Score another one for amyloid immunotherapy. The anti-pyroglutamated amyloid antibody donanemab has posted positive top-line results in the Phase 3 Trailblazer-Alz2 study, according to an Eli Lilly press release. The treatment slowed decline on the primary outcome, the combined cognitive and functional measure iADRS, by more than a third. Results on all secondary clinical endpoints were positive. Overall, donanemab cut the risk of progressing to the next stage of disease over 18 months by about 40 percent.

Donanemab slowed clinical progression by a third in Phase 3.
People with fewer tau tangles benefitted most.
ARIA remains a problem, with two deaths in the trial ascribed to it.

Eli Lilly said it will apply for approval from the U.S. Food and Drug Administration this quarter. If approved, donanemab will become the third amyloid immunotherapy in the U.S., after Aduhelm and Leqembi. Lilly plans to submit in other countries as well.

Alzheimer’s researchers hailed the news. "The donanemab Phase 3 trial is a major accomplishment on many levels," wrote David Knopman, Mayo Clinic, Rochester Minnesota. “The Trailblazer results represent a major advance in the treatment of AD,” agreed Stephen Salloway at Butler Hospital in Providence, Rhode Island. As did others, Salloway noted innovative trial features such as stratifying participants by tau tangle load, which helped target treatment to the right disease stage. Eric Siemers at Siemers Integration LLC, previously at Lilly, pointed out the consistency of findings between donanemab, lecanemab, and aducanumab. “Given these results, it can no longer be said that we have no treatments to slow the inexorable progression of AD,” he wrote. Indeed, the amyloid hypothesis is no longer a hypothesis, wrote Dennis Selkoe at Brigham and Women’s Hospital, Boston (comments below).

Others hoped that the accumulating weight of evidence in favor of anti-amyloid antibodies will prompt the Centers for Medicare and Medicaid Services to reconsider its decision limiting coverage (Apr 2022 news). “It is clear that this class of amyloid-removing therapies is effective, and this should encourage CMS to fully cover these treatments,” said Paul Aisen at the University of California, San Diego (comment below).

Benefit More Robust in Phase 3 Than 2
Often in drug development, a potential efficacy signal in Phase 2 vanishes in Phase 3. Not here. The earlier Phase 2 Trailblazer1 study of donanemab was positive on its primary outcome, reporting a 32 percent slowing of decline on the iADRS. However, it missed significance on some secondary measures such as the CDR-SB, notching only a 23 percent slowing there. That trial enrolled 257 people, half of whom received placebo (Mar 2021 news).

The Phase 3 trial enrolled 1,182 people in its primary analysis population. As in Phase 2, these participants had a tangle burden that fell between 1.1 and 1.46 SUVR on tau PET, the range where Lilly scientists predicted removing plaque would be most effective. In this group, about a third of people became amyloid-negative by six months, and 71 percent by one year. This was slightly more than in Phase 2, where 60 percent of participants reached this threshold by one year.

Cognitive and functional decline slowed on all outcome measures: 35 percent on the iADRS, 37 percent on the CDR-SB, 32 percent on the ADAS-Cog13, and 40 percent on ADCS-iADL. All results were statistically significant. Looking at the data another way, 47 percent of people on donanemab stayed stable on the CDR-SB over one year, compared with 29 percent on placebo.

Tangle Load May Be Key
Trailblazer-Alz2 also included 552 participants who had baseline tau PET scans above 1.46 SUVR. This “high tau” group was included to gather more data about the effects of immunotherapy later in disease. In its press release, Lilly reported that a combined analysis of the full 1,734 participant cohort was still positive, with slowing of 22 percent on the iADRS, 29 percent on the CDR-SB, 20 percent on the ADAS-Cog13, and 28 percent on the ADCS-iADL. The company did not report numbers for the high-tau subgroup alone; Lilly told Alzforum those data will be presented later.

“One can infer that they did not respond as well to the drug, which strengthens the argument for early treatment,” noted Erik Musiek at Washington University in St. Louis (comment below).

Some recent data hint that the 1.46 threshold may represent a point beyond which plaque removal does little good. In Lilly’s much smaller Trailblazer-Alz4 study comparing donanemab to aducanumab, participants above this threshold gained no benefit on downstream biomarkers after plaque clearance. Likewise, gantenerumab Phase 3 data found no benefit in women, whose baseline tangle load in the temporal lobe topped 1.46 SUVR, but a significant benefit in men, who started with a lower tau signal (Apr 2023 conference news). If the findings about such a threshold hold up, then the field might have an answer to the question of when to stop amyloid immunotherapy, as well as a clearer idea about when to start.

Some commenters noted clinical implications, especially if donanemab’s label requires baseline tau PET scans, as well as amyloid PET scans to determine when plaque is cleared, as was done in the trial. “If payers require explicit replication of the clinical trials for reimbursement, clinical use of donanemab could be limited to quaternary referral centers for the immediate future,” Siemers speculated. Others thought this might not be necessary. “While tau PET studies will not be available in the foreseeable future to the vast majority of early AD patients, the principal of treating at early clinical stages is achievable,” Selkoe wrote. Yet others saw this as another argument to get the leading fluid biomarkers ready for prime time.

The stopping provision of this trial—i.e., participants came off donanemab once their amyloid load fell below a predefined threshold—drew both questions and praise. Some asked how clinicians are to know when amyloid returns, and what to do then. But Gil Rabinovici of University of California, San Francisco, wrote “A limited duration of treatment, akin to induction of chemotherapy for cancer, could be a game-changer for patients, families, health systems, and third-party payers. This approach may also limit the time in which patients are exposed to the risk of ARIA, which is particularly critical given the reported deaths in the study.”

