Does the FDA‐approved Alzheimer drug aducanumab have a place in the Australian pharmacopoeia?

Aducanumab, a controversial Alzheimer disease therapeutic with questionable efficacy, is under Therapeutic Goods Administration review

Dementia, the second leading cause of death in Australia,1 affects nearly half a million people.2 It is the single greatest cause of disability in Australians aged 65 years and over and cost more than $15 billion in 2018.2 Alzheimer disease (AD) accounts for about two‐thirds of cases.3 The need for disease‐modifying treatment for AD is clearly one of the highest priorities for current medical research.

The dominant pathological model for AD has been the amyloid cascade hypothesis,4 which posits that the initial causative event in AD is deposition of amyloid‐β. While a substantial proportion of research funding and therapeutics trials has focused on aspects of the amyloid cascade hypothesis over recent years, it remains unclear whether amyloid‐β is toxic or merely a common epiphenomenon.5,6 Over 30 phase 3 trials of anti‐amyloid therapeutics have failed to meet primary endpoints, raising questions over whether this is indeed the correct therapeutic target.5

Aducanumab is a human monoclonal antibody that selectively reacts with amyloid‐β, robustly reducing amyloid plaque.7 Phase 3 studies were prematurely terminated in 2019 following a futility analysis. A subsequent post hoc reanalysis of additional data led to United States Food and Drug Administration approval on 7 June 2021 via the Accelerated Approval Program, not on the basis of demonstrated clinical efficacy, but on the unproven presumption that reduction of amyloid plaques is “likely to predict clinical benefit”.8 This approval took place despite the recommendations of 10 of the 11 members of the advisory committee (one abstained).9 Three FDA advisors subsequently resigned in protest10 and the FDA commissioner requested an independent inquiry into the process that led to approval of the drug.11

Proponents of the approval, including the FDA and the Alzheimer’s Association, defended the decision by arguing that AD is a serious disease where there is unmet need, and, if there is a chance that a drug could provide a benefit, patients should be allowed access while more data are collected. The FDA stated that aducanumab has “expectation of clinical benefit despite some residual uncertainty”.8 However, as discussed below, the evidence thus far does not provide a compelling case for clinical benefit. Treatment is costly and side effects are common, with unclear long term consequences. Proponents also argue approval may spur further investment into AD drug development. Yet promoting poorly performing medicines by lowering regulatory standards is an imprudent strategy for expediting the development of therapeutics. Having an approved drug in the community will likely hinder design and recruitment to future clinical trials of drugs that may actually be effective.

There were two phase 3 randomised controlled trials of low or high dose aducanumab versus placebo, named EMERGE (https://clinicaltrials.gov/ct2/show/
NCT02484547) and ENGAGE (https://clinicaltrials.gov/ct2/show/NCT02477800). Statistically significant results favouring the drug with a small effect size were found at 78 weeks for participants in EMERGE at high but not low dose. No statistically significant benefits were seen for the primary or any secondary endpoints on the other phase 3 study, ENGAGE.7 A negative trial clearly raises questions about reliability of results,12 and whether the positive findings in EMERGE could be due to chance alone.13 The discordance between EMERGE and ENGAGE is potentially explained by a type I error given the large number of negative phase 3 trials of anti‐amyloid therapeutics.14 It is possible that the different findings for EMERGE were due to a worsening of the placebo group that occurred only at the final time point in that study. The treatment groups progressed to the same degree in both studies, and indeed when the results were combined from both trials, no statistically significant benefit for high dose aducanumab was observed.13 Additionally, EMERGE and ENGAGE may have been unblinded by investigator, subject and caregiver knowledge of amyloid‐related imaging abnormalities, which were seen far more often in the treatment arm.15

Even if the findings from EMERGE are true, the magnitude of benefit is unlikely to be clinically meaningful. On the primary endpoint, the Clinical Dementia Rating sum of boxes (CDR‐SB), a difference of − 0.39 was found, favouring the treatment arm.7 Findings on secondary endpoints also favoured aducanumab, with a 1.7 point difference on the Alzheimer's Disease Cooperative Study scale for activities of daily living in mild cognitive impairment, 0.6 points on the Mini‐Mental State Examination (MMSE), and − 1.4 points on the Alzheimer's Disease Assessment Scale–Cognitive Subscale (13‐item) (ADAS‐Cog13).7 Although these post hoc analyses were statistically significant, the magnitude of benefit was much smaller than the minimum clinically important difference on the CDR‐SB, generally considered to be 1–2 points, or the MMSE at 1−3 points.16 The MMSE and ADAS‐Cog13 changes seen for aducanumab in ENGAGE were about half as large as the benefit seen with the cholinesterase inhibitor donepezil over a much shorter time frame (26 v 78 weeks).17

There are no data from clinical trials of aducanumab, or any other source, to indicate that lowering amyloid‐ β has clear clinically significant benefit.12 Data from other trials to date suggest that amyloid lowering does not have an appreciable effect on cognition,18 and in some studies, patients treated with amyloid‐ β‐lowering therapies were cognitively worse.19 A recent Bayesian analysis provided “strong evidence of absence of a therapeutic effect” for anti‐amyloid therapies, although aducanumab data were not included.20 Under the Accelerated Approval Program, the manufacturer has 9 years to run a phase 4 trial to demonstrate aducanumab’s efficacy. In the meantime, those who receive this drug will be exposed to side effects.

