world systems with different albumin availability or prescribing culture may observe different net effects.

Conclusion on External Validity: Generalisability is moderate for stable, carefully selected patients with advanced HRS managed in transplant-centre systems, but limited for broader ICU populations (sepsis, shock, mixed AKI) and for earlier-stage HRS-AKI.

Strengths & Limitations

  • Strengths:
    • Randomised, double-blind, placebo-controlled design across 52 North American sites.
    • Stringent, objective primary endpoint intended to reflect durable renal recovery (confirmed reversal).
    • Excellent follow-up completion through 90 days with minimal loss to follow-up.
    • Protocolised dose escalation and discontinuation criteria reduced clinician-driven treatment variation.
  • Limitations:
    • Primary endpoint not met; observed event rates were far lower than assumed in the power calculation, increasing risk of type II error.
    • Albumin use was recommended but not standardised; total albumin exposure was numerically higher in the placebo group (246.2 g vs 185.9 g), potentially diluting treatment separation.
    • Old HRS-1 definition (SCr ≥2.5 mg/dL) preferentially enrolled late-stage disease, where reversibility may be less likely.
    • Terlipressin increased treatment-related adverse events and discontinuations, including ischaemic events (7.5% vs 1.1% leading to discontinuation).

Interpretation & Why It Matters

  • Clinical implications
    • REVERSE provides high-quality evidence that terlipressin produces consistent haemodynamic and renal function improvements (MAP +5.3 vs −0.5 mmHg; SCr change −1.1 vs −0.6 mg/dL), but did not demonstrate a statistically significant increase in confirmed HRS reversal or any survival advantage.
    • The safety signal (higher treatment-related serious adverse events and discontinuations) reinforces that terlipressin should be delivered with ICU-level monitoring and careful selection, rather than as a “low-risk” ward intervention.
    • For trialists and methodologists, REVERSE highlights how endpoint stringency and optimistic event-rate assumptions can render a biologically effective intervention “negative” on the primary endpoint, even when physiological separation is clear.

Controversies & Subsequent Evidence

    • Endpoint choice and power assumptions: The stringent “confirmed reversal” endpoint reduced transient-response misclassification but substantially lowered observed event rates versus design assumptions, raising the likelihood of a false-negative primary result (editorial critique) 2.
    • Co-intervention imbalance: Total albumin exposure and duration were higher in the placebo group (246.2 g over 5.1 days vs 185.9 g over 3.9 days), complicating interpretation of an add-on vasoconstrictor effect and potentially biasing towards the null.
    • Safety–efficacy trade-off: Treatment-related adverse events (66.7% vs 40.0%) and treatment withdrawal owing to adverse events (20.4% vs 6.3%) were higher with terlipressin; ischaemic events leading to discontinuation occurred in 7.5% vs 1.1% (editorial emphasis on careful patient selection and monitoring) 2.
    • Effect modification signals (hypothesis-generating): CHRSR appeared higher with terlipressin among those with recent large-volume paracentesis (32.4% vs 8.3%; P=0.04), while subgroup estimates were otherwise imprecise; these findings were not adjusted for multiplicity and should not be over-interpreted.
    • Subsequent RCT evidence: The later CONFIRM trial evaluated terlipressin plus albumin in a similar disease space and is central to modern evidence synthesis and guideline recommendations 3.
    • Guideline evolution: Post-REVERSE guidance and consensus statements incorporate terlipressin within broader AKI/HRS frameworks (including earlier recognition and structured escalation), with increasing emphasis on patient selection and monitoring for complications 45678.

Summary

    • REVERSE was a multicentre, double-blind, placebo-controlled North American phase 3 trial (n=196) evaluating terlipressin (with albumin) for HRS-1.
    • The primary endpoint (confirmed HRS reversal) was not significantly improved: 19.6% (19/97) vs 13.1% (13/99); P=0.22.
    • Terlipressin produced clear physiological effects: greater improvement in SCr (−1.1 vs −0.6 mg/dL; P<0.001), creatinine clearance (23.0 vs 14.4 mL/min; P<0.0001), and MAP (+5.3 vs −0.5 mmHg; P<0.01).
    • No benefit was shown in survival or transplant-free survival through 90 days (survival probability 0.577 vs 0.537; P=0.61).
    • Harms were clinically relevant: treatment-related serious AEs (15.1% vs 2.1%) and discontinuations (withdrawal owing to AEs 20.4% vs 6.3%) were higher with terlipressin; ischaemic events leading to discontinuation occurred in 7.5% vs 1.1%.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • For some older studies and certain meta-analyses, DOI landing pages could not be confirmed from available source material in this build; PubMed record links/search links are provided as a practical fallback while preserving citation accuracy.
  • Where both terlipressin and norepinephrine are available, comparative effectiveness and safety are highly sensitive to patient selection (baseline hypoxaemia/volume status, infection, ACLF severity) and local thresholds for RRT and transplantation.

Overall Takeaway

REVERSE is a landmark North American phase 3 trial because it separated physiological efficacy from patient-centred benefit: terlipressin clearly improved renal haemodynamics and creatinine trajectories, yet failed to significantly increase confirmed HRS reversal or improve survival. Its stringent endpoint and unexpectedly low event rates shaped subsequent trial designs and highlighted the need for careful patient selection and safety-focused delivery when using potent splanchnic vasoconstrictors in advanced cirrhosis.

Overall Summary

  • In severe HRS-1 (SCr ~3.6–3.7 mg/dL; MELD ~33), terlipressin improved renal physiology but did not meet the confirmed reversal primary endpoint and increased treatment-related harms—making patient selection and monitoring central to any real-world implementation.

Bibliography