Safety Is a Concern
As in Phase 2, about a quarter of people taking donanemab developed the brain swelling known as ARIA-E. In 6 percent of people, it was symptomatic. The microhemorrhages known as ARIA-H occurred in 31 percent of people on drug, compared with 14 percent on placebo. Most concerning, 1.6 percent of participants had serious ARIA (either E or H). Lilly attributed two deaths in the trial to ARIA, and a third death is possibly related, as well, occurring after an episode of serious ARIA. Lilly's press release offered no further details about these deaths, but a company representative told Alzforum that antithrombotic medications, including anticoagulants, were not a factor.

In the open-label extension study of lecanemab, three deaths were associated with use of anticoagulants or acute thrombolytics, leading researchers to recommend excluding people who are on these therapies (Apr 2023 conference news). Both the lecanemab and donanemab Phase 3 trials allowed people on such drugs to enroll.

“The deaths are very worrisome,” Musiek said, calling for research to look for ways to lower the risk. On this, commenters were unanimous. Salloway asked for a thorough review of serious and fatal cases, and the relationship of APOE genotype to ARIA and safety outcomes. Selkoe noted that deep analysis of symptomatic versus asymptomatic ARIA cases in donanemab, lecanemab, and aducanumab might identify clinical and biomarker characteristics that could help predict its occurrence beyond the known effect of APOE4 gene dosage. On that, Siemers pointed out that across these three antibodies, despite differing amounts of ARIA with each, a consistent 20 to 25 percent of these cases were symptomatic. What might be going on there?

Despite these concerns, Michael Weiner at the University of California, San Francisco, noted that the overall risk/benefit ratio of donanemab remains favorable.

A fuller set of data from this trial will be presented at this summer’s Alzheimer’s Association International Conference in Amsterdam. “We all eagerly await full presentation of the data,” Aisen said.—Madolyn Bowman Rogers


  1. The donanemab Phase 3 trial is a major accomplishment on many levels: its unequivocal success in demonstrating clinical benefit, the fact that Lilly was able to conduct and complete a very large Phase 3 trial within two years of reporting the Phase 2 story, and the success using tau PET to enroll people with the biological profile most likely to respond.

    Much more information about the trial results will need to be disclosed to understand who the participants were, how response differed across clinical characteristic, and the details of the ARIA risks by genotype and baseline patient characteristics. 

    That said, these results will increase interest in anti-amyloid therapies, though it is not clear how long it will be before donanemab is available clinically. And, if tau PET characteristics are indeed important in selecting persons in the AD pathway most likely to benefit from donanemab therapy, the lack of access to tau PET will need to be addressed. Ultimately if donanemab is approved, weighing the many different characteristics of donanemab and lecanemab will be a challenging task for patients, families, and physicians.

  2. The Trailblazer results represent a major advance in the treatment of AD. The study included important innovations such as use of amyloid and tau PET to determine the disease stage and implementing stopping rules based on amyloid-lowering with PET.

    It remains to be seen how these approaches will be utilized in clinical practice. Three new deaths related to ARIA and macrohemorrhage were reported, and a thorough review of the safety data, serious and fatal cases, and the relationship of ApoE genotype to ARIA and safety outcomes is necessary. Also, a careful review of adverse outcomes related to patients on anticoagulation is needed.

    We look forward to seeing the impact of treatment on plasma p-tau and other plasma biomarkers. There is an urgent need to bring plasma biomarkers into the clinic to help to screen for cerebral amyloid and to validate tests such as p-tau217 to help measure target engagement and treatment response. We salute Lilly’s decades of dedication to AD research and we are excited that this treatment provides new opportunities to build on.

  3. This represents another major advance in the field of AD research. In the target population—people with intermediate amounts of tau based on tau PET at baseline—a slowing of 36 percent based on the CDR-SB or 35 percent based on the iADRS was demonstrated, both of which were highly significant.

    The field has progressed incrementally over time to get to this point. In hindsight, even the small effects of solanezumab in patients with mild dementia due to AD in the Expedition3 study would appear to be consistent with a drug effect. This was followed by the controversial results from the Engage and Emerge studies with aducanumab, although many people in the field concluded based on the totality of the data that a drug effect was present. Subsequently the results from the Clarity study using lecanemab were clearly positive. And now we again have clearly positive results from Trailblazer-ALZ2 using donanemab. Given these results broadly, it can no longer be said that we have no treatments to slow the inexorable progression of AD.

    Despite these broad advances, much is still to be learned regarding these different antibodies, their different mechanisms of action, and the differences in efficacy and safety. Solanezumab targeted Aβ monomers, had very limited efficacy, but also had rates of ARIA that were essentially the same as placebo. Donanemab and aducanumab primarily target plaque, and when plaque is reduced to the point where it is essentially no longer present on amyloid PET scans (<25-30 centiloids), as was seen in Trailblazer-ALZ, Trailblazer-ALZ2, and Emerge, slowing of disease progression can be demonstrated. 

    These plaque-targeting antibodies also cause rates of ARIA-E of 24 percent to 35 percent. The point has been appropriately raised that since most cases of ARIA-E are asymptomatic, only symptomatic cases of ARIA-E should be a focus of attention. This point may be valid, but note that across the aducanumab, donanemab, and lecanemab trials, although the overall rate of ARIA-E may be different, about 20 to 25 percent of ARIA cases are symptomatic with each drug. Lecanemab was developed to target Aβ protofibrils, although plaque reduction certainly does occur. Interestingly, the ARIA-E rate for lecanemab is substantially less than for the plaque-targeting antibodies (12.7 percent in Clarity), which would suggest that the different target of lecanemab results in a lower frequency of ARIA-E. 

