John A. Hardy and Gerald A. Higgins and tau phosphorylation can be controlled by intracellular calcium (22). Thus, APP may induce neurofibrillary tangle formation as a consequence of its ability to increase the concentration of intracellular calcium, leading to phosphorylation of tau and the formation of paired helical filaments (6). The intervening steps by which APP affects calcium homeostasis still remain to be elucidated. However, the overall mechanism is consistent with what we know about calcium-mediated neuronal death. The mutations in APP sofar described are responsible only for a small proportion ofcases ofAlzheimer's disease (23). Indeed, most cases of Alzheimer's seem to occur in a sporadic fashion, suggesting that there must be other causes of the disease. The cascade hypothesis suggests that other causes of Alzheimer's act by initially triggering APP deposition. For example, there is an association betweenhead trauma and Alzheimer's (24). Dementia pugilistica, ex-hibited by boxers, may be thought of as a variant of Alzheimer's disease because these individuals exhibit both AIP deposits and neurofibrillary tangles (25). Furthermore, amyloid deposition occurs as an acute re-sponse to neuronal injury in both man and animals (26). This deposition could be caused by an induction of the APP gene through an interleukin-mediated stress re-sponse (27) because APP increases in re-sponse to a number of neuronal stresses (28). Although acute effects may only lead to transitory disruption ofAPP metabolism, it is possible that in some individuals the entire pathological cascade leading to Alzheimer's would be initiated. The evidence we have described supports the hypothesis that the AI3P molecule initiates the pathological cascade of Alzheimer's disease. APP-containing COOH-terminal derivatives of APP seem the most likely molecular candidates for initiation of the cascade, withthe process presumably taking several decades to produce the fullblown pathology of the disease. The ongoing development of transgenic animals that express APP or APP and exhibit Alzheimer's-like pathology should provide good models for experimental testing of key elements in the cascade. The identification of additional mutations in APP and other genes that cause Alzheimer's pathology will allow refinement of the amyloid cascade hypothesis and point to targets for therapeutic intervention.