The Neurobiology and Pharmacotherapy of Alzheimer's Disease

Alzheimer's disease (AD), the most common cause of dementia,¹ is a devastating illness characterized by cognitive deterioration as well as behavioral, affective, and psychotic disturbances. Over the past decade, dramatic progress has been made in understanding the neurogenetics and pathophysiology of AD. Four different genes have been associated with AD, and others are likely to be discovered. The mechanisms by which altered amyloid and tau protein metabolism, inflammation, oxidative stress, and hormonal changes may produce neuronal degeneration in AD are being identified, and rational pharmacological interventions based on these discoveries are being developed. The rapid pace of these developments is underscored by the more than 11,000 medline references for AD since 1990. In this article, we review recent advances in understanding the neurobiology of Alzheimer's disease and discuss corresponding therapeutic strategies.

NEURITIC PLAQUES AND NEUROFIBRILLARY TANGLES

In 1907, Alois Alzheimer reported the neuropathologic findings of Auguste D., a woman from Frankfurt who died at the age of 51 with progressive dementia.² Alzheimer eloquently described a constellation of neuropathological abnormalities including amyloid plaques and neurofibrillary tangles. Although some controversy persists regarding the proper diagnosis of Auguste D.,^3,4 the neuritic plaques (NPs) and neurofibrillary tangles (NFTs) described by Alzheimer have been considered the major pathological hallmarks of AD ever since.

Neuritic plaques are composed of extracellular spherical deposits of beta-amyloid protein (AΒ),⁵ which are intimately associated with dystrophic axons and dendrites as well as activated microglia and reactive astrocytes.⁶ The protein AΒ is composed of 39 to 43 amino acids and is produced under normal metabolic conditions from amyloid precursor protein (APP). The deposition of AΒ appears to be the earliest morphological change in the formation of neuritic plaques.⁷ Variants of AΒ, which differ only at their C-terminus, are found in extracellular and cerebrovascular plaques. This hydrophobic C-terminal sequence is a critical determinant of solubility, affecting the rate of amyloid formation. Soluble AΒ is composed primarily of a 40–amino acid protein (AΒ 40) with only a small amount of the 42 or 43 amino acid isoforms (AΒ 42/43). The AΒ 42/43 variant forms insoluble fibrils much faster than AΒ 40 in vitro.⁸ Furthermore, coincubation of small amounts of AΒ 42/43 with AΒ 40 results in immediate aggregation, suggesting that amyloid plaques may be seeded by trace amounts of this component. The term “diffuse plaques” refers to deposits of AΒ without the surrounding neuritic degeneration.⁹ It is believed that diffuse plaques contain the soluble, nonfibrillary form of AΒ and may represent an early stage of senile plaques.¹⁰

Neurofibrillary tangles consist of intraneuronal bundles of abnormal filaments, composed of a highly phosphorylated form of the microtubule-associated protein tau. They form either parallel bundles or randomly arranged paired helical filaments, occupying the perikaryotic cytoplasm and extending to the dendritic processes.¹¹ The critical phosphorylation of tau depends on kinases that are still being identified.

The pathological features of AD occur in a consistent pattern. On a cellular level, pyramidal neurons, the main component of corticocortical pathways,¹² are the major cortical cell type lost in AD.¹³ The majority of the NFTs are seen in cortical layers III and V, whereas neuritic plaques, present throughout the depth of the cortex, are more numerous in layers II and III.¹² NPs are initially found predominantly in the amygdala and in high-order temporal and parietal association cortex. Later in the course of the disease, they develop in other mesial temporolimbic structures such as the hippocampus.¹⁴ In addition, NPs are found in meningeal and cortical blood vessels. Primary visual and sensorimotor areas remain relatively spared.¹² NFTs have a slightly different distribution. Although NFTs are abundant in high-order association cortex, the entorhinal cortex, hippocampus, and amygdala are affected early in the disease process.¹⁵ NFTs are not specific for AD and may develop in response to various cerebral insults such as postencephalitic parkinsonism, subacute sclerosing panencephalitis, and dementia pugilistica, among other conditions. Both NFTs and NPs are seen in normal aging, although their density and distribution typically differ from that seen in AD. Current neuropathologic criteria for the diagnosis of AD are based on the quantitative distribution of NPs.^16,17