    Two small biotech companies, Acumen Pharmaceuticals and ProMIS Neurosciences, are developing monoclonal antibodies that target Aβ oligomers, ACU193 and PMN310. Phase 1 results for ACU193 are expected in 3Q2023 and an IND was recently awarded for PMN310. Additional clinical trials will determine whether targeting Aβ oligomers might further reduce the risk of ARIA-E while maintaining efficacy.

    Finally, how these disease-modifying antibodies will be used in clinical practice is not yet entirely clear. The label for donanemab will be of great interest since the inclusion/exclusion criteria for the studies not only includes evidence of amyloid positivity but also an intermediate amount of tau. The trials also stopped therapy with donanemab if a follow-up amyloid PET scan became negative. If payers require explicit replication of the clinical trials for reimbursement, then clinical use of donanemab could be limited to quaternary referral centers for the immediate future. 

    The two or three fatalities in the trial related to ARIA are not without consequence. Nevertheless, after the press release from Lilly today, patients and their families have added hope that the progression of their AD can be slowed.

  4. Lilly’s Phase 3 top-line results on donanemab are further good news for patients with early Alzheimer’s disease. Meeting the primary and all secondary endpoints with high statistical significance leaves no doubt that antibody-mediated amyloid lowering can robustly slow the biological and clinical progression of AD, even in a brief period of 12 to 18 months. The results revealed so far support the thoughtful design Lilly had used for their Phase 2 trial: stratifying patients by degree of tau aggregate load in the brain and finding that patients with an intermediate level appear to do better than those with higher levels. While tau PET studies will not be available in the foreseeable future to the vast majority of early AD patients, the principle of treating at early clinical stages is achievable.

    The “amyloid hypothesis” of AD (now no longer a hypothesis) indicates that accrual of abnormal forms of tau in tangles and neurites is a key downstream response (as is neuroinflammation) to the accumulation of Aβ-protein in fibrillar plaques and in more diffusible toxic forms like small, fibrillar oligomers. Once fluid biomarker data for donanemab Phase 3 become available (presumably at AAIC in July), one can predict that tau/phospho-tau levels in CSF and plasma declined on average, and plasma GFAP may have also gone down. The fact that 47 percent of the participants on donanemab experienced no significant clinical progression (per CDR-SB) at one year (vs. 29 percent on placebo) clearly supports the occurrence of disease modification at both the biological and clinical levels.

    ARIA remains the central adverse effect of amyloid-lowering antibodies, and donanemab’s rate of 6.1 percent symptomatic ARIA (with three deaths) is higher than was seen with lecanemab in the CLARITY AD study (2.8 percent). The field will need to deeply analyze the symptomatic vs. asymptomatic cases of ARIA from all three disease-slowing antibodies to date to identify clinical and biomarker characteristics that may help predict its occurrence, beyond the known effect of ApoE4 gene dosage.

    Regarding the much-debated clinical meaningfulness of lowering amyloid, the Trailblazer-ALZ 2 outcome of about 40 percent slowing of decline on ADL suggests that, like lecanemab, donanemab can provide benefit to many early AD patients on measures that matter in their everyday lives.

    Together with the earlier results for aducanumab and lecanemab, the consistent donanemab outcomes strongly suggest that we can ameliorate the progression of AD while watching carefully for and managing symptomatic ARIA in experienced clinics. And the results in the primary analysis population (intermediate tau) strongly support moving even earlier to presymptomatic AD. Prevention of AD now appears achievable, but it will take a lot of further hard work, including on other targets.

  5. The Phase 3 donanemab results are great news for the field, still further confirmation that disease progression can be slowed by robust removal of fibrillar amyloid from brain. We all eagerly await full presentation of the data from this innovative study to clarify issues around early discontinuation of therapy after amyloid PET normalization, stratification by tau PET signal, safety details, and what to expect in the anticipated label.

    But it is clear that this class of amyloid-removing therapies is effective, and this should encourage CMS to fully cover these treatments. The donanemab data supports early intervention, adding to our expectation that amyloid removal in preclinical AD, now under study with lecanemab and donanemab, will yield a greater impact on disease progression. These are exciting times for AD therapeutics.

  6. The top-line results from Lilly today are very exciting to me for several reasons. First, these data really strengthen the argument that plaque removal has a meaningful clinical effect in mildly symptomatic patients. With Phase 1b/2 trials for aducanumab, lecanemab, and donanemab all showing a signal, and now Phase 3 trials of lecanemab and donanemab (and one of the aducanumab trials) all meeting primary clinical endpoints, it is become increasing clear that there is a clinical benefit from this class of drugs.

    Second, donanemab has some very appealing characteristics. It appears to have similar or better efficacy than lecanemab (possibly due to faster plaque clearance), but its dosing is once a month, and the drug is stopped after plaque clearance. These two factors could make this drug much more attractive to patients (as opposed to two-week dosing that continues indefinitely).

    The stratification by tau PET is also really interesting. While the data was not shown for the “high tau PET” group alone, one can infer that they did not respond as well to the drug, which strengthens the argument for early treatment.

    However, ARIA is a concern. The rate of serious ARIA is higher than lecanemab’s, and the deaths are very worrisome. More information about these events and how they related to known risk factors (such as ApoE genotype or medical factors) is needed.

