25 Apr 2025

Talk about a closely watched study. Conference after conference, Roche scientists running the ongoing Brain Shuttle AD trial of trontinemab are dispensing morsels of news from sequential interim analyses—and they are filling large auditoriums each time. The latest bit came at the AD/PD meeting, held April 1-5 in Vienna. It featured the latest cut from the Phase 2 trial of this anti-Aβ antibody equipped with a transferrin receptor-targeting domain that propels it into every nook and cranny of the brain.

As reported at previous conferences, the antibody managed to remove amyloid plaques quickly while triggering but a few mild cases of ARIA. In Vienna, the audience learned that the treatment dramatically shifted AD biomarkers, and that it made some regions of the brain shrink temporarily. Scientists attributed this to the extent of amyloid removal in those regions.

This ostensibly good news remains tempered by the previously reported death of one participant, who suffered a macrohemorrhage after receiving trontinemab. This person had superficial siderosis, which indicates cerebral amyloid angiopathy and a weakened cerebrovasculature, and her death made Roche add this condition to the trial’s exclusion criteria. In Vienna, scientists were relieved to learn that no further serious adverse events had occurred, including no new cases of ARIA-E. This is crucial, given that better safety is one of the two main goals scientists are pursuing for amyloid immunotherapy. (The other one: a bigger clinical benefit).

In Vienna, Roche’s Luka Kulic presented data from ongoing dose-finding studies in participants with mild cognitive impairment and mild to moderate AD. Kulic has been teasing snippets from the trial, most recently at November’s CTAD, where he shared findings from a September 2 interim analysis (Nov 2024 news). The Brain Shuttle AD trial initially enrolled 60 participants split equally into four ascending dose groups, then added a dose-expansion study with two cohorts of 60 people each, for more safety and biomarker data on the two highest doses of 1.8 and 3.6 mg/kg. At CTAD, Kulic reported that all volunteers had completed Part 1, while 100 had enrolled in the dose-expansion cohorts in Part 2, having received a few doses so far.

Back then, two serious adverse events had occurred the expansion study, including the fatality above. Two participants in Part 2—both in the 1.8mg/kg or placebo group—had developed ARIA-E. Amyloid-PET plummeted in the two highest-dose groups in completed Part 1, dipping below the threshold for most participants by 28 weeks.

So, what was new in Vienna? An interim analysis from a November 4 data cut. By this time, all participants in the 1.8 mg/kg cohort, and a third of the 3.6 mg/kg cohort of Part 2 had completed 28 weeks. Including participants from the two highest-dose cohorts of Part 1, the data Kulic presented came from 76 participants who had received 1.8 mg/kg trontinemab or placebo, and 38 who had received 3.6 mg/kg or placebo, for the full 28 weeks. In each cohort, participants were randomized to drug or placebo at a 4:1 ratio.

From this combined cohort, Kulic displayed amyloid-PET data. Everyone in Parts 1 and 2 had amyloid-PET scans at baseline and day 196, after receiving seven infusions of drug or placebo. In addition, Part 1 participants had an interim scan at day 78, while Part 2 participants were randomized 1:1:1 to have an interim scan at day 50, 78, or 106 after the second, third, or fourth dose, respectively. Kulic reported a precipitous drop in amyloid burden in response to both doses, with 3.6 mg/kg triggering an earlier and more extensive reduction than did 1.8 mg/kg (image below). By the trial’s end, 81 percent of participants on 3.6 mg/kg had crossed below the 24 centiloid threshold of brain-wide amyloid positivity, compared to 65 percent of those who received 1.8mg/kg.

Kulic reported no new safety issues, or instances of the known safety issues, beyond those previously reported at CTAD. The trial counted 19 cases of mild and transient anemia—presumably caused by trontinemab binding TfR-expressing reticulocytes. Infusion-related reactions plagued half of participants on the two highest doses. On that, Kulic said a protocol amendment adding a pretreatment with IV corticosteroids prior to each infusion had substantially reduced the frequency and severity of these reactions, which typically occur in response to the first dose.

There were no new incidences of ARIA, beyond the three mild cases of ARIA-E and the five cases of ARIA-H reported previously. Because the study is ongoing, both safety and ARIA findings remain blinded, and numbers come from the combined placebo and active groups from each cohort. So far, all of the ARIA-E and four ARIA-H cases occurred in the 1.8 mg/kg cohort, with one ARIA-H detected in the 3.6 mg/kg cohort.