APP METABOLISM

Amyloid precursor protein is an integral membrane glycoprotein (see review⁶). Within APP, AΒ encompasses 28 residues of the extracellular and 12 to 15 residues of the putative transmembrane domains. APP occurs in different isoforms, the 695–amino acid form being expressed almost exclusively in neurons. The two other common forms (751 and 770 amino acids) are expressed in both neural and non-neural cells and carry an additional domain with protease inhibitor functions. Transported to the cell membrane via secretory vesicles, some of these molecules undergo proteolytic cleavage by an unidentified protease, termed α-secretase. This process releases a soluble ectodomain, designated “α APP.” It also cleaves the precursor within the AΒ domain, precluding AΒ protein formation.¹⁸ The basal release of the soluble form of APP appears to be regulated in vitro by receptor-coupled activation of protein kinase C.^19,20 Alternatively, uncleaved surface APP may follow a proteolytic pathway. APP is internalized and recycled by endosome-lysosomes^21,22 or via an endocytic route.²³ This latter pathway seems to be the site of subsequent cleavage by the putative β- and γ-secretases, which results in production and release of AΒ.²³ Increased intracellular calcium levels enhance AΒ production, suggesting that calcium-requiring protease(s) may be involved in the production of AΒ.²⁴

APP GENE AND CHROMOSOME 21

The observation that patients with Down's syndrome (trisomy 21) develop cognitive deterioration and typical pathological features of AD by middle age led to the discovery of the APP gene on chromosome 21.²⁵ Simultaneously, a locus segregating with a minority of early-onset familial AD (FAD) kindreds was mapped to this chromosome²⁶ in the same region as the APP gene.²⁷ Later, several missense mutations within the APP gene that resulted in amino acid substitutions in APP were identified in these FAD kindreds.^28–32 It appears that such mutations alter the previously described proteolytic processing of APP, generating amyloidogenic forms of AΒ. Skin fibroblasts from subjects carrying APP mutations produce increased AΒ 42/43.³³ Increased plasma concentration of AΒ 42/43 is also seen in these subjects³⁴ regardless of age, sex, or clinical status.^33,34 Interestingly, some patients with sporadic AD may exhibit similar elevations of plasma AΒ 42/43.³⁴

Improved understanding of amyloid metabolism provides several theoretical targets for pharmacological interventions. Since the basal release of the soluble form of APP is dependent on putative α-secretase and regulated by protein kinase C activity, agents that enhance this metabolic pathway may directly impede AΒ formation. Conversely, drugs that decrease cleavage of APP by β- and γ-secretases might also decrease AΒ deposition. In addition, agents that inhibit the aggregation of AΒ may diminish its neurotoxicity. Transgenic mice expressing human mutant APP have recently been developed. They will likely prove useful in assessing these and other potential therapeutic agents. These mice develop memory deficits^35,36 and exhibit many of the pathologic abnormalities seen in AD, including AΒ deposits, neuritic plaques, synapse loss, astrocytosis and activated microglial cells.³⁷ They do not produce NFTs.

