The amyloid cascade, a deterministic chain of events that leads from amyloid deposition, then tau, to neurodegeneration and progressive cognitive impairment, is the most widely accepted model of AD. According to this model, the 30% of people over 65 years of age without memory impairment who have amyloid in their brains have early Alzheimer's disease and need to be treated with anti-amyloid drugs to prevent the development of cognitive impairment. This model is suitable for autosomal dominant AD, the rare familial forms that begin around 40-50 years of age, but is less applicable to sporadic AD, the common form that affects people over 65 without significant heredity.

Due to the challenges of developing drugs targeting amyloid, the amyloid hypothesis needs to be reconsidered. New information regarding the epidemiology and biology of AD allows us to propose a probabilistic model of AD, in which amyloid remains a key player, but as one of several risk factors such as the e4 allele of the APOE gene, which promote the development of cognitive impairment. Cognitive impairment develops when the weight of these risk factors overcomes the resilience of the brain. The latter is itself determined by protective factors of genetic and environmental origin.

The model suggests that:

• people at risk (with brain amyloidosis) should be treated before they develop memory impairment (secondary prevention), as we do today for the prevention of vascular disease
• all risk factors, not just amyloid, need to be treated to maximise brain protection; research should develop drugs for the most potent genetic risk factor, the e4 allele of the APOE gene

Article to be published in Nature Reviews Neuroscience under the title The probabilistic model of Alzheimer disease: the amyloid hypothesis revised, Giovanni B. Frisoni, Daniele Altomare, Dietmar Rudolf Thal, Federica Ribaldi, Rik van der Kant, Rik Ossenkoppele, Kaj Blennow, Jeffrey Cummings, Cornelia van Duijn, Peter M. Nilsson, Pierre-Yves Dietrich, Philip Scheltens and Bruno Dubois