That no ARIA-E cases have yet cropped up in the highest-dose cohort is striking, Kulic said, given that 75 percent of its participants are ApoE4 carriers, including 18 percent who carry two copies. ApoE4 homozygotes have had higher ARIA rates in trials of conventional anti-Aβ antibodies, which are also given at higher doses due to their poor brain penetrance.

Roche will start a Phase 3 trial later this year, Kulic said.

Next up, Roche’s Gregory Klein reported interim biomarker data from the same November 2024 cut point. First, Klein broke open the amyloid-PET data. He showed average images of all amyloid-PET scans from participants on 3.6 mg/kg, taken at baseline, 78, and 196 days. These people started out with many brain regions carrying a heavy amyloid burden, which had dropped a lot by 78 days, and nearly vanished by 196 days. Klein emphasized that trontinemab appeared to have removed amyloid throughout the brain. He attributed regional differences in removal to amyloid load at baseline, not differences in access by trontinemab.

Regarding the 81 percent of participants on 3.6 mg/kg who were amyloid-negative by the end of the trial, Klein said that the other 19 percent had started with the heaviest amyloid loads. By trial’s end, they were hovering just above the 24 centiloid positivity threshold but hadn’t yet crossed it. “There were no non-responders,” he said.

Lon Schneider, University of Southern California, Los Angeles, put the results in perspective. “Six months of monthly IV treatments seem to result in a relatively precise, predictable 90-100 centiloid plaque-lowering effect and a residual plaque level of about 10 centiloids, maybe lower. By comparison, it could take lecanemab and donanemab 12 to 18 months or longer to get to that level,” he wrote to Alzforum.

Whether greater amyloid removal will help patients remains unexplored. “To the extent that rapid plaque clearance and lower residual plaque load result in greater clinical benefit, trontinemab would be seen as a therapeutic advance. But without efficacy trials we won’t know,” Schneider noted (see comment below).

As did other anti-Aβ drugs, trontinemab caused a dose-dependent dip in brain size, as measured by volumetric MRI. This shrinkage happened fastest at the beginning of treatment, when large amounts of amyloid were being whisked away. A regional analysis revealed that the brain areas with the most amyloid removal also shrank the most, Klein reported in Vienna. The prefrontal cortex, where the most amyloid vanished, shrank by an average of 3 percent throughout the trial, while the lateral and medial temporal cortices, where less amyloid went away, shrank by 2.5 and 1.5 percent, respectively. The size of the hippocampus, which harbored little amyloid to begin with, remained constant throughout the trial among people who received the 3.6 mg/kg dose, but shrank ever so slightly in those on placebo.

Klein next compared the rate of atrophy before and after amyloid removal in the 3.6 mg/kg cohort. He found that the rate of shrinkage accelerated while amyloid was actively being cleared, and came to zero after amyloid was gone. This again was most pronounced in the prefrontal cortex (image below).

“We believe that the data suggests the volume changes we’re seeing are largely due to amyloid reduction, not a sign of neurodegeneration,” Klein said in Vienna.

Finally, Klein divulged bits of interim fluid biomarkers data. In CSF, the Aβ42/40 ratio increased by 84 percent throughout the trial in the 3.6 mg/kg group, whereas p-tau181, neurogranin, and total tau dropped by 29, 21, and 22 percent, respectively. In plasma, values were more variable among participants but moved in the same directions as CSF. Among people on 3.6 mg/kg, plasma Aβ42/40 nudged up by 13 percent, while p-tau181 and p-tau217 went down by 36 and 51 percent, respectively. Klein said these changes were trending toward normative levels.

What are these normal levels? David Morgan of Michigan State University in Grand Rapids asked. Klein acknowledged that while normal values are established for amyloid- and tau-PET, for fluid markers they are not. Ongoing research is still seeking to nail down cut points; as of now, he could only say that trontinemab treatment was pushing them back in that direction.

Another attendee asked how trontinemab might be affecting atrophy. Klein said that, based on histopathology, it’s clear that it is not driven purely by the volume of the amyloid lost, but connected to processes involved in amyloid removal. According to work by Nick Fox of University College London, so-called “pseudo-atrophy” might be accounted for by removal of plaques along with their “suburban sprawl” of other proteins, dystrophic neurites, and surrounding immune cells (Aug 2024 conference news).  