    A few questions for the future come to mind:

    • Are there ways to limit severe ARIA risk with these drugs? Biomarkers? Changes to the dosing regimen?
    • Is lecanemab actually safer than donanemab? If so, might it be preferentially used in higher-risk patients?
    • Is the tau PET stratification scheme used in this trial feasible in clinical practice?
    • How will we detect when plaques re-accumulate, and what can we do then? If donanemab cannot be given again due to antibody responses, would a different agent be used? Is a maintenance dose needed? The details of long-term therapy are unclear.
  7. The top-line results of Trailblazer-ALZ2 are very encouraging and important to the field. First and foremost, we now have converging evidence from two well-designed and well-executed Phase 3 randomized controlled trials that significant amyloid removal at an early clinical stage of AD can slow cognitive and functional decline. Importantly, this slowing is shown to be clinically meaningful, as reflected by a larger proportion of patients remaining stable on the CDR-SB, and a hazard ratio of 0.61 for progression in disease stage in treated patients vs. placebo.

    Even more groundbreaking are the innovative elements of this trial. Patients were stratified by baseline tau PET, and the data released so far suggest that patients with intermediate tau benefited more than those with high tau tracer uptake. These results provide strong support for the biological approach that defines the 2018 NIA-AA research framework, and suggest that biomarker-based staging of disease will be an important aspect of future patient care.

    Second, treatment was halted and patients were switched to placebo once there was evidence of amyloid clearing on PET. A limited duration of treatment, akin to induction chemotherapy for cancer, could be a game-changer for patients, families, health systems, and third-party payers. This approach may also limit the time in which patients are exposed to the risk of ARIA, which is particularly critical given the reported deaths in the study.

    Of course we have only seen the top-line data, and there are still several questions that remain unanswered and will hopefully be presented at AAIC and in an upcoming publication:

    • Was the trial population representative in terms of racial and ethnic diversity?
    • How did risk of ARIA, and treatment response, interact with APOE4 genotype?
    • Did amyloid lowering modify fluid and PET measures of tau?
    • Is there clinical or biomarker evidence of true disease modification, as defined in the excellent recent article by Planche and Villain (2023)?

    The risk of severe clinical outcomes, including death, are small but real for donanemab and the overall class of anti-Aβ monoclonal antibodies. These risks require a careful discussion about risks and benefits, and a conservative approach to treatment eligibility and safety monitoring, as highlighted in the Appropriate Use Recommendations for aducanumab and lecanemab (Cummings et al., 2021, 2023).

    Overall, in my opinion there is no doubt at this point that potent anti-Aβ monoclonal antibodies provide clinically meaningful benefit but also pose non-negligible risks. The real question is whether patients and their physicians will have the opportunity to make an informed choice about whether these drugs are right for them. In the U.S. this will in large part depend on CMS coverage decisions, if and when lecanemab and donanemab receive full FDA approval based on evidence of clinical efficacy.

    Furthermore, Trailblazer-ALZ2 highlights the needs for CMS to cover baseline tau PET and longitudinal amyloid PET to optimize treatment decisions. The science in the field is moving forward fast—will CMS allow patients and their families to benefit?


    . Advocating for Demonstration of Disease Modification-Have We Been Approaching Clinical Trials in Early Alzheimer Disease Incorrectly?. JAMA Neurol. 2023 Jul 1;80(7):659-660. PubMed.

    . Aducanumab: Appropriate Use Recommendations. J Prev Alzheimers Dis. 2021;8(4):398-410. PubMed.

    . Lecanemab: Appropriate Use Recommendations. The Journal of Prevention of Alzheimer's Disease

  8. While a detailed peer-review publication will help put the top-line reported results in perspective, I suspect a lot of people are thinking the removal of amyloid plaques from MCI or mild AD patient brains slows decline cognitive decline over an 18-month period. The efficacy data mentioned in the press release are more encouraging than the very-difficult-to-interpret aducanumab Phase 3 trial data. Given the donanemab rationale is to remove plaques, the concept of terminating the intervention once the plaques are eliminated to a specified degree would seem an advantage over lecanemab. I suspect patients, families, and physicians will continue to struggle with the risk-benefit calculation.

    The donanemab data, in conjunction with the lecanemab data, seem consistent with the idea that plaque amyloid is not inert. Why removing plaque amyloid affects ongoing decline in MCI/mild AD patients is an interesting question, as the effects could arise through a number of different reasons, for example by altering peri-plaque microglial activation states.

    One thing I would love to see would be the data only from those who did not develop ARIA at any point in the trial, with the number of E4 carriers balanced between the treatment and placebo groups. That would help address the question of how much unblinding or any differences in APOE4 allele frequencies affected the spread between the groups.

  9. I am extremely encouraged by the top-line findings reported in Lilly’s announcement. The amyloid antibody therapy dramatically reduced PET measurements of amyloid plaques and significantly reduced decline in memory and thinking abilities and activities of daily living. As expected, the drug was also associated with amyloid-related imaging abnormalities (ARIAs), which can have occasional serious and sometimes life-threatening effects.

    This study provides another shot in the arm for the fight against Alzheimer’s disease. It provides additional hope for mildly impaired patients and their families, further support for the role of amyloid plaques in the development, treatment and potential prevention of this disease, and a chance to galvanize the successful development of Alzheimer’s disease-modifying and prevention therapies.

    The field is eager to learn more about the study results when they are presented at the Alzheimer’s Association International Conference (AAIC) this summer and reported in a peer-reviewed journal. For instance, there is interest in understanding the drug’s particular benefits versus risks in carriers and noncarriers of the APOE4 allele, the major genetic risk factor for Alzheimer’s disease, its effects on different Alzheimer’s disease biomarkers, and the extent to which its biomarker effects were associated with a clinical benefit.

