Six randomized trials of a cytokine filter in the bypass circuit. The kidney endpoint did not move. That is not the same sentence as the filter does not work.
The Research
Abou-Arab O, Beyls C, Ibrahima A, et al. Hemoadsorption to prevent acute kidney injury in elective cardiac surgery: a systematic review of randomized controlled trials with meta-analysis and trial sequential analysis. Anaesthesia Critical Care & Pain Medicine, 2026. doi:10.1016/j.accpm.2026.101931. · PubMed 42906746
Six randomized trials of a cytokine filter in the bypass circuit. The kidney endpoint did not move. That is not the same sentence as the filter does not work.
Pulled every randomized trial of hemoadsorption during bypass in elective cardiac surgery. Six trials, 798 patients. Four of those reported kidney injury, covering 723 patients.
Kidney injury, relative risk 0.93, interval 0.71 to 1.21, p 0.57. Heterogeneity of 42 percent, so the trials were not telling wildly different stories. Nothing else moved either: interleukin-6, TNF, 28-day mortality, SOFA, ICU days.
The Hallway
We have half the patients the math says we need, and the running result has not moved far enough either way to call it. This question is not finished.
Palm TreeHowever, the accrued sample represented only 48% of the heterogeneity-adjusted required information size (1,491 patients), and the cumulative Z-curve crossed neither the monitoring nor the futility boundary.
The protocol was prospectively registered, so the endpoint was not chosen after the data. They used a hard outcome, kidney injury, rather than a lab value. And they ran a trial sequential analysis, which most meta-analyses skip, which is the step that stops you reading a flat result as a final answer.
Elective cases only. The kidney endpoint rests on four of the six trials, so the pooled group is smaller than the headline number suggests. The cytokine surrogate did not move, which raises the question whether the cartridge was doing the job it is sold for. And a pooled null can hide benefit in one subgroup and harm in another.
The InFlow-mation
Critical Conclusion
No kidney benefit was shown. The trials collected 48 percent of the patients they calculated they needed, so the question stays open rather than closed.
Take It Back to the Pump
Not strong enough to buy on. I am not buying it yet. If a rep brings it, I will ask how many patients and how many events. If the answer is about fifteen hundred, I will look again.
Pump off.
On the source
Peer reviewed, PROSPERO CRD42025634991. Elective adult cardiac surgery only. The kidney endpoint rests on four of the six trials. Do not carry this to emergency cases or to children.