Polymyxin B haemoadsorption in endotoxic septic shock (Tigris): a multicentre, open-label, Bayesian, randomised, controlled, phase 3 trial

Background: Endotoxic septic shock, a subset of septic shock with high endotoxin activity and multiorgan failure, is associated with high risk of death. We sought to identify the effect of endotoxin removal from the blood with polymyxin B haemoadsorption on mortality. Methods: We conducted an open-label, randomised, controlled, phase 3 trial at 19 US hospitals, enrolling adults (aged ≥18 years) with septic shock requiring vasopressors, with multiple organ dysfunction, and with endotoxin activity between 0·60 and 0·89 units. Patients were randomly assigned (2:1) to receive two sessions of polymyxin B plus standard of care or standard of care alone (control), with block randomisation stratified by site. Study personnel were masked to treatment allocation. Polymyxin B haemoadsorption was given via haemodialysis at a blood flow rate of 80–120 mL/min for 90–120 min per session, 22 h apart. The primary outcome was 28-day mortality in the intention-to-treat cohort. The safety cohort included all participants exposed to any amount of study treatment. Design and analysis followed a Bayesian framework, using a prior for the treatment effect based on a subgroup of the earlier EUPHRATES trial. 90-day mortality was the key secondary outcome. The trial was registered at ClinicalTrials.gov and is completed (NCT03901807). Findings: Between Sept 30, 2019, and April 10, 2025, we screened 14 890 patients, of whom 157 were enrolled (106 assigned to polymyxin B and 51 to control; 66 [42%] women and 91 [58%] men). At 28 days, 41 (39%) patients in the polymyxin B group and 23 (45%) in the control group had died, yielding a posterior probability of benefit of 95·3% (APACHE-II adjusted odds ratio 0·67 [95% credible interval 0·39–1·08]). At 90 days, the posterior probability of benefit was 99·4% (0·54 [0·32–0·87]). Of 100 patients in the polymyxin B group assessed for safety, 30 (30%) had a serious adverse event compared with 11 (22%) of 51 patients in the control group (difference –8 percentage points [95% CI –22 to 6]). Two (2%) serious adverse events in the polymyxin B group were treatment-related, one related to polymyxin B and one to catheter placement. Interpretation: In patients with endotoxic septic shock defined by high endotoxin activity and multiorgan failure, polymyxin B haemoadsorption was associated with a high probability of lower mortality at 28 days and 90 days. Funding: Spectral Medical.