    In addition to his other roles, Dr. Reiman is CEO of Banner Research, which operated three of the clinical trial cites involved in Trailblazer ALZ-2. He and his Alzheimer’s Prevention Initiative co-lead Trailblazer ALZ-3, a prevention trial of the same drug in cognitively unimpaired persons with blood test evidence of amyloid plaques in collaboration with Eli Lilly. He is an uncompensated consultant to Eli Lilly and a co-founder and advisor to ALZPath, a startup company involved in the development of Alzheimer’s disease blood tests that were not used in these studies.

  10. The donanemab Phase 3 outcomes—according to Lilly’s press release—are remarkably like its Phase 2 trial outcomes and the Clarity-AD lecanemab Phase 3 trial. The iADRS mean differences were approximately 3.26 in the targeted intermediate tau group and 2.88 in the combined intermediate/high tau group. By comparison, the iADRS difference in the Phase 2 trial of a similar intermediate tau population was 3.20. The ADAS-cog13 difference can be estimated from the iADRS as approximately 1.6 and compared to 1.8 in the Phase 2 trial; and to a 1.44 ADAS-cog14 difference in the lecanemab trial. 

    The press release says negative amyloid-PET scans were achieved in 71 percent by 12 months. This can be compared to 68 percent by 18 months in the Phase 2 trial, and 68 percent negative scans by 18 months with lecanemab. (The donanemab Phase 2 trial started out at 107.6 centiloids while lecanemab patients started at a much lower 77.9 centiloids.) Nevertheless, after 12 to 18 months they show about the same level of negative amyloid scans.

    In Phase 3, ARIA-E occurred in 24 percent, symptomatic in 6.1 percent, and serious in 1.6 percent. In Phase 2, it was 26.7 percent, 6.1 percent symptomatic, and 1.5 percent serious (Mintun et al., 2021). The lecanemab rates were half that of donanemab, 12.6 percent ARIA-E and 2.8 percent symptomatic (van Dyck et al., 2023). Could the higher incidence of ARIA-E with donanemab be due to this substantially greater brain amyloid at baseline acting as substrate/antigen?

    The only seemingly exciting effect with donanemab is an approximately -0.69 difference on the CDR-SB compared to -0.36 in Phase 2 and -0.45 with lecanemab. Even here, the three CDR-SB mean differences are within overlapping confidence intervals and are not substantially different from each other.

    It is reassuring to see a large Phase 3 trial that mirrors its smaller Phase 2 ancestor as this frequently does not happen. Identical, large pairs of Phase 3 trials–e.g., aducanumab and gantenerumab–often give discordant outcomes. But to heap "never before" superlatives on this Trailblazer-ALZ2 trial as though the trial results are a major paradigm shift directs attention away from the many uncertainties within the trial that the press release avoided. These include the number and effect of dropouts, protocol violations, site and country differences, COVID, and other potential biases such as functional unblinding of patients who developed ARIA that create high risks for bias.

    The elephant in this room is the three deaths out of about 850 people randomized to donanemab. A potential incidence of 0.35 percent with an upper limit of a 95 percent confidence interval approaching 1 percent is not “small,” as some would have it, and may not be acceptable under any circumstance. If this is a true incidence, then of 100,000 patients that some Wall Street analysts predict could be prescribed donanemab (or lecanemab right now) could result in 70 to 1,000 deaths. Although we surveil for ARIA, we don’t really know how to predict or treat it.


    . Donanemab in Early Alzheimer's Disease. N Engl J Med. 2021 May 6;384(18):1691-1704. Epub 2021 Mar 13 PubMed.

    . Lecanemab in Early Alzheimer's Disease. N Engl J Med. 2023 Jan 5;388(1):9-21. Epub 2022 Nov 29 PubMed.

  11. The results of the Trailblazer 2 study are unassailable. They confirm what was seen in the Trailblazer-ALZ study (Figure 1). These results are not unexpected but are the culmination of years of progress with monoclonal antibodies. The key to being able to see this progress relies on three insights:

    1. We must reject “The 0.06 Fallacy,” concluding that a study has demonstrated a null treatment effect if the primary endpoint has a p-value >0.05 (such as 0.06).
    2. Recognize that endpoints with low correlations provide independent evidence of effect. Expecting them to all achieve significance hides all but very large treatment signals.
    3. Combine results across cognitive, functional and global outcomes to calculate a better estimate of overall treatment effect than any one outcome alone.

    Solanezumab Expedition 3 had a 13 percent slowing over placebo in 2017, aducanumab (averaged across Engage and Emerge) had 19 percent slowing in 2019, and lecanemab had 30 percent slowing in 2022, leading up to the donanemab 2023 results (Figure 2) that show a 38 percent slowing (average of 36 percent, 35 percent, and 40 percent)—a nearly linear progression in effect over year of result. Without the insights noted above, solanezumab and aducanumab look like failures, and even the lecanemab results seem unexpected. But in reality, effects have been progressively larger with each MAB. This culminates in the lecanemab and donanemab results achieving significance on all 3 key endpoints—cognition, function and global—reflecting three different perspectives of disease: a direct measure of patient performance, a clinician’s assessment, and a study partner assessment. These accumulating effects over time, observed on multiple clinical outcomes, are consistent with disease modification.

    Disease-modifying effects result in permanent time savings, meaning that 18 months of treatment avoids approximately six months of progression (Dickson 2023). This time savings will be maintained even if treatment is stopped. Some would claim this is a victory for the amyloid hypothesis and for MABs targeting amyloid, but this is also a victory for the field of Alzheimer’s disease and for Alzheimer’s patients. Each piece of the puzzle is important, but we still need to add to this effect. Combination therapies will allow us to add on top of this nearly 40 percent slowing of progression.

