Table 1.

Findings that appear to undercut the amyloid hypothesis of AD and counterarguments that could explain these discrepancies

Findings	Counterarguments
Amyloid plaque burden correlates much less well with degree of cognitive impairment than do neurofibrillary tangle counts	Aβ deposits appear to be a very early and widespread event that is distant to the clinical dementia and can lead to many downstream cellular and molecular changes (e.g., microgliosis, neuritic dystrophy, tangles, etc.) that are more proximate to and causative of neuronal dysfunction
Many humans show sometimes abundant Aβ deposits at death but were not noticeably demented	Some or many of these deposits are diffuse plaques (not rich in abnormal neurites and glia); the patients were often not tested rigorously before death; and Aβ oligomer levels per plaque are much lower than in AD brains (Esparza et al, 2013), suggesting that plaques can effectively sequester oligomers in a non‐diffusible, less neurotoxic state, at least up to a point
Some human neuropathological studies suggest tangles may precede amyloid plaques	Such studies may not have searched systematically for diffuse plaques or soluble Aβ oligomers in the brain. Human genetics proves that Aβ‐elevating APP mutations lead to downstream alteration and aggregation of wild‐type tau, whereas tau mutations do not lead to Aβ deposition and amyloid‐related dementia
A hypothesis that AD is fundamentally due to loss of presenilin function has been put forward	AD‐causing presenilin mutations may indeed act through partial loss of function of this protease, but these heterozygous mutations do not produce clinically detectable loss of presenilin function (e.g., Notch phenotypes), and organismal development and function are normal until the carriers develop typical AD symptoms in mid‐life, heralded by elevated Aβ42/43 to Aβ40 ratios. Moreover, 99.9% of all AD patients have wild‐type presenilins
Numerous clinical trials of anti‐amyloid agents have not met their pre‐specified endpoints	Several of these agents had inadequate preclinical data, poor brain penetration, little human biomarker change, and/or low therapeutic indexes (e.g., tramiprosate; R‐flurbiprofen; semagacestat). Most such failed trials enrolled many patients in the late‐mild and moderate stages of AD, whereas other trials conducted in very mild or mild AD produced suggestive evidence of clinical benefit. AD trials done prior to obligatory amyloid‐PET imaging turned out to have up to ~25% of subjects that were amyloid‐negative (i.e., did not have AD)