O P T I M I S M A B O U T therapeutic interventions in Alzheimer disease is based largely on a growing confidence that we understand theunderlyingpathobiologyof the illness.Breakthroughs ingenetics ledtotheidentificationof3early-onset Alzheimer disease autosomal dominant genes, each of which affects the metabolism of the small -amyloid (A ) peptide. This peptide is the core component of senile plaques, 1 of the 2 major histopathological hallmarksof thedisease.Biochemical and molecular analyses ultimately led to the identification of the and -secretases, enzymes important in generating A , and innovative approachesaimedatclearingA , including immunotherapy, have emerged. Giventhis intensive investigationinto the pathology of A , and the emergence of pharmacological targets directly relatedtoA metabolism,hope was strong that effectivewaysofhaltingor reversing thediseasewerenear at hand. However, a series of disappointing results from large clinical trials have ensued. Perhaps it would havebeenunrealistictoexpectthefirst generation of pharmacologic agents to be blockbusters, but even modest signals of clinical efficacy have been elusive.Thefirstactive immunization program was halted because of meningealencephalitis inasmallpercentage of the patients, but long followup of patients who were in the trial to autopsy showed no clear change in clinical course despite dramatic clearance of amyloid. In a more recent failure, the development of a -secretase inhibitor was halted because of systemic adverse effects, as wellasbecausepatients’cognitionnot only did not improve, but apparently got worse. Whatwentwrongwiththeserecent clinical trials? Is the “A hypothesis” in need of revision? The strong case from genetics and the fact that all patientswithAlzheimerdiseasehave innumerable amyloid plaques in their brains and elevated levels of synaptotoxic soluble amyloid still make the strongestpossiblecasethatA isacritical factor in the disease. However, it hasbeenrecognizedforyearsthatthere arepoorclinicalpathological correlations between plaques and the extent of cognitive impairment in patients with Alzheimer disease, whereas the otherhistologic featuresofAlzheimer disease—neuronal loss, synaptic loss, and neurofibrillary tangles—tend to match up much better with clinical symptoms. I propose in this commentary that the amyloid hypothesis needs to be reformulated to posit a 2-stageprocess inwhichamyloid initiates a cascade of pathophysiologic consequences,which in turnbecome increasingly independentof the initiatingamyloid.Thisrevisedhypothesis wouldexplain theapparent lackofefficacyof targetingA after thedisease isestablishedandsupports thesearch for therapies aimed at downstream events thatare theproximatecauseof neurodegeneration and dementia. One of the pillars of this new formulationis theobservationthat innumerableindividualshaveaccumulated amyloid plaques in their brains but donothavedementia. Asmanyas 30% of cognitively normal individuals in their 70s have amyloid plaques by positron emission tomography scans. These data concur with previous autopsy studies that had highlighted the possibility that amyloid plaques could accumulate even in “normal” individuals. Moreover, the numberofamyloidplaques thataccumulate in the brain appears to reach a near maximal level by the time patients have clinically apparent symptomsofcognitiveimpairment. Inan autopsy-based cross-sectional study, we foundthat amyloiddepositionaccumulated prior to the development of cognitive impairment, and thereafter, the amount of amyloid in the brain was similar across a wide range ofclinical severity, remainingatanalmostflatplateau.Bycontrast,atrophy, neuronal loss,synaptic loss,andnumberof tangles increasewithseverityof illness (Figure 1). The first phase of the disease, in which we now suggest that soluble oligomeric and fibrillar A accumulates, leads to local disruptionoftheneuropil, lossofdendritic spines, remodeling of neurites, and inflammatory responses. The second phase would consist of the (further)developmentoftangles,neuronal loss, synaptic loss, and glial responses in limbicandassociationcortices, and ultimately throughout the corticalmantle,withneurodegeneration ultimately leading to end-stage organ failure. What would happen if the second phase of the disease proceeds somewhat independently of A ? Dementia in Alzheimer disease would then be envisioned to be similar to the symptoms that a patient who has had heart attacks may display due to congestiveheartfailure.Forthesepatients, lowering their cholesterol level for a fewmonthsorayearwillnot improve their performance on the treadmill. Therapeutic strategies may be necessary to improve the symptoms of organfailureotherthanprimaryprevention strategies aimed at lowering the riskofheart attacks.Medicinehas innumerableexamplesofconditionsthat evolve over time and in which the ultimate cause of disability and illness Author Affiliation: Department of Neurology, Massachusetts General Hospital/Harvard Medical School, Charlestown.