PRESENILIN-1 AND PRESENILIN-2

Fifty to seventy percent of early-onset autosomal dominant AD cases appear to be associated with a locus (AD3) mapped by genetic linkage to the long arm of chromosome 14 (14q24.3).³⁸ Numerous missense mutations have been identified on a strong candidate gene called presenilin-1 (PS1).^39,40 At the same time, another autosomal dominant locus responsible for early-onset AD was localized to chromosome 1. Two mutations were identified on the candidate gene, designated presenilin-2 (PS2). One was demonstrated in an Italian pedigree,⁴¹ and the other one was found in Volga German kindreds.⁴² The physiological role of presenilins and the pathogenic effects of their mutations are not yet well understood. Their two putative products, PS1 and PS2, share substantial amino-acid and structural similarities, suggesting that they may be functionally related. In addition, the expression patterns of PS1 and PS2 in the brain are very similar, if not identical. Both PS1 mRNA and PS2 mRNA are detectable only within neuronal populations.⁴³ Mutations result in single amino-acid substitutions in PS1 and PS2.⁴³ PS1 and PS2 exhibit strong homology with two peptides of a nematode (Caenorhabditis elegans) involved in the intracellular trafficking and recycling of proteins. Immunochemical analyses indicate that PS1 localize to intracellular compartments such as the endoplasmic reticulum and Golgi complex^43,44 that are involved in similar functions. Recent evidence supports the role of presenilins in AΒ metabolism. Mice deficient in the expression of PS1 exhibit dramatic decrease in proteolytic cleavage by γ-secretase of the transmembrane domain of APP.⁴⁵

Presenilin-1 is immunoreactive with NPs.^46,47 Asymptomatic and demented subjects carrying the PS1 mutation have increased production of the amyloidogenic AΒ 42/43 isoform in skin fibroblasts and plasma.³⁴ Prominent deposition of AΒ 42/43 is found in many brain regions of patients with PS1 mutations.⁴⁸ These findings, in suggesting that inhibiting presenilin function might decrease AΒ amyloid production, offer new therapeutic avenues.

APOLIPOPROTEIN ALLELES AND CHROMOSOME 19

The gene encoding the cholesterol-carrying apolipoprotein E (APOE) on chromosome 19 has been linked to late-onset familial⁴⁹ and sporadic⁵⁰ AD. This gene is inherited as an autosomal codominant trait with three alleles. APOE E3 is the most frequent allele (allele frequency=0.78), followed by APOE E4 (allele frequency=0.14) and APOE E2 (allele frequency=0.07). Three homozygous genotypes (APOE E4/E4, E3/E3, and E2/E2) and three heterozygous genotypes (APOE E3/E4, E3/E2, and E2/E4) result in the respective phenotypes.⁵¹ APOE E4 gene “dose” is correlated with increased risk^49,52 and earlier onset of AD. Subjects with 2 copies of the APOE E4 allele (4/4 genotype) have significantly greater risk of developing AD than subjects with 2/3 or 3/3 genotypes. On the other hand, the APOE E2 allele may confer a protective effect.^53,54 Mean age at onset is significantly lower in the presence of two APOE E4 copies.^49,54 A recent collaborative study suggested that APOE E4 exerts its maximal effect before the age of 70.⁵⁵ Many APOE E4 carriers do not develop AD,⁵⁶ and many patients with AD do not have this allele. Therefore, the presence of an APOE E4 allele does not secure the diagnosis of AD, but instead the APOE E4 allele acts as a biological risk factor for the disease,^52,56 especially before the age of 70.

How APOE genotype influences AD susceptibility remains unclear. Subjects with one or two copies of APOE E4 allele exhibit more beta amyloid staining at autopsy than subjects homozygous for APOE E3,⁵⁷ suggesting that APOE genotype may affect the rate of amyloidogenesis. APOE is synthesized in various organs including the brain.⁵¹ The APOE E4 isoform binds to AΒ with higher affinity than the other APOE E isoforms.^58,59 APOE is also immunoreactive with plaques in situ.⁵⁷ Studies have suggested that the pathogenic role of APOE E4 may be the result of inefficient suppression of amyloidogenesis⁶⁰ or increased AΒ polymerization.⁶¹ Impaired clearance of AΒ with sequestration in the extracellular space⁵⁰ has also been suggested.

