Intended for healthcare professionals


New treatments for Alzheimer’s disease

BMJ 2023; 382 doi: (Published 22 August 2023) Cite this as: BMJ 2023;382:p1852
  1. Robert Howard, professor of old age psychiatry,
  2. Helen C Kales, Joe P Tupin, chair of psychiatry 2
  1. 1University College London, London, UK
  2. 2University of California, Davis, California, USA
  1. Correspondence to: H C Kales: hckales{at}

Blazing trails or the road to nowhere?

Donanemab is a new monoclonal antibody treatment that binds to amyloid β to stimulate microglial removal of amyloid plaques from the brain in people with Alzheimer’s disease.1 In the recently published Trailblazer-ALZ 2 trial in 1736 adults with early symptomatic disease, participants treated with donanemab had a mean decline over 18 months that was 3.03 points less than those given placebo on a 144 point scale of cognition and functional ability (integrated Alzheimer disease rating scale (−6.02 v −9.27; difference 3.25, 95% confidence interval 1.88 to 4.62). All participants had evidence of brain amyloid and tau protein accumulation on positron emission tomography at recruitment.2 Of a further 24 secondary and exploratory trial outcomes, 23 showed statistically significant benefits with donanemab. Notably, the benefit on the mini-mental state examination was 0.48 points; patients with Alzheimer’s disease dementia typically show an annual decline of 3 points on this measure.3

The most important adverse events associated with this class of drugs are amyloid related oedema and effusion or haemorrhage on magnetic resonance imaging (MRI). Oedema was seen in 24% of 853 participants treated with donanemab and haemorrhage in 19.7%, necessitating monitoring with serial MRI and leading to three deaths.

Clinical benefit is unclear

The trial results were received with excitement at the Alzheimer’s Association International Congress in July 2023 and attracted considerable media attention. Although the large number of participants gave the trial power to show statistical significance of small differences between drug and placebo, most commentators agree that these modest absolute differences are smaller than conventionally defined minimum clinically important differences.4

The progression of symptoms and functional impairment in early symptomatic Alzheimer’s disease is gradual, making it challenging to show any slowing of this already slow process over 18 months. The temptation to convert modest differences in quantitative outcome measures into differences in rates of decline is understandable. Thus, a 3.03 point difference on a 144 point scale (2.1% of the total range of the scale) has been translated into a 22.9% slowing of disease compared with placebo, when considered as a fraction of the decline seen over 18 months in participants who received placebo.

Because there are no agreed minimum clinically important values for slowing of symptom decline, it is harder for clinicians and patients to judge what such changes might actually mean in their lives when weighing up the benefits and risks of treatment. Percentage slowing has been used to convey an impression that the clinical course of disease has been modified, that benefits can be visualised in terms of months of better function saved, and even that it is possible to extrapolate beyond the 18 months of trial data to predict cumulative benefits with longer follow-up. Claims that treatment with these drugs would give people with Alzheimer’s disease several months longer engaging with hobbies, driving safely, or recognising grandchildren will have reached desperate patients and their families, but are not supported by data.5

Donanemab is likely to join lecanemab6 in receiving marketing approval from the US Food and Drug Administration, and applications to other regulatory bodies across the world are in progress. Limitations in the capacity of health systems to safely provide these treatments to real world patients mean they will be inaccessible to all but a tiny minority. Only 20% of screened patients met entry criteria for Trailblazer-ALZ 2; participants were young (mean age 74 years), fit, and unlikely to belong to minority populations.

Longer trials are needed to evaluate whether removal of amyloid plaques, undeniably achieved by these drugs, shifts the clinical decline curve in ways that are meaningful for patients, their families, and health and social care systems. Without these data and with emerging real world experience of lifechanging brain oedema and haemorrhage in recipients (the FDA has already placed a black boxed warning on lecanemab7 and will do so for all drugs in this class), amyloid antibodies are not going to make the kind of impact on Alzheimer’s disease claimed in recent headlines.89

Meanwhile, the expectations of people with dementia and their families (who are already bitterly familiar with such false dawns) about access to these treatments must be responsibly managed, including more and clearer information about the real benefits and risks associated with their use. Healthcare systems for people with dementia, which have fragile and often limited funding structures, will be at risk of damaging distortion if resources are redirected to support access to such expensive and clinically ineffective treatments.



Log in

Log in through your institution


* For online subscription