On June 10th, the FDA convened an advisory panel to discuss the status of donanemab, drug maker Eli Lilly’s latest beta-amyloid targeting monoclonal antibody for Alzheimer’s disease. The hours-long meeting included slides early on specifying that discussion points would include whether the advisory committee thought donanemab was effective, dosing considerations, and whether the benefits of the drug outweighed the risks. At the end of the meeting, the panel voted unanimously 11-0 that donanemab is effective at slowing Alzheimer’s in early-stage disease and that the benefits outweigh the risks. While this decision alone does not represent approval of the drug, the FDA is expected to act on the recommendation of its advisory panel in the coming months.
Presenting the evidence
In March the FDA had originally announced it would put off a decision regarding approval to further study and ask for additional information regarding trial data. Arguably the biggest trial in question, the TRAILBLAZER-ALZ 2 trial, came out last year. It concluded that donanemab statistically significantly slowed clinical progression at 76 weeks in those with baseline low/medium tau and in the combined low/medium and high tau pathology population.
The advisory panel was convened in June to discuss a number of issues, including the high rates of side effects (24% brain swelling and 31% brain bleeding in the donanemab groups). Patients in the donanemab group were only treated until amyloid plaque levels reached a pre-specified set point at which point treatment was stopped, representing the “limited-duration dosing regimen” referred to. Baseline tau levels were used to stratify patients.
In the trial results, clinical progression of Alzheimer’s was slowed in those in the low/medium tau group and in the combined low/medium and high tau group. But the “high” tau group alone did not have an improved integrated Alzheimer Disease Rating Scale score (iADRS), the primary outcome measure in the trial (though participants did show improvement with a secondary outcome measure, the clinical dementia rating scale sum of boxes or CDR-SB score). Patients with higher tau burdens would be more at risk for developing Alzheimer’s to begin with. If the drug were to be approved, questions surrounding who to offer it to and how to offer them the drug need to be addressed.
The meeting included presentations on efficacy and safety results by officials at Eli Lilly, opportunities for clarifying questions from the committee, and an open public hearing with speakers unaffiliated with the FDA having the opportunity to chime in. At the end of the session, each of the 11 advisory committee members had an opportunity to share their thoughts briefly when discussing their votes.
Reservations
Although the votes of affirmation for donanemab were unanimous, not all committee members shared an unbounded enthusiasm. For example, Dr. Nilufer Ertekin-Taner from the Mayo Clinic gave her vote of affirmation with the understanding that the data aren’t clear for all populations of interest, including those with two copies of the APOE4 gene (a strong Alzheimer’s risk factor), and in African American and Latin American patients.
If approved, donanemab would join fellow beta-amyloid targeting monoclonal antibody lecanemab (Leqembi™) as the third FDA approved disease-modifying Alzheimer’s treatment while seeking to avoid the fate of the first approved therapy, aducanumab (Aduhelm™), which has since been pulled from the market by maker Biogen due to difficulties generating enthusiasm and obtaining insurance coverage for their drug. Both lecanemab and aducanumab were approved under the FDA’s accelerated approval pathway, where drug approval is based not on clinical efficacy but on showing efficacy in achieving a surrogate endpoint. In these cases, the endpoint was a reduction of beta-amyloid plaques. Critically, however, the link between beta-amyloid and cognitive decline is far from certain.
The amyloid hypothesis and the tau hypothesis
The hypothesis that beta-amyloid is linked to cognition, or the amyloid hypothesis, posits that the accumulation of beta-amyloid fibrils under pathological conditions (such as with Alzheimer’s disease) leads to the development of plaques that are neurotoxic and lead to cognitive decline. Initial excitement for this hypothesis was linked to the fact that individuals with Down syndrome who have three copies of the amyloid precursor protein gene instead of the regular two have a higher rate of developing Alzheimer’s Disease. This protein is involved in the generation of beta-amyloid peptides.
However, not everyone with Down syndrome who develops beta-amyloid plaques has gone on to develop Alzheimer’s disease. Additionally, in other experiments using genetically modified mice with increased beta-amyloid production, nerve cell death was not observed as anticipated. Also, with the more recent advent of imaging for amyloid plaques in humans, many patients without known cognitive decline have extensive amyloid deposits, while many patients with Alzheimer’s Disease have few amyloid deposits. There are however mutations in amyloid genes identified which are linked to early onset familial Alzheimer’s, though these represent no more than 10% of Alzheimer’s cases.
Altogether, though the initial enthusiasm for the amyloid hypothesis was high, there are holes. Another protein called tau has also been implicated in the development of Alzheimer’s disease. Specifically, when tau is hyperphosphorylated (termed neurofibrillary tangles if they are formed in neuronal cell bodies), it is strongly correlated with the development of cognitive symptoms. Imaging techniques to study the patterns of tau pathology as well as post-mortem studies of patients showed much better correlation between tau pathology and cognitive status. The tau hypothesis has today garnered more support for direct correlation with cognitive decline in Alzheimer’s disease.