    Figure 1: Results of Trailblazer ALZ study (Phase 2) with Trailblazer 2 results (Phase 3) represented with stars.

    Figure 2: Progress with Monoclonal Antibodies over the past 6 years.

  12. The donanemab Phase 3 randomized clinical trial results confirm what has recently, and consistently, emerged from high-clearance anti-amyloid immunotherapies RCTs in early AD: a significant, undisputable if small, clinical effect, and a small risk of serious life-threatening ARIAs (van Dyck et al., 2023; Villain et al., 2022). 

    The design of these RCTs in early AD has been constantly optimized since the first failure of the AN1792 anti-amyloid vaccine. It now definitely confirms that the high clearance of amyloid plaques in early and biologically-confirmed AD impacts clinical symptoms. Here, Eli Lilly has innovated by proposing a stratification by tau status, as measured by tau PET. This, together with a higher amyloid plaque clearance, might be responsible for the apparent larger numerically favorable effect on cognitive/functional decline, compared to the other high-clearance anti-amyloid immunotherapies (~0.68 points 18-month difference on CDR-SB vs. 0.45 for the lecanemab Phase 3 Clarity AD trial and 0.39 for the aducanumab Phase 3 Emerge trial). Still, as underlined recently in a previous Alzforum comment for lecanemab, “The 18-month [donanemab] data provided so far do not demonstrate an important impact on the global disease course, do not reach the [minimal clinically important differences], nor outperform the previously established effect of AChEIs.” Even if not disclosed by the sponsor in any recent high-clearance anti-amyloid immunotherapies RCT, estimations of the Cohen’s d of the lecanemab Clarity-AD RCT was ~0.20, i.e., a small size effect.

    More worrying, beyond questions of practicability, this design is probably the most optimal and thoughtful one could currently think of to evidence an effect of anti-amyloid immunotherapies. Still, it failed to generate a higher size effect. Thus, we may be observing here the maximum efficacy of anti-amyloid immunotherapies as a single therapy in any symptomatic stage of AD.

    Interestingly, even if not disclosed in the press release, calculations can be performed to estimate the effect in the intermediate and high tau subgroups (see Google doc). The absolute effect on CDR-SB in these two groups looks quite similar (0.68 vs. 0.72 points respectively), even if the corresponding placebo groups worsened differently during the 18-months duration of the RCT (1.9 vs. 3.5 points on the CDR-SB respectively). Does it indicate that the drug works less well in the high tau subgroup? Or that the anti-amyloid effect is similar whatever the biological stage of the disease, hence aligned with a symptomatic effect of the drug?

    From a mechanistic point of view, I would respectfully beg to differ with Eric Siemers and Dennis Selkoe when they write, “It can no longer be said that we have no treatments to slow the inexorable progression of AD … The 'amyloid hypothesis' of AD ([is] now no longer a hypothesis).” I would like to submit that RCTs do not demonstrate any mechanistic disease hypotheses. The effect of anti-amyloid immunotherapies could be symptomatic-like, mediated by the alleviation of Aβ synaptic toxicity, but unrelated to the amyloid cascade hypothesis, as we recently detailed with Vincent Planche in JAMA Neurology (Planche and Villain, 2023). Unfortunately, as underlined by Gil Rabinovici in his comment, with currently disclosed data (and previously from the other anti-amyloid immunotherapies RCTs), there is no clinical or biomarker evidence of true disease modification.

    Regarding safety, the three deaths in the core trial add to the three deaths in the OLE of the lecanemab Clarity AD RCT (Reish et al., 2023; Sabbagh et al., 2023; Solopova et al., 2023). The rate of serious ARIA seems relatively high (1.6 percent), higher than in the lecanemab trial (1.1 percent, 10/898, Eisai, personal communication) and than in any of the high-clearance anti-amyloid immunotherapy Phase 2 or 3 RCT, as we highlighted in our recent meta-analysis (Villain et al., 2022) and updated with the Clarity AD, Graduate 1 and 2, and Trailblazer-ALZ2 data. If not fatal, serious ARIAs can sometimes have long-term sequelae (Villain et al., 2022). We should not consider ARIAs as 100 percent benign and manageable side effects.

    Overall, these results confirm the efficacy of high-clearance anti-amyloid immunotherapies RCTs in early AD. The size effect remains small after 18 months, and currently available data do not allow us to be sure that the benefits will increase over time. On the other hand, risks are not negligible, with fatalities now indisputably related to these drugs and long-term sequelae of serious ARIAs. In my opinion, the risk/benefit ratio of high-clearance anti-amyloid immunotherapies remains questionable. The demonstration of actual disease modification—e.g., by using a delayed-start RCT design and a prespecified mediation analysis of biomarkers, as proposed in Planche and Villain, 2023as well as identification of subgroups with better risk/benefit ratios, use of combination therapies, and improvement of practicability are indispensable for large-scale use of these drugs.


    . Advocating for Demonstration of Disease Modification-Have We Been Approaching Clinical Trials in Early Alzheimer Disease Incorrectly?. JAMA Neurol. 2023 Jul 1;80(7):659-660. PubMed.

    . Multiple Cerebral Hemorrhages in a Patient Receiving Lecanemab and Treated with t-PA for Stroke. N Engl J Med. 2023 Jan 4; PubMed.

    . Lecanemab Phase 3 Clarity Ad Trial : Aria With the Use of Antiplatelets or Anticoagulants in Early Alzheimer's Disease. Meeting Abstract, International Conference on Alzheimer’s and Parkinson’s Diseases.

