Alzheimer’s disease (AD) is a complex neurodegenerative disease that causes progressive dementia. Patients have impaired memory and as the disease progresses they have increasing impairment of all cognitive function. Life expectancy is typically 3-11 years following diagnosis. Worldwide more than 25 million people suffer from AD, with a prevalence of 4.4% in those over 65. In the US for those over 70 the prevalence is 9.7%.
At present there are no treatments which are generally considered to be disease modifying, and nothing approaching a cure. Existing treatments are considered symptomatic in that they reduce the severity of impairment but do not prevent progression. It is clear there is a great need for an effective treatment for AD, and this has apparently driven the decision by the FDA (discussed by Scott Gavura last month) to give “fast track” approval to a new drug, aducanumab (brand name Aduhelm), based on preliminary evidence. But the decision has proven to be highly controversial.
In the FDA announcement they report:
Researchers evaluated Aduhelm’s efficacy in three separate studies representing a total of 3,482 patients. The studies consisted of double-blind, randomized, placebo-controlled dose-ranging studies in patients with Alzheimer’s disease. Patients receiving the treatment had significant dose-and time-dependent reduction of amyloid beta plaque, while patients in the control arm of the studies had no reduction of amyloid beta plaque.
They also explain that, given the great need, they allowed for approval based upon a biomarker only, without clinical data. They argue this is because there is a plausible connection between the biomarker (reduction in amyloid beta plaque) and clinical disease. Further, they now require the company, Biogen, to conduct a clinical trial to prove benefit. If the trial does not show clinical benefit, then the FDA will withdraw approval.
That all sounds reasonable, or at least they give a justifiable reason for their decision. However, they leave out many details, such as the fact that out of their 11 member advisory panel of neurologists, 10 voted not to approve aducanumab and one abstained. There are several reasons offered for their advice to reject the application. One is the fact that of the two clinical trials, one did not show a significant reduction in amyloid plaque while the other did. The results were mixed at best.
The majority of FDA reviewers (outside the neurology advisory panel) did recommend approval of the drug (which is why it was approved). This included the pharmacological review and the clinical review. However, the statistical review recommended denial, because the statistical results were weak. One of the members of the advisory panel, Dr. Aaron Kesselheim, who resigned in protest after the approval, said:
If the FDA allows companies to get drugs approved on the basis of trials that are stopped early, trials that are reanalyzed, that sets a precedent.
An independent statistical analysis published in March in JAMA also took issue with the statistical analysis, concluding that the results were not statistically significant. Therefore, we have two clinical trials, one negative, the other mixed with questionable statistical significance at best. That seems like a low bar, especially if we are relying entirely on a marker for AD rather than net clinical outcomes.
Further, use of that biomarker, amyloid beta plaque, has also been criticized. This has been one of the challenging aspects of AD science – researchers find lots of biomarkers that correlate with disease progression. However, as is always the challenge with biomarkers, it is unclear if this is contributing to progression of the disease or is just a consequence of it. Specifically, we cannot assume that simply reducing the presence of the amyloid beta plaque will reduce progression of the disease.
As other critical experts have pointed out, there have been prior monoclonal antibody treatments which demonstrably reduce amyloid beta plaque in AD but which failed to produce any clinical improvement, specifically bapineuzumab and solanezumab (monoclonal antibody drug names all end in “mab”). While there is still debate on this issue, the evidence so far does not clearly support reduction in amyloid beta plaque as a predictive biomarker for clinical improvement.
Scientifically, therefore, aducanumab may not even work as advertised in terms of meaningful reduction in amyloid beta plaque in AD, and even if it does this may not result in clinical improvement. But there are some pragmatic considerations as well. One is that the company announced that the cost of the drug will be $56,000 per year. Monoclonal antibodies are not chemicals, as the name implies they are specifically created antibodies with a therapeutic target. They are expensive to make, hence the high price tag just for the drug (although critics claim the company is including a hefty profit in that price tag) but also are expensive to give. They are delivered through intravenous infusions. A reasonable estimate for total cost, therefore, is about double the drug cost, or over $100k per year.
Given the prevalence of AD this is a hefty price tag for a treatment based on such thin evidence. Further, there are likely to be inequities in availability and reimbursement.
Another concern (one we often point to with questionable medical treatments) is that treatment with aducanumab, because it is FDA approved, may displace other experimental treatments. Patients in clinical trials (and who may be getting a placebo) may want to pull in order to get aducanumab, or may never enter a clinical trial in the first place. If aducanumab truly worked, this would be OK, it would become the standard of care and the baseline with which further research would be compared. But given the dim view that many experts have for this treatment, the fear it the only net effect, in addition to wasting a lot of health care dollars, will be to slow further AD clinical research.
The one good thing about the aducanumab controversy is that the public debate on the topic has been nuanced and well-informed. The principles I have outlined above are all important concepts that are broadly applicable. We invoke them frequently in many contexts at SBM: biomarkers may not predict clinical outcome, statistical rigor is critical, preliminary results are poor predictors of ultimate efficacy, and diverting resources based on false hope can cause much harm. Let’s hope these lessons are learned well. They can be applied to just about the entire supplement industry and much of the alternative medicine industry, for example.