However, APOE may be involved in microtubular metabolism as well. APOE is immunoreactive with neurofibrillary tangles in situ,⁶² and electron microscopic studies have localized APOE to the cytoplasm (see review⁶³). There are also APOE-isoform differences in binding to tau with higher avidity for APOE E3 than for APOE E4.⁶² APOE E3 binds to the receptor-binding domain of tau thought to be involved in the assembly of paired helical filaments.⁶³ There are also APOE-isoform differences in neurite outgrowth. APOE E4 is associated with depolymerization of microtubules and inhibition of neurite outgrowth.⁶⁴ By contrast, APOE E3 is believed to prevent the hyperphosphorylation of tau and thereby to protect tau's abilities to interact with cellular microtubules, stabilize the cytoskeleton, and support neurite elongation.⁶⁵

THE AMYLOID HYPOTHESIS VERSUS THE TAU HYPOTHESIS

A central and controversial issue in the pathogenesis of AD is the relationship between amyloid deposition and NFT formation. That abnormal amyloid metabolism plays a key pathogenic role is supported by several lines of evidence. The fibrillar form of AΒ has been shown to be neurotoxic to cultured neurons at high concentrations.^66,67 Apoptosis, self-regulated cell destruction, is one possible mechanism of cellular death in AD.⁶⁸ Cultured cortical and hippocampal neurons treated with AΒ protein exhibit changes characteristic of apoptosis, including nuclear chromatin condensation, plasma membrane blebbing, and internucleosomal DNA fragmentation.⁶⁸ The fibrillar form of AΒ has also shown to alter the phosphorylation state of tau protein.⁶⁹ The identification of several point mutations within the APP gene in some patients with early-onset familial AD and the development of transgenic mice exhibiting cognitive changes and NPs also incriminate AΒ in AD. Additional evidence comes from recent experimental data supporting the role of presenilins in AΒ metabolism as well as findings of abnormal production of AΒ protein in presenilin-mutation familial Alzheimer's disease.

Although very popular, the “amyloid hypothesis” is not uniformly accepted. Dementia severity correlates better with the number of neocortical NFTs than with NPs.¹⁵ Tau is a protein whose function is to stabilize neuronal microtubules.⁶⁵ As mentioned previously, the APOE E3 allele may protect tau from hyperphosphorylation. It has been speculated that destabilization of the microtubular system may disrupt the Golgi apparatus, in turn inducing abnormal protein processing and increasing production of AΒ.⁷⁰ In addition, this destabilization may decrease axoplasmic flow, generating dystrophic neurites and contributing to synaptic loss.

OXIDATIVE STRESS

Oxidative stress is believed to be a critical factor in normal aging⁷¹ and neurodegenerative diseases such as Parkinson's disease and amyotrophic lateral sclerosis. There is evidence that oxidative stress is also implicated in AD. Formation of free carbonyls⁷² and thiobarbituric acid-reactive products,⁷³ an index of oxidative damage, are significantly increased in AD brain tissue compared with age-matched controls. Plaques and tangles display immunoreactivity to antioxidant enzymes.⁷⁴ There are multiple mechanisms by which cellular alterations may be induced by oxidative stress, including production of reactive oxygen species (ROS) in the cell membrane (lipid peroxidation). This in turn impairs the various membrane proteins involved in ion homeostasis^75,76 such as N-methyl-d-aspartate receptor channels or ion-motive adenosine triphosphatases.⁷⁵ The subsequent increase in intracellular calcium, along with the accumulation of ROS, damages various cellular components such as proteins, DNA, and lipids and may result in apoptotic cellular death.^75,76 Increased intracellular calcium may also alter calcium-dependent enzyme activity⁷⁷ such as the implication of protein kinase C in amyloid protein metabolism and the phosphorylation of tau. Therefore, blocking the increase in free intracellular calcium may diminish neuronal injury. Nimodipine is a lipophilic calcium channel blocker whose activity is mediated through inactivation of voltage-dependent L-type (long-lasting) calcium channels. Clinical trials of nimodipine for AD have yielded generally disappointing results.⁷⁸