    . Fatal Iatrogenic Cerebral Amyloid-Related Encephalitis in a patient treated with lecanemab for Alzheimers disease: neuroimaging and neuropathology. 2023 Apr 29 10.1101/2023.04.26.23289061 (version 1) medRxiv.

    . Lecanemab in Early Alzheimer's Disease. N Engl J Med. 2023 Jan 5;388(1):9-21. Epub 2022 Nov 29 PubMed.

    . High-clearance anti-amyloid immunotherapies in Alzheimer's disease. Part 1: Meta-analysis and review of efficacy and safety data, and medico-economical aspects. Rev Neurol (Paris). 2022 Dec;178(10):1011-1030. Epub 2022 Sep 29 PubMed.

  13. The donanemab Phase 3 results are an exciting development for the Alzheimer's research community and for patients. These findings, alongside the aducanumab and especially the lecanemab results, should increase confidence in anti-amyloid therapy as a viable treatment for Alzheimer's disease.

    Determining the generalizability of the treatment effects will be a critical question. Selection of participants based on intermediate tau was an innovative strategy designed to select individuals who were progressed enough, but not too progressed, to benefit from treatment. 

    The finding that patients with intermediate tau burden had greater benefits than those with higher tau burden suggests that this strategy was successful, but it is unclear how this will translate to clinical practice, since the amount of tau burden in Alzheimer's patients is extremely heterogeneous, even among those with a given degree of impairment. In MCI or cognitively normal individuals, intermediate tau burden is less common. Will these patients just need to be treated or followed longer to see the benefit of the drug, or will they respond differently?

    There are also important questions about the safety of this drug.  Given the frequency of ARIA in this trial, it will be critical to determine whether there are patient characteristics that predict the likelihood of ARIA severity.

  14. The Phase 3 donanemab results are very encouraging for our field, providing further evidence that clearing aggregated species of amyloid from the brain at an early clinical stage of AD can slow cognitive and functional decline.

    A particularly valuable innovation of Trailblazer-ALZ2 was the full incorporation of tau PET to determine the disease stage at which participants were most likely to respond. Lilly, following the Goldilocks principle, required that the tau PET signal be “not too hot and not too cold, but just right.” By limiting the primary analysis to these ideal intermediate tau subjects, they maximized their chance of success. However, a population more comparable to those in the lecanemab and aducanumab trials—and, more importantly, to patients in the clinic—is the overall sample (i.e., the combined intermediate and high tau population), for whom decline was slowed by 22 percent for the primary outcome (iADRS, compared to 35 percent in the intermediate tau population) and by 19-29 percent for other outcomes. The large discrepancy between the numerical effects in the intermediate tau population and the overall population raises the question of whether the high tau subjects did not benefit, or simply benefitted less. This will be important information for the full presentation of data.

    In hindsight, Lilly could have benefitted our field even further by the inclusion of participants with low/normal tau as well. These individuals were fully excluded as less likely to benefit. However, their inclusion in the study for secondary analyses would have provided valuable information for the clinical setting. The study is well-positioned to examine the predictive (theragnostic) value of this biomarker. And those data will likely be relevant for the other antibodies as well. I am certainly looking forward to seeing the complete results at the AAIC meeting in Amsterdam.

  15. The trial will likely help thousands of patients. It will also help dozens of future trials, because the donanemab data supports amyloid as a therapeutic biomarker. Two years ago, this idea was controversial and polarizing. Given the data across the amyloid mAbs, the wisdom of amyloid surrogate biomarkers is now undeniable. Drugs that markedly decrease amyloid on PET work. Drugs that don't, don't.

    Obviously this theory is limited to IGG1 antibodies against amyloid. Outside this class, with tau biomarkers, we can do the same: get go/no-go decisions quickly. As a field, we can avoid doomed shots on goal and fund second-generation amyloid mAbs to reduce ARIA and increase effectiveness.

  16. This announcement represents another huge leap in our field. Coming shortly after the successful Clarity AD trial of lecanemab, it convincingly furthers our previous excitement that we have finally arrived at a therapeutic breakthrough strategy. These are exciting times to be an AD scientist and a cognitive neurologist.

    To date we have witnessed two monoclonal antibodies that successfully reached FDA provisional accelerated approval. This raises the question of how the two compare? It is a hard question to answer without having seen the full clinical trial safety and efficacy data. Further, it is hard to compare donanemab to the two MABs that have received accelerated approval to date without a true head-to-head comparison of efficacy. With what we know to date, it does seem that donanemab is probably at least just as effective as lecanemab and more effective than aducanumab. The latter is supported by the head-to-head aducanumab vs. donanemab Trailbalzer ALZ4 trial results presented at CTAD last year, which showed greater and more rapid removal of amyloid by donanemab relative to aducanumab, and yet comparable safety profiles. Such data for lecanemab does not exist.

    In regard to donanemab’s safety profile, ARIA-E and ARIA-H are known side effects of most of the drugs of this class. In some individuals, these could be so serious as to contribute to long-term disability or, unfortunately, death. These cases are luckily extremely rare, but the risk is real. The recently published appropriate use recommendations for lecanemab (and those for aducanumab) appropriately call for extensive pretreatment patient and family counseling that covers the treatment requirements, therapeutic expectations, as well as the risks and benefits from MAB therapies. Most physicians might also be inclined to obtain APOE genotyping as APOE4 carriers are at increased risk of ARIA. The pros and cons of starting anti-amyloid MAB therapy are critical components of the provider-patient pre-treatment counseling session.

