Terlipressin versus norepinephrine as infusion in patients with septic shock

Liu ZM. ICM 2018; published first online 3rd July 2018. Doi:10.1007/s00134-018-5267-9

Clinical Question

  • In critically ill patients with septic shock, does the use of terlipressin compared to norepinephrine reduced 28-day mortality?


  • Sepsis is one of the most common conditions treated by critical care clinicians worldwide
  • The best management strategies e.g. target parameters, fluids, vasopressors, etc remain widely debated
  • Traditional vasopressors such as norepinephrine and adrenaline are now being joined by newer agents, such as vasopressin, terlipressin and angiotensin 2, with mixed results
  • Terlipressin, being more selective for V1 receptors, has theoretical benefits over non-selective vasopressin
  • Previous trial in patients with liver cirrhosis showed that terlipressin is as effective as norepinephrine as a vasopressor in patients with cirrhosis and septic shock


  • Randomised, multicentre
  • Double-blinded (patient, clinicians and assessors)
  • Stratified by investigating centres combined with block randomisation
  • Randomisation by sequential numbered, opaque, computer-generated sealed envelopes
  • Modified intention-to-treat analysis
  • Power calculation
    • Baseline mortality 50% at 28 days
    • Powered at 80% to detect absolute 10% difference
    • 10% estimated drop-out rate
    • Two-sided alpha error of 0.05
    • Target recruitment 1100 pts


  • 21 ICUs in China
  • January 2013 to February 2016


  • Inclusion: Over 18 years old with septic shock
    • 2 SIRS criteria
    • Proven or suspected infection
    • Hypotension despite adequate fluid resuscitation
      • Hypotension defined as SBP<90mmHg, MAP<70mmHg or SBP decrease >40mmHg or > 2 standard deviation below normal for age
  • Exclusion: unstable coronary syndrome, previous use of terlipressin for BP support during current ICU admission, malignancy or disease for which mortality was estimated to be very high, acute mesenteric ischaemia (suspected or proven), Raynaud’s phenomenon, pregnancy, organ transplantation
  • 684 pts screened -> 617 randomised -> 526 analysed (260 terlipressin, 266 noradrenaline
    • 14% lost to follow up, withdrew or excluded post-randomisation
  • Baseline characteristics similar in both groups
    • Mean age 61
    • Mean SOFA 11
    • Mean APACHE2 score 19
    • Mean BP 67mmHg


  • Terlipressin
    • Made as 0.02 mg/ml in 50mls syringes
    • Commenced at 1 ml/hr (0.33 mcg/min)
    • Titrated 1–8 ml/hr (max 2.66 mcg/min)


  • Norepinephrine
    • Made as 0.22 mg/ml in 50mls syringes
    • Commenced at 1 ml/hr (3.66 mcg/min)
    • Titrated 1–8 ml/hr (max 29.33 mcg/min)

Management common to both groups

  • Titrated to achieve target BP (suggested MAP of 65 to 75 mmHg but left to discretion of treating ICU physician)
  • Open label norepinephrine was allowed if target BP not achieved with maximum rate of study drug
  • Other open label vasopressors if maximum dose of study drug AND open label norepinephrine failed to achieve target BP
  • Tapering of open label drug was conducted first before study drug tapered
  • Tapering of study drug only when target BP maintained for 12 hrs without open label drugs
  • Study drug was interrupted if any serious adverse events occurred (defined)
  • If subsequent recurrence of shock required vasopressors, the study drug was preferentially used


  • Primary outcome:
    • No significant difference in 28 day all cause mortality
    • Terlipressin group: 40%
    • Norepinephrine group: 38%
    • Absolute risk increase (ARI): 2% (95% CI -9.8% to 18.8%; P = 0.633)
  • Secondary outcome:
    • No significant difference in:
      • Changes in SOFA score on day 7 after randomisation
      • Days alive
      • Days free of vasopressor during 28 days after randomisation
    • More serious adverse events in terlipressin group
      • Terlipressin group 30% vs noradrenaline group 11.65%
      • Mainly digital ischaemia

Authors’ Conclusions

  • There was no significant reduction in 28-day mortality in patients with septic shock treated with terlipressin compared to norepinephrine


  • Good external generalisability – multicentre trial with inclusion and exclusion criteria comparable to other trials
  • Study protocol published on
  • Consideration of possible heterogeneity in patients and inter-hospital care in randomisation process, by using blocks of random allocation
  • Very thorough concealment of sequence, randomisation, and blinding of those involved, which reduces systematic biases and strengthens the internal validity


  • Trial was terminated (50% recruitment) due to futility – the trial was under-powered as the observed mortality rate was considerably less than the expected mortality rate
  • Significant number of patients lost post-randomisation which could introduce bias
  • Significant proportion of the terlipressin group required addition of norepinephrine and dopamine (86%) with the mean norepinephrine dose of 0.48 mcg/kg/min
  • Baseline mean BP at randomisation was 67mmHg which suggests that the patients in this study may not have required vasopressor therapy
    • Data in the supplementary material showed that both groups had fluid intake of positive four litres per day in the first 2 days after randomisation
  • Short term outcomes (28 day and hospital mortality) were assessed and therefore any long-term differences between treatment groups cannot be assessed

The Bottom Line

  • This trial does not show a mortality benefit from using terlipressin compared to norepinephrine as a first-line vasopressor
  • Use of terlipressin as first line led to significantly more serious adverse events, especially digital ischaemia
  • I shall continue my current practice of using norepinephrine as first-line vasopressor, and add vasopressin in those patients with vasopressor-refractory septic shock

External Links


Summary author: Adrian Wong
Summary date: 19 July 2018
Peer-review editor: Duncan Chambler

One comment

  • Duncan Chambler

    This is a frustrating trial!

    They concluded at ~50% recruitment that it was futile and stopped the trial. What they mean is that they were never going to find p<0.05 given the results so far. We have already criticised the excessive use of P values in our blog.

    They state the absolute risk increase is 2% (95% CI -9.8 to 18.8%). I calculate slightly different 95% CI of -6.3% to 10.4%.
    What they conclude is that the true absolute difference is somewhere between 9.8% in favour of terlipressin to 18.8% in favour of noradrenaline. That's a massive range and both extremes are clinically important.

    Using my calculated 95% CI values, it could be that terlipressin is beneficial with a number-needed-to-treat (NNT) of up to 16, or it could be harmful with a number-needed-to-harm (NNH) of up to 10. Those are potentially game changing in my opinion.

    I use terlipressin (second line to norad though) and I want to know if that terli is good, bad or indifferent for my patients, and this trial proves neither. This conclusion is worse than futile; it is ethically wasteful in my opinion.

    I understand resources must be justified, but I am disappointed that this trial was terminated early despite the wide and clinically important 95% Confidence Intervals.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.