The apoptotic pattern of cellular death seen in oxidative stress is similar to that produced by AΒ exposure,⁶⁸ and AΒ neurotoxicity is attenuated by antioxidants such as vitamin E.⁷⁵ AΒ may induce its toxicity by engaging several binding sites on the membrane surface.⁶⁶ The receptor for advanced glycation end products (RAGE) may be one of these receptors.⁷⁹ RAGE is a member of the immunoglobulin superfamily of cell surface molecules known for its capacity to bind advanced glycation end products. This receptor is highly expressed by cortical neurons, especially in the hippocampus and cerebellum, during rat brain development.⁸⁰ RAGE is also expressed in a variety of other cell types, including endothelial cells and mononuclear phagocytes. Activation of this receptor is believed to trigger cellular oxidative reactions. In addition, RAGE has been shown to mediate the interaction of AΒ protein with glial cells, which may be one of the first steps in the inflammatory cascade (see below).

The long-term effects of alpha-tocopherol (vitamin E) and l-deprenyl (selegiline) on AD progression were recently assessed.⁸¹ Vitamin E has antioxidant properties, and selegiline, an irreversible inhibitor of monoamine oxidase B (MAO-B), is believed to act as free radical scavenger and modulator of monoaminergic neurotransmission. Selegiline is marketed for the treatment of Parkinson's disease. In a controlled trial, 341 patients with AD were treated with selegiline (10 mg/day), vitamin E (2,000 IU/ day), both, or placebo over a 2-year period.⁸¹ After adjustment for baseline differences among the groups on the Mini-Mental State Examination, both vitamin E and selegiline were found to delay the time of occurrence of death, institutionalization, and loss of the ability to perform activities of daily living. However, there was no significant difference between the different groups on the cognitive section of the Alzheimer's Disease Assessment Scale (ADAS-Cog), a 70-point cognitive battery designed to provide a global index of cognitive performances.⁸² A series of short-term studies of selegiline has suggested a beneficial effect in AD (reviewed⁸³), but long-term studies have showed no or very mild improvement on cognitive measures.^84–86 Milacemide and lazabemide are two other MAO-B inhibitors. Milacemide has not proven to be effective in AD.⁸⁷ Lazabemide is now under investigation in the United States. A study from Bordeaux investigated the relationship between alcohol consumption and risk of AD. A negative association was found between “moderate” alcohol consumption (3 to 4 glasses of red wine per day) and the incidence of AD.⁸⁸ One hypothesized explanation rests on the antioxidant properties of red wine.⁸⁹ The effects of ginkgo biloba, a compound with putative antioxidant properties, were recently evaluated in a study of 236 AD patients.⁹⁰ A modest but statistically significant improvement was seen on the ADAS-Cog in the ginkgo-treated group. Idebenone is an antioxidant that inhibits lipid peroxidation in vitro. This compound is currently in phase III clinical trials in the United States. Recent experimental data have suggested that melatonin may have neuroprotective effects by preventing AΒ-induced oxidative damage.⁹¹ Melatonin prevented increased intracellular calcium, lipid peroxidation, and apoptotic death in neuroblastoma cells exposed to AΒ. It is interesting to note that serum⁹² and cerebrospinal fluid⁹³ melatonin levels are reduced in demented subjects compared with elderly control subjects. Further trials are needed to delineate the clinical application of this compound.

INFLAMMATORY REACTIONS

A growing body of evidence indicates that inflammatory and immune mechanisms accompany the degenerative process in AD. Reactive microglia, cells phenotypically related to peripheral monocytes, are densely embedded in neuritic plaques.^94–96 Increased cytokines are seen in the serum, cortical plaques, and neurons of patients with AD compared with aged-matched control subjects.⁹⁷ Interestingly, the anti-inflammatory cytokine-transforming growth factor beta 1 (TGF-β1) has recently been found to promote or accelerate the deposition of AΒ.⁹⁸ Classical complement pathway fragments are also found in the brains of AD patients, and AΒ may directly activate the classical complement pathway in an antibody-independent fashion.⁹⁶ One recent study found an association between AD and the major histocompatibility complex, located on chromosome 6.⁹⁹ The age at onset of AD was significantly lower in patients carrying the A2 allele. The possibility of linkage disequilibrium with another gene requires further study.