In Vienna, Klein asked if the low ARIA rate in response to trontinemab is starting to resolve controversies about the underlying cause of ARIA (Aug 2023 conference news). For example, the data might rule out the idea that rapid clearance of amyloid through perivascular drainage is the prime cause, Fox said. Klein did not opine about ARIA mechanisms, but instead noted that the low dose of trontinemab required to remove amyloid likely plays a strong role. “We need to study this further,” Klein said.

Also at AD/PD, Alector’s Gary Romano presented Phase 2 data showing that a TREM2 agonist antibody had caused ARIA without also removing amyloid (see Part 5 of this series).

All Aboard the TfR Train?
Trontinemab is the farthest along, but not the only such investigational drug. Many scientists at AD/PD seemed to be of the mind that cellular active transport machinery—including but not limited to TfR—is the ticket to getting therapies to neurons and glia more safely and effectively. While some groups are strapping new drugs to transport vehicles, others are fitting the vehicles with existing drugs. Case in point: the obvious one.

In Vienna, Dag Sehlin of Uppsala University in Sweden, who worked on lecanemab’s early development, presented preclinical data on “bi-lecanemab.” This version of lecanemab has a TfR binding domain, an analogy to trontinemab as a souped-up version of gantenerumab. Sehlin said that although lecanemab’s full removal of plaques within 18 months was impressive initially, expectations for the next generation of drugs have since risen.

Sehlin and colleagues have for years been studying this bispecific version of lecanemab, whose mouse version is mAb158. In 2018, they reported that the TfR-targeted version distributed broadly throughout the brain of amyloid-ridden mice, while conventional mAb158 was waylaid in the lateral ventricles and perivascular Aβ deposits, a loitering Sehlin now believes leads to ARIA (Syvänen et al., 2018). As TfR is most highly expressed on endothelial cells lining capillaries, this promotes passage across these tiny vessels into the parenchyma, avoiding the main sites of ARIA, Sehlin said.

In Vienna, Sehlin presented unpublished data on how bi-lecanemab distributes and engages its target in the mouse brain. First, he noted that lecanemab is thought to bind protofibrils, soluble precursors to fibrillar Aβ. However, when the scientists injected the TfR-equipped, radiolabeled bi-lecanemab into amyloidosis mice and analyzed brain extracts, they found the antibody overwhelmingly associated with insoluble, formic acid-extracted Aβ, likely derived from plaques.

Audioradiography experiments in amyloidosis mice supported bi-lecanemab’s penchant for plaques, as it lingered in plaque-ridden brain areas even after perfusion. PET scans of radiolabeled bi-lecanemab hinted at this as well, as the antibody lingered in amyloidosis mouse brain, but cleared out of wild-type brain within 72 hours.

This means it binds to something anchored in the tissue, probably plaques,” Sehlin said. He showed no comparison with conventional lecanemab on this question.

Sehlin closed with striking immunofluorescence images of bi-lecanemab engaged with a plaque. “I […] use this enormous screen here to blow up probably the biggest plaque of the conference,” he told the crowd, and then flashed an image of a green monster ringed by a blue halo of bi-lecanemab. Microglia, labeled red, swarmed into the plaque’s center.

Suzanne Hendrix, Pentara Corporation, asked the presenters whether efforts to curb ARIA risk of approved anti-Aβ antibodies might dampen their efficacy. To Sehlin, the answer is clear. “Shuttle versions [of amyloid-targeted antibodies] clear amyloid even more efficiently, quicker, and at lower dose.” He thinks more companies will adopt this strategy as positive clinical trial data continues to roll in. “This is the wave of the future,” he said.

BioArctic and Eisai started evaluating BAN2802, a “BrainTransporter” version of an undisclosed candidate AD drug (20 April 2024 press release), and earlier this year licensed BAN2803, an anti-pyroglutamated Aβ BrainTransporter antibody, to Bristol Myers Squibb (20 February press release). Neither is in clinical trials yet.