Regardless, interpretation of trial data for all three amyloid targeting drugs purported to show not just clearance of beta-amyloid but also less cognitive decline compared to placebo. These observations raised many questions, including the extent to which beta-amyloid is directly related to cognition, the extent to which modulation of beta-amyloid modulates tau, and – crucially — the extent to which the trials showing benefit were run well and had their data interpreted correctly.
What happened with aducanumab?
In the case of aducanumab, only one of the phase 3 trials showed slight benefit, but only in males over the age of 70 with a copy of the APOE4 gene. Though the other trial did not show benefit, a post-hoc analysis that studied only those receiving at least 14 treatment sessions with a “high” dose of the drug did show some benefit.
Although an FDA advisory committee voted heavily against recommending aducanumab via the traditional pathway based on clinical efficacy, the FDA switched its review to that of the accelerated approval pathway based on the surrogate endpoint that showing removal of beta-amyloid on its own was good enough and then approved the drug. This was done without consultation with the advisory committee, leading 3 members of the advisory committee to resign. A congressional probe later found atypical collaboration between Biogen and the FDA. Biogen eventually discontinued development of aducanumab in January of this year.
The devils in the details
The “Lecanemab in early Alzheimer’s Disease” trial was published in late 2022 and concluded that lecanemab reduced markers of amyloid in early Alzheimer’s disease and resulted in moderately less decline on measures of cognition and function than placebo at 18 months. Not widely publicized was that per the trial data, lecanemab did not slow cognitive decline in women, and in fact enhanced decline in those with 2 APOE4 genes. Interestingly, trial data for aducanumab, lecanemab, and donanemab do show reduction in phosphorylated tau levels, keeping open the question of a threshold level of amyloid which influences tau pathology.
Another issue plaguing the trials had been the nagging question of functional unblinding which may be contributing to the positive results. In phase 3 trials of amyloid-targeting drugs, 21% to 44% of patients on active treatment developed brain bleeding or swelling, given the acronym ARIA for amyloid-related imaging abnormalities. A recent analysis postulated that the development of ARIA would have triggered protocol changes that may have unblinded patients and those around them to their treatment. This same analysis found a positive correlation between the incidence of ARIA-E (E for edema; brain swelling) and reported efficacy of the drugs in question.
An additional point of criticism was how a sensitivity analysis aimed to account for functional unblinding in the TRAILBLAZER ALZ-2 Trial, in which patients with ARIA-E had their post ARIA data points replaced with extrapolated data points from the average course of patients receiving donanemab who did not develop ARIA-E. It is also well known that ARIA are more likely to occur in those with the APOE4 gene mutation and, as previously mentioned, that APOE4 is more likely to lead to Alzheimer’s. This would have effectively removed individuals most at risk for developing Alzheimer’s disease from the donanemab group.
Finally, while levels of statistical significance have been hit in the aforementioned trials, an additional point of contention is when these changes become clinically meaningful. It’s been suggested that the threshold for a meaningful difference is a within-patient change of CDR-SB score of 1. With this in mind, neither donanemab nor lecanemab during their respective trial periods improved scores this much. However, the CDR-SB scale may also not capture one’s cognition very well.
So, is it worth it?
Now that the FDA advisory committee panel has unanimously voted to give donanemab the green light, it seems close to a sure bet that donanemab will be approved in the next few months. However, with the above reservations in mind, there remains uncertainty. To drill this down properly and instill public confidence, more transparency is needed. This includes more data on within-trial subject level records. Without increased transparency, it’s hard to ignore the reservations regarding donanemab and its predecessors.
If, as is the case with donanemab, patients with higher baseline tau pathology do not see a benefit in their iADRS, then those most at risk of developing cognitive decline stand to benefit the least. A big question would be if it is worth it to pre-emptively suggest a drug like donanemab to patients with suspected early Alzheimer’s disease before symptoms progress.
With the reservations discussed above, the side effect profile, and the tenuous nature of the amyloid hypothesis on which the premise of clinical efficacy is based, it should not be suggested to patients at this time. Nevertheless, if already approved, it’s also hard for clinicians to put a full stop on offering the drug to patients asking for it. In this case, as long as clinicians are open about the uncertainties, the remainder of the decision-making should fall to the patient.
Saami Zakaria Saami Zakaria is a neurology resident physician in New York City who has long been interested in the intersection of the brain and the machine and how it can be responsibly shaped. Additionally, he is enthusiastic about appraising evidence for clinical trials and the various methodologies used in medicine to ensure our practices are truly patient-centric. Many of his writings can be found on his substack. He can also be found on X (formerly known as Twitter) as @SaamiJZ View all posts