The potential role of anti-inflammatory agents in the treatment of AD was first suggested by the discovery of an unexpectedly low prevalence of AD among patients with rheumatoid arthritis.^94,100 Additional case-control,¹⁰¹ cross-sectional,¹⁰² and longitudinal studies¹⁰³ have found an inverse relation between the risk for AD and nonsteroidal anti-inflammatory drug (NSAID) use. The relative risk for AD decreased with duration of NSAID use, from 0.65 for a less than 2-year exposure to 0.40 for 2 or more years of treatment.¹⁰³ The protection conferred by aspirin appears weaker than that of NSAIDs, and no association was found between AD risk and use of acetaminophen. A 1-year longitudinal study found that AD patients using NSAIDs declined less on several cognitive measures.¹⁰⁴ A 6-month double-blind placebo-controlled trial of indomethacin has been conducted in 44 patients suffering from mild to moderate AD. Given at a dose of 100–150 mg/day, indomethacin was found to improve cognitive function.¹⁰⁵ However, about 20% of patients developed drug-related side effects, primarily gastrointestinal intolerance. The toxicity of long-term NSAID use, especially in elderly patients, may hinder the potential use of these agents in patients with, or at risk for, AD. Recent observations suggest that inhibition of cyclo-oxygenase (COX)-1 may cause much of this toxicity, whereas inhibition of COX-2 may confer the clinical anti-inflammatory effect. Interestingly, COX-2 is expressed in neocortex, particularly in limbic structures such as the hippocampus that are preferentially involved in Alzheimer's disease.¹⁰⁶ Selective inhibitors of COX-2 are currently under investigation for the treatment of AD.

Whether markers of immune and inflammatory processes actively participate in the neurodegenerative process or whether they represent an epiphenomenon remains unclear. Brain specimens from elderly patients with arthritis treated with NSAIDs have similar numbers of senile plaques, as do control brains. However, less microglial activation is seen in the arthritis patients' brains.¹⁰⁷ This suggests that although NSAIDs may not impede senile plaque formation, they may delay or prevent clinical symptoms by limiting the associated inflammation.

As mentioned above, RAGE has been shown to mediate the interaction of AΒ and glial cells, producing cellular activation and an inflammatory response with cytokine production, chemotaxis, and haptotaxis. The expression of this receptor appears to be upregulated in neurons, vasculature, and microglia in affected regions of AD brains.⁷⁹ The unrelated class A scavenger receptor (Class A SR) also mediates the adhesion of microglial cells to AΒ fibrils. Senile plaques contain high concentrations of microglia that express class A SRs.¹⁰⁸ RAGE and class A SRs may represent novel pharmacological targets for diminishing the inflammatory and oxidative reactions associated with AD.

Genetic factors associated or potentially associated with Alzheimer's disease are summarized in Table 1 and Table 2.

ESTROGENS AND ESTROGEN REPLACEMENT THERAPY

Experimental studies have demonstrated that estrogen loss predisposes to cognitive decline and neuronal degeneration (see review¹¹²). For instance, ovariectomized female rats have diminished learning abilities. Neuronal choline uptake and choline acetyltransferase activities, indices of cholinergic function, are shown to be reduced in the brain of these animals. The expression of nerve growth factor and brain-derived neurotrophic factor mRNA is decreased as well. These cognitive and biological changes are reduced by administration of estrogen replacement therapy (ERT). In addition, estrogen has been shown to exert cytoprotective effects and to prevent AΒ toxicity in human neuroblastoma cell cultures.