The most common adverse event in ENGAGE and EMERGE was amyloid‐related imaging abnormality (ARIA). ARIA includes signal hyperintensities on magnetic resonance imaging (MRI) fluid inversion recovery sequences, thought to represent vasogenic oedema or sulcal effusion (ARIA‐E), and signal hypointensities on gradient recalled echo T2*‐weighted imaging, thought to represent haemosiderin deposition such as microhaemorrhage or superficial siderosis (ARIA‐H).21 ARIA‐E or ARIA‐H was seen in 41% of subjects on 10 mg/kg compared with 10% of those on placebo in pooled results from ENGAGE and EMERGE.7 Clinical symptoms of ARIA, such as headache, delirium, dizziness and visual disturbance, were present in 24% of those on high dose and remained unresolved in 12% over the study period.7 While serious symptoms were only seen in 0.3%, many of these subjects required hospital admission and treatment with steroids or antiepileptics.22 One of the few detailed reports of the course and management of severe ARIA describes a case of malignant hypertension and epileptiform activity requiring intensive care admission.23

The question of whether ARIA might have long term consequences is critical. Spontaneous ARIA in PSEN1 mutation carriers is associated with poor clinical outcome.24 ARIA‐like MRI changes are also seen in cerebral amyloid angiopathy‐related inflammation,21 which is associated with anti‐amyloid‐β antibodies, suggesting the possibility of a shared pathophysiology with ARIA.25 In a non‐human primate model of cerebral amyloid angiopathy, histopathological examination revealed reactive astrocytosis, microgliosis, and damage to axons and myelin, raising the possibility of similar changes in ARIA‐E.26 Microhaemorrhage seen in settings other than ARIA is a potential risk factor for haemorrhagic stroke, and is associated with increased incidence of future cognitive impairment, functional dependence and death.27

An independent analysis concluded that existing evidence is insufficient to determine the net health benefit of aducanumab.13 In the US, aducanumab costs $28 200 (currently about AU$40 000; reduced from US$56 000 in January 2022) per patient per year, many times higher than the base‐case cost‐effectiveness threshold price of US$2560 to $8290.13 Assuming half of Australian AD cases are mild and two‐thirds have AD pathology, about 150 000 Australians will be eligible for aducanumab. If 10% of those eligible for treatment received the drug, the annual cost would be approximately AU$600 million (if the price is similar to the US), nearly 5% of the $12.6 billion Pharmaceutical Benefit Scheme expense for the 2019–20 financial year.28 If all Australians with mild AD were treated, this would increase the Pharmaceutical Benefit Scheme expenditure by about 50%. Real costs would be higher considering that, in addition to the drug, staff time and infrastructure are needed for infusions, monitoring for ARIA requires repeat MRI scans, and severe ARIA may require hospitalisation.7,22 Donepezil, in contrast, which has a clinically important treatment effect and a relatively benign side effect profile, costs about $200 per patient per year.

Given that about a quarter of patients with a clinical diagnosis of AD do not have amyloid pathology,29 treatment without a positive cerebrospinal fluid or positron emission tomography biomarker will expose people unnecessarily to the risk of ARIA. A scarcity of dementia specialists to make AD diagnoses, limited access to infusion centres, and poor availability of amyloid imaging are bottlenecks. An increase in demand for diagnosis would likely overwhelm our limited dementia diagnostics services. Prescribing a drug approved without evidence of consistent clinical benefit is complex even for specialists, and there are currently no data on safety or efficacy in moderate to severe disease, or guidance on when to stop treatment.12

Regulatory approval based on a post hoc analysis is analogous to winning a Texas sharpshooter contest by drawing a bull’s‐eye around a bullet hole. Premature approval without evidence of clinically meaningful benefit from controlled clinical trials is costly to government as well as patients (who are exposed to side effects), may make it harder to recruit to experimental placebo‐controlled clinical trials, and could divert research funding away from the development of more effective treatments. Disease‐modifying therapies for AD are urgently needed, but science, not desperation, should guide the approval process.