possible to maintain normal brain function. This must be true, since some individuals maintain function for this period of time. Indeed, it appears that cognitive function can be preserved in some individuals even though some sensory functions may be compromised. Thus, as illustrated in Figure B-1, the successful aging line can be relatively flat.

Let us further postulate that the mechanism of decline can be divided into two general phases: initiation and propagation (Johnson et al., 1998). The initiation phase is distinct in terms of relative risk factors (e.g., APOE genotype) and protective factors. According to this hypothesis, the phase is relatively reversible and very amenable to interventions; it is relatively stable even though decline does occur. This idea is consistent with epidemiology data (e.g., Breitner, 1996) and histochemical data on factors affecting the accumulation of β-amyloid (see Johnson et al., 1998). For example, APOE at low levels of amyloid accumulation can determine the average age of decline onset. Once amyloid accumulation (or a similar process) reaches a certain level, however, the propagation phase is set in motion.

The propagation phase is distinct in that self-reinforcing molecular cascades are the net driving force; these cascades supersede contributing risk factors and accelerate pathogenesis. In terms of β-amyloid accumulation, this phase is largely independent of APOE-ε4, although it may have gender factors. Examples of possible autocatalytic cascades that could contribute to a propagation phase include the ability of amyloid to induce the amyloid precursor protein (APP) and chronic inflammation (see Cotman and Su, 1996). Such a propagation phase could be reversible up to a threshold point; however, once past this point, it may be irreversible.

This hypothetical model must also be considered in the context of the microenvironment. Thus, in the initiation phase, one would postulate that pathology is focused on vulnerable regions, but within those regions is largely confined to local domains, perhaps even individual cells. Subsequent transition/entry into the propagation phase results in the spread of pathology throughout the network. This propagation mechanism is unknown but may represent a breakdown of the local microenvironment. This is an important issue and is not addressed in the context of most current molecular mechanisms.

The progression of pathology is embodied in the Braak and Braak staging, in which the induction of tangles spreads through the limbic system network and is in essence the morphological equivalent of this set of events. Recently, my colleagues and I have established morphological evidence for transsynaptic propagation of neurofibrillary tangle pathology. We examined a series of neuropathologically staged cases and traced the temporal induction of AT8 and PHF staining in a well-established trisynaptic pathway: entorhinal stellate neurons, dentate gyrus granule cells, and CA4 pyramidal neurons. Cellular changes along this circuit appeared to initiate in the entorhinal cor-