Human studies suggest that ERT may prevent, delay, or reduce the symptoms of AD. In a case control study, treatment with ERT at the time of enrollment was found to be significantly less common in women meeting criteria for AD than in control subjects.¹¹³ In addition, the demented women using ERT performed significantly better on the Mini-Mental State Examination. Confounding factors included retrospective analysis and the lack of information regarding dose and previous history of ERT. Another study, with a similar design, found comparable results.¹¹⁴ A beneficial role for postmenopausal ERT is more convincingly supported by longitudinal studies of normal aging. In a group of 1,124 elderly women,¹¹⁵ age at onset of AD was significantly later among estrogen users. The relative risk of AD was significantly reduced even after adjustment for education, ethnic origin, and apolipoprotein E genotype. This was especially true for women on ERT for longer than 1 year. Another study of 472 post- and perimenopausal women provided additional support for a protective effect of ERT.¹¹⁶ The observational design of these two studies resulted in several limitations involving nonrepresentative populations and variability in dose and route of administration. Prospective trials of estrogen in normal perimenopausal women are in progress.

Little information is available on the effect of ERT in AD patients. In a double-blind placebo-controlled multicenter clinical trial of tacrine for AD, a subgroup of women receiving ERT had a significantly better response than women receiving tacrine alone or placebo.¹¹⁷ A randomized clinical trial of ERT in women with AD is currently in progress (Alzheimer's Disease Cooperative Study).

CHOLINERGIC NEUROTRANSMISSION AND ALZHEIMER'S DISEASE

Cholinergic systems are involved in normal memory functions,¹¹⁸ and cholinergic deficiency has been implicated in the cognitive and behavioral¹¹⁹ manifestations of AD. Activity of the synthetic enzyme choline acetyltransferase (CAT) and the catabolic enzyme acetylcholinesterase (ACE) are significantly reduced in the cerebral cortex,^120,121 hippocampus, and amygdala in AD patients.¹²² The nucleus basalis of Meynert, medial septum, and diagonal band of Broca, which provide the main cholinergic input to the hippocampus, amygdala, and neocortex, demonstrate a fairly selective pattern of neuronal degeneration in AD.¹²³ Loss of cortical CAT¹²⁴ and decline in acetylcholine synthesis in biopsy specimens have been found to correlate with cognitive impairment¹²⁵ and reaction time performance.¹²⁶ The relationship between basal forebrain dysfunction and amyloid precursor protein metabolism remains poorly understood. Experimental lesions of the nucleus basalis in rats result in elevated cerebrospinal fluid APP secretion.¹²⁷ The effects of cholinergic agents on APP release have yielded conflicting results in animal studies.

Because cholinergic dysfunction may contribute to the symptoms of AD patients,¹¹⁸ enhancing cholinergic neurotransmission constitutes a rational basis for symptomatic treatment. Attempts to augment cholinergic activity have employed acetylcholine (ACh) precursors, cholinergic agonists, and drugs inhibiting acetylcholinesterase (AChE), the enzyme responsible for the breakdown of ACh. Precursor therapy, using choline or phosphatidylcholine (lecithin), has been largely unsuccessful.¹²⁸ Thus far, cholinergic agonists (bethanechol, pilocarpine, arecoline, RS-86, xanomeline tartrate) have yielded modest results or have been associated with significant side effects.^129,130

Various classes of AChE inhibitors have been studied for the treatment of AD. They differ in their mechanism of inhibition (reversible, pseudo-irreversible, or irreversible), metabolism, and brain selectivity. Physostigmine, a short-acting cholinesterase inhibitor, was one of the first drugs offering promising results. In the initial study, intravenous physostigmine enhanced performances on a recognition memory task.¹³⁰ A review of 11 studies using physostigmine¹³¹ found improvement in all five investigations that involved parenteral administration and in four of the six involving oral administration. Tacrine (THA or tetrahydroaminoacridine) is a centrally active reversible AChE inhibitor with a moderately long duration of action. It is thought to exhibit a variety of other pharmacological properties such as enhancement of monoaminergic activity. In 1986, Summers et al.¹³² reported significant improvement in the performance of psychometric tests after long-term treatment with tacrine. Larger multicenter studies confirmed these findings, with statistically significantly improvement on various outcome measures such as the ADAS-Cog,^133–135 clinician's impression scales,^134,135 and time until nursing home placement.¹³⁶ Elevation of serum levels of alanine aminotransferase (ALT), greater than three times the normal value, occurs in about 25% of patients.¹³⁷ Adverse events also included gastrointestinal complaints, and resulted in frequent dosage adjustment and a high dropout rate.^135–137 Overall, these studies demonstrated that tacrine produced modest, dose-dependent, symptomatic improvement in a portion of patients who were able to tolerate it.

