Neurology, perhaps more than any other field of medicine, relies on a close association between clinical phenotype and underlying pathology. This is certainly the case for the neurodegenerative diseases where despite advances in genetics and biomarkers a careful clinical assessment remains the key to accurate diagnosis. In this context, Alois Alzheimer's description of the association between progressive cognitive decline centred on episodic memory and the presence of amyloid plaque and tau tangle pathology remains central to our formulation of Alzheimer's disease—although his original case was highly atypical, being a young woman now known to have an autosomal dominantly inherited form of the disease.1

A fundamental question that underlies all neurodegenerative disorders is why certain pathologies associate fairly reliably with certain clinical phenotypes. Seminal pathology studies in Alzheimer's disease showed that amyloid plaque and tangle pathology spreads through the brain in a fairly predictable sequence, starting in the medial temporal lobe before involving other neocortical structures;2 this sequential spread—particularly of tau pathology—correlates broadly with the typical progression of symptoms. More recently, the availability of biomarkers has allowed for other aspects of the pathological cascade to be assessed in vivo. Cerebrospinal fluid allows for measurement of brain-enriched proteins, inflammation and synaptic dysfunction; positron-emission tomography allows for visualisation of abnormal deposits of …