ALBIOS

Discussed in Oxford 30 May 14 – report by Joel Meyer

Citation: N Engl J Med 2014; 370:1412-1421April 10, 2014 DOI: 10.1056/NEJMoa1305727 (NEJM link)

The use of albumin in the resuscitation and treatment of patients with shock, burns, hypovolaemia and other critical illness was commonplace in the second half of the 20th century. This was largely based on the observed association between hypoalbuminaemia and organ dysfunction/death.
Following a Cochrane meta-analysis in 1998 which showed lack of benefit and possibly an increased risk of death from albumin, the use of albumin in the UK ICU community fell sharply. However the subsequent SAFE trial showed absence of harm from 4% albumin (except in head injured patients) and possibly a signal towards benefit in the septic cohort. Since then albumin has remained a therapeutic option in the treatment of severe sepsis, and features in the Surviving Sepsis Guidelines “when patients require substantial amounts of crystalloids” with a grade 2C recommendation.
The Italian investigators behind the ALBIOS trial set out to pragmatically test the efficacy of albumin as an adjunctive resuscitation fluid in severely septic patients within 24 hours of diagnosis. 1818 patients in 100 Italian ICUs were enrolled. The intervention group received 300 ml of 20% albumin in addition to crystalloids until ICU discharge. The target albumin level was 30 g/L. Crystalloids were administered in both groups according to goal directed therapy for sepsis initially, and then at the discretion of the clinician. The trial ran from 2008 to 2013. Importantly, the administration of albumin was an unblinded intervention.
The primary outcome, 28d mortality, was unchanged: 31.8% in the albumin group and 32% in the crystalloid only group. There were no significant differences in the secondary outcomes either. In terms of clinical measures there was a negligible difference in MAP (77 v 79 mmHg), a very slightly lower heart rate (94 v 99 bpm), a very slightly higher CVP (11 v 10 mmHg), a slightly more negative fluid balance in the first seven days, and a higher serum albumin (30 g/L versus 23 g/L) in the albumin group. The daily total fluid administered to the two groups did not differ.
It is worth noting that the authors did a retrospective subgroup analysis of the 1121 patients who had septic shock, showing survival benefit from albumin (43.6% versus 49.9%), and correspondingly, a trend towards increased mortality with albumin in the patients without septic shock.
What can we make from these results? Most importantly, although there may be a physiological advantage of giving albumin in severe sepsis, and that 20% albumin may be safe, there isn’t any overall impact on patient outcome. The retrospective subgroup analyses probably aren’t interpretable due to bias.
This was a rigorously conducted single-country multicentre trial with clear objectives and a practical real-world approach. The patients probably reflect our UK practice, with a mortality from severe sepsis of 30%, age about 70 years, 60% medical and 35% emergency surgical, 80% mechanically ventilated and with a moderate severity of illness. However there are problems with the design. Firstly, the use of albumin was unblinded which makes it hard to assume that the patients were treated similarly in all other ways by the clinicians. Secondly, the design was based on a projected mortality rate was 45% whereas the actual mortality was just over 30%. Thirdly, the paper states that “synthetic colloids were not allowed” whereas the supplementary appendix details the protocol violations which include 23% of patients in each group receiving synthetic colloids at least once. Moreover, 37% patients in the crystalloid only group received albumin at least once, and the rules for albumin administration were violated at least once in 42% of patients in the intervention group. So perhaps the conduct was less rigorous than implied in the paper. Finally, the investigators used 20% albumin which differs from the 4% albumin used in the SAFE trial.
The other useful data from the supplementary appendix is the microbiological analysis, which shows: primary site of infection to be lung 40%, abdomen 40%, urinary tract 15%; rate of positive site culture 63%; rate of positive blood culture 33%; organism type to be Gram positive 14%, Gram positive 20%, mixed bacteria 6%, virus 1%, fungus 3%; antibiotics given at randomisation to 93%. These data are always of interest when comparing the study population to one’s day to day practice.
The message seems to be that albumin is relatively safe in severe sepsis but without appreciable benefit. There’s no benefit from targeting a higher serum albumin of > 30g/L per se. We await a response from the Surviving Sepsis Campaign, but given the expense of albumin, it is unlikely to be recommended for routine practice.