Donepezil inaugurates a new generation of AChE inhibitors with greater selectivity for AChE in the central nervous system. Donepezil is a reversible AChE inhibitor with a long plasma half-life of 70 hours. A recent double-blind study enrolled 161 patients with mild to moderate AD for a period of 15 weeks.¹³⁸ Subjects received either placebo or daily doses of 1 mg, 3 mg, or 5 mg of donepezil. There was statistically significant, dose-related improvement in ADAS-Cog scores in the donepezil group. Another 30-week study of 473 patients yielded similar results. After 6 months, the mean improvement in the ADAS-Cog scores for the donepezil group was 3 points compared with the placebo group.¹³⁹ Donepezil has not been associated with hepatotoxicity. Nausea, vomiting, and diarrhea led to discontinuation in less than 3 percent of patients.

Tacrine and donepezil have been approved by the Food and Drug Administration (FDA) for the treatment of mild to moderate AD. For tacrine, initial FDA-approved dosing strategy recommended starting at 10 mg four times daily and increasing by 40 mg every 6 weeks to a maximum of 160 mg. Transaminase levels were monitored every other week, and tacrine was to be stopped if ALT levels exceeded 5 times their normal value. New recommendations are expected. They should allow increasing the doses every 4 weeks, and include a monitoring of transaminases every 4 weeks, to be discontinued if ALT levels exceed 20 times their normal value. For donepezil, dosing recommendations suggest starting the treatment with 5 mg daily. The possibility of additional benefits with a higher dose allows increases up to 10 mg after 4 to 6 weeks.

Although still under investigation, rivastigmine (ENA 713), a carbamate AChE inhibitor, and metrifonate, an organophosphorus compound used as an insecticide and for the treatment of schistosomiasis, produce no liver toxicity and may hold promise in the treatment of AD.

The locus ceruleus provides the principal noradrenergic innervation of cerebral cortex, whereas the dorsal raphe nucleus and central superior nucleus provide serotonergic projections. In AD, neuronal loss and NFTs are seen in both these structures.^140,141 Some studies have shown an association between locus ceruleus or central superior nucleus involvement and depression in patients with AD.^141,142 A variety of other neurotransmitter and neuropeptide systems are also abnormal in AD, including dopamine, somatostatin, corticotrophin-releasing factor, GABA, and substance P, suggesting that repleting any one neurotransmitter is unlikely to produce dramatic clinical improvement.

CONCLUSION

Alzheimer's disease results from multiple pathogenic events, including protein processing abnormalities, inflammation, and oxidative stress, leading to the destruction of selective neuronal populations. Figure 1 diagrams the main pathologic events and potential pharmacological targets in AD. Until recently, no effective treatments were available for patients with Alzheimer's disease. Cholinergic enhancement with acetylcholinesterase inhibitors has now been conclusively shown to provide mild symptomatic benefit in some patients with AD. Estrogen replacement therapy may provide additional symptomatic benefit and may slow disease progression. Anti-inflammatory and antioxidant medications may also directly interfere with the underlying disease process. Improved understanding of amyloid metabolism offers a theoretical framework for multiple pharmacological interventions. Ultimately, the optimal pharmacological management of AD will likely consist of polypharmacy with agents that enhance deficient neurotransmitters and target multiple steps of the underlying disease process.

TABLE 1. Genetic factors associated with Alzheimer's disease

Enlarge table

TABLE 2. Other locus or susceptibility genes potentially associated with Alzheimer's disease

Enlarge table