    While we can convincingly say that we are making history, our work is not finished. Alzheimer’s disease continues to progress despite the anti-amyloid MAB therapeutic advances as other disease mechanisms remain unchecked. The next phase of clinical trials will be even more complicated as combinatorial trial designs are called for to address the tauopathy, immune system and bioenergetic derangements, and the vascular aspects of Alzheimer’s. Yet, we should celebrate this moment of now several therapeutics with consistent and convincing evidence for disease modification. 

  17. It's great to see this. With these data, the lecanemab data, and the aducanumab data, we can see what treatments need to do to achieve efficacy. It looks also that earlier is better (and it may also be safer). Much practical work is required to get these drugs into clinical practice, but the route ahead is clear.

    Great also to see the Phase 2 data on tau antisense looking encouraging (Mummery et al., 2023). It really seems like 40 years of research since Glenner and Wong and Masters and colleagues are now poised to reap clinical benefit.


    . Tau-targeting antisense oligonucleotide MAPTRx in mild Alzheimer's disease: a phase 1b, randomized, placebo-controlled trial. Nat Med. 2023 Jun;29(6):1437-1447. Epub 2023 Apr 24 PubMed. Correction.

  18. As understandable as it is to chalk up the results of the latest batch of anti-Aβ antibodies to a slowdown of disease progression, that is still premature in view of what we see with Aβ removal in transgenic mice. There, treatment with Aβ antibodies has been reported to have rapid effects on cognitive performance (Kotilinek et al., 2002Dodart et al., 2002). Since these mice do not have the overt neurodegeneration seen in AD, the short-term cognitive effects are actually more of the cognition-enhancer type, i.e., offsetting an underlying dynamic impairment. The most plausible structural correlate to that is the rapidly reversible synaptic impairment by Aβ causing synaptic density to settle on a lower equilibrium. Aβ removal is then best understood as relaxing this impairment acutely.

    Plaque removal in patients evidently takes months. If the type of impairment seen in mice occurs in humans, as well, and is correlated with plaque density or possibly occurs faster than that, as seems to happen in mice, then the relaxation may in fact only look like progression modification until plaques are finally cleared or neutralized, and may thereafter present as merely an offset typical for a cognition enhancer.

    The observation that, above certain levels of tau pathology, Aβ removal is no longer effective on clinical measures could be due to the relaxation by Aβ removal being balanced out by faster-growing tau pathology in later stages (as expected from prion behavior), which is the true mechanism of neurodegenerative progression.

    Biomarkers will be less mechanistically interpretable than some want to believe, as the origin of such markers is not clear-cut. If these markers mainly reflect the shedding of intra-synaptic content from enhanced steady-state turnover of synapses as a result of Aβ effects thereon, then they report essentially acutely reversible synaptic density effects as the material substrate of the cognition signal and are not to be understood as progression markers.

    It will require longer clinical observation periods substantially beyond the kinetics of unambiguous plaque removal to sort this out. It is therefore premature to declare that the Aβ hypothesis is no longer a hypothesis. It would be a disservice to the field to fall into another one of the interpretation traps it has suffered from for decades.

    There may be a time when synaptic turnover promotes the propagation of tau pathology when the neurodegenerative activity of tau pathology is too low to compete, but as we can see in pure tauopathies, one can expect that there is a time when Aβ as a player becomes fully irrelevant once prion-type spread of tau is self-sustaining.


    . Reversible memory loss in a mouse transgenic model of Alzheimer's disease. J Neurosci. 2002 Aug 1;22(15):6331-5. PubMed.

    . Immunization reverses memory deficits without reducing brain Abeta burden in Alzheimer's disease model. Nat Neurosci. 2002 May;5(5):452-7. PubMed.

  19. It is great to see some positive news with anti-amyloid antibodies therapy. Of course this is only a news release, and full publication of the data is needed for proper evaluation. The main concern remaining is that with such high prevalence of ARIA in the treatment group and not in the placebo group, can these trials can really be considered blinded? And how much does this "partial unblinding" affect relatively subjective cognitive readouts used in these trials? It will be interesting to see if there is a correlation between groups with higher ARIA prevalence (APOE4 positive?) and cognitive scores.

  20. Donanemab was dosed at 700 mg monthly for the first three months, then 1,400 mg for up to 18 months. Is this dosage concerning?

    For example, Evusheld (formerly AZD7442) is a long-acting antibody (LAAB) combination for the prevention of symptomatic COVID-19. The recommended dosage is 300 mg of Evusheld administered as two separate 1.5 mL, sequential injections of 150 mg of tixagevimab and then 150 mg of cilgavimab. For ACTEMRA® (tocilizumab), the recommended single IV dosage for COVID-19 patients is 8 mg per kg. If clinical signs or symptoms worsen or do not improve after the first dose, one additional infusion of ACTEMRA may be administered at least eight hours after the initial infusion. In COVID-19, doses exceeding 800 mg per infusion are not recommended.

    Did the control group get a nonspecific antibody as placebo? If not, how do we know that the outcomes (both ARIA and the benefits) are not influenced by placebo effect?

Make a Comment

To make a comment you must login or register.


Therapeutics Citations

  1. Donanemab
  2. Aduhelm
  3. Leqembi
  4. Gantenerumab

News Citations

  1. Drilling Down into the CMS Aduhelm Decision—A Primer
  2. Donanemab Confirms: Clearing Plaques Slows Decline—By a Bit
  3. What Happens After Amyloid Plaque Removal? Who Benefits Most?
  4. Next Goals for Immunotherapy: Make It Safer, Less of a Hassle

External Citations

  1. Eli Lilly press release

Further Reading