Adding a blood-cleansing filter to dialysis cut deaths, but only just

Adding a blood-purifying step to routine dialysis modestly lowered the risk of death over more than three years in people with kidney failure, according to a large randomised trial from Shanghai — but the headline result sat right on the edge of statistical significance, and the trial's open-label design leaves room for doubt [s1][s2]. The finding is intriguing because death rates on long-term dialysis remain stubbornly high, and much of that mortality is cardiovascular, driven in part by toxins that standard dialysis clears poorly [s1].

For all the refinements of the past two decades, people on long-term dialysis still die at rates that would be scandalous in most other chronic diseases, and remarkably few interventions have been shown in randomised trials to change that trajectory [s1]. Conventional haemodialysis filters small molecules well but struggles with larger, protein-bound compounds that accumulate in advanced kidney disease and are thought to inflame blood vessels [s1]. Hemoadsorption works differently: blood is passed over a cartridge whose material binds those larger molecules directly. The question this trial set out to answer was whether bolting that step onto usual dialysis would translate into fewer deaths [s1].

What the trial did

The trial was randomised, open-label and run across 11 haemodialysis centres in Shanghai, China [s1][s2]. It enrolled adults who had been on maintenance haemodialysis for at least three months and were receiving an adequate dose of dialysis (a clearance marker, Kt/V, of at least 1.2) [s1]. In all, 1,362 patients were randomly assigned 1:1 to hemoadsorption combined with haemodialysis (683 patients) or to haemodialysis alone (679 patients, mostly low-flux dialysis plus intermittent haemodiafiltration) [s1]. Every randomised patient was included in the analysis [s1]. The primary outcome was all-cause mortality; secondary outcomes included cardiovascular death and major cardiovascular events [s1].

What it found

Over a median follow-up of 39.5 months, all-cause mortality occurred in 117 of the combined-therapy patients (17.1%) versus 144 of those on dialysis alone (21.2%) [s1]. That worked out to a hazard ratio of 0.778, with a 95% confidence interval of 0.609 to 0.994 and a P value of 0.045 — a reduction that clears the conventional significance line, but barely [s1]. The cardiovascular results were firmer: cardiovascular mortality fell with a hazard ratio of 0.659 (95% CI 0.481 to 0.901; P=0.009), and major cardiovascular events with a hazard ratio of 0.772 (95% CI 0.621 to 0.959; P=0.019) [s1]. Important adverse events, chiefly infections and abnormal blood pressure, were comparable between the two groups, with no new safety signal attributable to the added step — a reassuring finding, since any extra procedure in frail dialysis patients carries its own risks [s1].

How to read it

A mortality benefit in dialysis is notable, because few interventions have moved that needle, and the consistency across all-cause death, cardiovascular death and cardiovascular events is reassuring that the signal is not a fluke of one endpoint [s1]. But two caveats deserve equal billing with the result. First, the primary comparison was borderline: a confidence interval whose upper bound reaches 0.994 and a P value of 0.045 describe an effect that is statistically present but fragile, the sort that can evaporate in a second trial [s1]. Second, the trial was open-label — patients and staff knew who was getting the extra treatment — which can subtly shape care and the ascertainment of softer endpoints, even if it is harder to bias death itself [s1].

There are also questions the single-city setting raises. All 11 centres were in Shanghai, the comparator was a particular mix of lower-intensity dialysis, and whether the same benefit would appear against high-flux dialysis or haemodiafiltration as the default comparator elsewhere is unknown [s1]. Hemoadsorption cartridges also add cost and complexity to every session, so even a real benefit of this size would need to be weighed against resources in systems where dialysis capacity is already stretched [s1].

What to watch

The result makes a strong case for a confirmatory trial — ideally multinational, with a modern dialysis comparator and endpoints adjudicated by assessors blinded to treatment — to test whether the mortality signal holds [s1]. Also worth watching is mechanism: if the benefit is real and driven by clearing vessel-damaging toxins, biomarker substudies could show which patients stand to gain most, turning a population-level average into something clinicians can act on [s1][s2]. For now, the trial is a promising lead rather than a reason to change practice [s1].

This article describes research and is not medical advice. Decisions about dialysis are for patients and their kidney specialists.

Sources

• Hemoadsorption combined with hemodialysis versus hemodialysis alone on mortality — Nature Communications, 28 March 2026
• Trial registration, NCT03227770 — ClinicalTrials.gov