A first-in-human trial of an immune checkpoint antibody in early Alzheimer's disease has reported that the drug was safe enough to continue testing, with signs that it engaged its biological target. The randomised, double-blind, placebo-controlled phase 1b of IBC-Ab002, run at 11 sites in the United Kingdom, Israel and the Netherlands, enrolled 40 people with mild cognitive impairment or mild dementia and spinal fluid evidence of amyloid. Five ascending dose cohorts received 1, 3, 6, 15 or 30 mg/kg intravenously, four times at three month intervals over 48 weeks, with two placebo recipients per cohort. Safety and tolerability were the primary endpoint. The results were published in Nature Medicine on 14 July 2026, timed to a late-breaking presentation at the Alzheimer's Association International Conference in London, with Catherine Mummery of University College London as principal investigator. The trial (NCT05551741) was funded in part by the US National Institute on Aging and the Alzheimer's Association.

The mechanism is not the amyloid-stripping approach behind the currently approved antibodies. IBC-Ab002 is an Fc-silenced anti-PD-L1 antibody engineered by ImmunoBrain, the company co-founded by Michal Schwartz of the Weizmann Institute of Science, and it is designed to be short-lived: brief, intermittent release of the PD-1/PD-L1 brake on peripheral immunity, on the hypothesis that recruited immune cells then help resolve inflammation in the brain. That hypothesis rests on rodent work. In this trial, PD-1 expressing memory T cells in the blood rose dose-dependently a week after dosing and fell back between doses, and the antibody's half-life was about four days without accumulation, which is the pharmacodynamic evidence the company points to for target engagement.

The biomarker findings were exploratory and thin. A second lumbar puncture at week 48 was optional, and the comparison rests on five people in the 30 mg/kg group and six on placebo: in that highest dose group, cerebrospinal fluid neurogranin, total tau and p-tau181 trended down while the placebo group did not move. Pooled lower dose cohorts showed no consistent change. CSF neurofilament light, GFAP, amyloid beta 42 and soluble TREM2 did not change at any dose, nor did plasma markers, nor did measures of cognition. All participants had at least one treatment-emergent adverse event. About a third of those on drug had immune-related adverse events, roughly half of them picked up only by laboratory monitoring; thyroiditis was the most common and resolved on its own in all but one person, who needed thyroid replacement. One severe, grade 3 event was an asymptomatic liver enzyme rise that resolved. No MRI abnormalities such as ARIA were reported. The only serious adverse event in the treated groups was a fall, judged unrelated to the drug.

Those autoimmune signals are the central question the programme now carries. Checkpoint inhibitors are established in oncology precisely because they lift immune restraint, and the same action produces inflammatory toxicity in other organs; the short half-life and 12 week gaps are the design intended to limit that risk in older adults dosed for years, and 40 people over one year cannot test whether it does. The company has said it is designing the next stage of clinical development with cognitive endpoints.