Per-Ola Freskgảrd of BioArctic in Stockholm noted that the trontinemab clinical data is fueling an overall interest in the pharma industry to apply this approach even beyond Alzheimer’s disease. “We are likely to reach a point soon where active delivery across the blood-brain barrier will be both required and the standard when using large molecules like antibodies to treat brain disorders where the target is behind the blood brain barrier,” he wrote to Alzforum.

Signs of this were apparent at AD/PD. They weren’t limited to amyloid-targeted drugs, to antibodies, or even to TfR. For one, Can Kayatekin of Sanofi reported preclinical findings on SAR446159, an α-synuclein antibody affixed with an insulin-like growth factor receptor binding domain that acts as a shuttle. A Phase 1 study in healthy volunteers is ongoing. For another, a complement-blocking antibody hitched to a TfR-binding domain purportedly shut down the inflammatory cascade in mouse models, according to Wioleta Zelek of Cardiff University.

Meanwhile, a curious crowd huddled around a poster presented by Shane Lofgren of Boston-based startup Manifold Bio. This company uses high-throughput screens to hunt for novel brain-shuttling alternatives to TfR, which come with risks due to their peripheral binding of reticulocytes. In Vienna, Lofgren reported the discovery of a novel target on the brain’s vasculature, which actively delivered an anti-Aβ antibody into the brain without harming blood cells or spiking pro-inflammatory cytokines.

It seems companies old and new are switching horses. They are racing, but have not had clinical trials yet.—Jessica Shugart

Comments

1. • Lon S. Schneider
     University of Southern California Keck School of Medicine
   • Posted: 25 Apr 2025

   Amyloid-targeting monoclonal antibodies are now established Alzheimer's treatments, approved for marketing in the U.S., despite their small clinical effects, controversy over whether the effects are clinically meaningful, and their lower-than-expected adoption by physicians and patients.

   The hope for amyloid-targeting antibodies in development is that they show greater effectiveness than lecanemab and donanemab such that some of this controversy is mitigated. Trontinemab may have accomplished the first step, perhaps showing more rapid and somewhat greater plaque clearance than donanemab. Six months of monthly IV treatments seem to result in a relatively precise, predictable 90-to-100 centiloid plaque-lowering effect and a residual plaque level of about 10 centiloids, maybe lower. By comparison, it could take lecanemab and donanemab 12 to 18 months or longer to get to that level.

   To the extent that rapid plaque clearance and lower residual plaque load result in greater clinical benefit, trontinemab would be seen as a therapeutic advance. But without efficacy trials we won’t know. Trontinemab, like aducanumab and lecanemab, could receive accelerated marketing approval based only on its plaque lowering before any clinical efficacy studies read out. This is because amyloid PET is an FDA-accepted surrogate marker that can substitute for a clinical endpoint like the CDR-SB. (If trontinemab is given accelerated approval, then it would be wholly up to CMS whether they will reimburse).

   Two notable characteristics of trontinemab are the very low rates of ARIA-E due to its transferrin-assisted, active transport across the BBB, an obvious advantage, and its very high rate of infusion-related reactions in most patients, requiring that all patients be pretreated with high-dose dexamethasone. That requirement essentially makes trontinemab part of a two-drug combination treatment with dexamethasone. As always, there are interesting tradeoffs to be explored.

   View all comments by Lon S. Schneider

Make a Comment

To make a comment you must login or register.

References

Therapeutics Citations

1. Trontinemab
2. Leqembi
3. Gantenerumab

News Citations

1. Trontinemab Data Strengthen Hope for Brain Shuttles 8 Nov 2024
2. Is the ‘Atrophy’ of Immunotherapy Just the Dismantling of Plaque ‘Suburbia’? 28 Aug 2024
3. Trialists Grapple with How to Outsmart TREM2 17 Apr 2025

Paper Citations

1. Syvänen S, Hultqvist G, Gustavsson T, Gumucio A, Laudon H, Söderberg L, Ingelsson M, Lannfelt L, Sehlin D. Efficient clearance of Aβ protofibrils in AβPP-transgenic mice treated with a brain-penetrating bifunctional antibody. Alzheimers Res Ther. 2018 May 24;10(1):49. PubMed.

Other Citations

1. Aug 2023 conference news

External Citations

1. 20 April 2024 press release
2. 20 February press release
3. Manifold Bio

Further Reading

News

1. What About the Brain Shrinkage Seen with Aβ Removal? 3 Apr 2023