Background

This is an anti-tau IgG4 antibody that grew out of a research collaboration between AC Immune and Genentech, part of the Roche Group. The collaboration focused on approaches that target extracellular, not intracellular, tau. The aim was to explore antibodies with reduced effector function in an effort to limit microglial activation leading to inflammatory responses (Lee et al., 2016). At the 2017 AD/PD conference, Genentech reported that RO7105705 binds the N-terminus of all six isoforms of human tau, both monomeric and oligomeric, regardless of phosphorylation status, and that 13 weeks of treatment with either 3, 10, or 30 mg/kg of the antibody dose-dependently reduced brain pathology while raising plasma tau levels in P301L mice. Chronic dosing was safe in mice or cynomolgus monkeys, Genentech also reported (Apr 2017 conference news). 

Findings

From June 2016 to June 2017, Genentech ran a Phase 1 study of 74 volunteers comprising both healthy controls and people with mild to moderate Alzheimer's disease. Conducted in Tennessee, this trial compared the antibody to placebo on safety, tolerability, pharmacokinetics, and preliminary activity outcomes. It combined single doses, dose escalation, and multiple dosing, given both intravenously and subcutaneously. At the 2017 AD/PD and AAIC conferences, Genentech reported that single doses in healthy volunteers went as high as 16,800 mg, with a 15-day window observed between a given dose and the next-higher dose, and that 70 percent of the subcutaneous doses were bioavailable. Results remained blinded at this point in time, but the trial had not generated serious adverse events; minor AEs related to the drug included bruising and pain at the injection site. RO7105705 plasma half-life was 32 days, and plasma and CSF concentration increased with dose (Apr 2017 conference news; Aug 2017 conference news). Phase 1 data was subsequently published (Ayalon et al., 2021).

In October 2017, Genentech started TAURIEL, a Phase 2 study in 457 people with prodromal or probable AD ascertained by a positive amyloid PET or CSF Aβ42 finding, and mild symptoms. Participants were randomized to an 18-month course of monthly infusions of placebo or 1,500, 4,500, or 8,100 mg of RO7105705, with a 96-week open-label extension option for those who completed the blinded portion of the trial. This trial used Genentech's tau PET tracer GTP1 at baseline and to measure treatment response at week 73. Primary outcomes were change on the CDR sum of boxes and safety; secondary measures include the Repeatable Battery for Assessment of Neuropsychological Status (RBANS), the new Amsterdam Instrumental Activity of Daily Living questionnaire, the ADAS-Cog13 and ADCS-ADL Inventory, as well as serum drug concentration and anti-drug antibodies. This trial ran at 97 locations in North America, Australia, and various European countries.

The placebo-controlled phase completed in July 2020 and in September, Genentech reported top-line results. Semorinemab missed the primary efficacy endpoint of reducing decline on the CDR sum of boxes, and missed on both of two secondary endpoints, the ADAS-Cog13 and ADCS-ADL (press release). According to results presented at the 2020 CTAD, there was no change in any other secondary clinical endpoint, either, nor did the antibody slow tangle accumulation in brain relative to placebo. Tau-PET showed a significant increase from baseline in all groups by 12 months, and continued accumulation out to 18 months. Antibody pharmacokinetics were dose-proportional, and the two highest doses elicited maximal increases in plasma tau. The drug was safe. The company terminated the trial during the open-label period, and it ended in January 2021. Additional biomarker data were presented at AD/PD 2021 (Mar 2021 conference news). CSF total tau and phosphorylated tau decreased in all treated groups, but downstream markers of neurodegeneration and inflammation did not change. Trial results were published after peer review (Teng et al., 2022).

In February 2019, Roche started LAURIET, another Phase 2 study in 272 people with a diagnosis of probable AD confirmed by amyloid positivity via PET or CSF testing, and with moderate dementia. The study comprised a screening, double-blind treatment, optional open-label extension period, and a safety follow-up period. The study was to have up to three cohorts, receiving 4,500 mg monthly for 48, 60, or 72 weeks, though the last cohort was never enrolled. Co-primary endpoints were change from baseline on both ADAS-Cog11 and ADCS-ADL; secondary endpoints included CDR-SB, MMSE, adverse events, serum concentration of RO7105705, and anti-drug antibodies to RO7105705. This trial used GTP1 to monitor tauopathy. It ran at 49 sites across the United States, France, Poland and Spain. The placebo-controlled part of the study ran until July 2021, and the study was completed in August 2023.

On August 31, 2021, the sponsors announced topline results indicating a 43.6% slowing of decline on the ADAS-Cog11 co-primary, though no benefit of the other cognitive or functional outcomes (see Sep 2021 news). A Phase 3 decision is pending additional data from LAURIET's ongoing long-term extension study and biomarker measurements. In November at CTAD, Genentech showed this trial's data, and reported the treatment did not change tangle accumulation measured by GTP1 PET (Nov 2021 conference news). According to published results, most of the benefit on the ADAS-Cog was driven by the memory domain, specifically word recognition (Monteiro et al., 2023). This could explain the lack of effect on other outcome measures, or may be a false positive result (see accompanying 2023 editorial by Edland and Llibre-Guerra). As in TAURIEL, the LAURIET CSF substudy found significant reductions in total tau and ptau181, suggesting target engagement. Modeling of semorinemab pharmacokinetics and plasma tau measures across the trials suggested greater than 90 percent peripheral target engagement (Ramakrishnan et al., 2024). A separate computational modeling study claimed that target engagement in the synaptic cleft, the location of tau propagation, could be as low as 1 percent (Geerts et al., 2023).

On January 22, 2024, Genentech and Roche ended its collaboration with AC Immune to develop semorinemab (press release).

An evaluation of putative biomarkers for neuroinflammation identified multiple complement proteins to be elevated in AD patient CSF compared to cognitively unimpaired elderly. However, CSF samples from TAURIEL and LAURIET showed no difference in these proteins in treated versus placebo groups (Sandoval et al., 2024).

For all trials with this antibody, see clinicaltrials.gov.