Biological Markers in Alzheimer's Disease

Trey Sunderland, Susan E. Molchan, and George S. Zubenko

Biological markers in medicine are usually considered diagnostic or prognostic tools in the treatment of serious illness. However, Alzheimer's disease (AD) remains largely a clinical diagnosis of exclusion. While elegant neuropathologic procedures allow final diagnoses at autopsy, the current consensus histopathological criteria have not been validated. Because patients rarely undergo a diagnostic brain biopsy because of the general lack of subsequent therapeutic options, discovering a readily accessible biological marker in AD which is firmly associated with the diagnostic neuropathology would be a major scientific major step forward.

In this chapter, several approaches in the study of biological markers will be reviewed:

1. Cerebrospinal fluid (CSF)

2. Peripheral tissue markers

3. Pharmacologic and neuroendocrine probes

4. Behavioral and biochemical correlates

Because the expanding study of biologic markers in AD is such a vast topic, not all approaches can be included. These areas each represent a fertile research field, and many of the major studies will be highlighted. Although the underlying neurobiology of AD is still elusive, these biologic probes offer a way to study potential etiologic mechanisms in affected and at-risk subjects, for both diagnostic and prognostic purposes. They may also provide a mechanism to follow treatment response once more effective therapies are available.

Given the physiologic proximity of the CSF to the brain and the usual lack of biopsy confirmation of AD, CSF studies have long held promise for the antemortem diagnosis of the illness. Unfortunately, that promise has mostly been unfulfilled, because conflicting results have been the rule rather than the exception. There are multiple reasons for these conflicting results, not the least of which is the diagnostic uncertainty in any clinical study of AD. Other obvious clinical sources of variance include the severity of dementia, concomitant medical and neuropsychiatric variables, status of the subject at the time of the lumbar puncture (i.e., inpatient or outpatient), concurrent medications, and recent dietary history. There are also technical factors which may play a role in the final results, including the delay between sampling and assays, aliquot sampling, possible freezer thaw effects, and, of course, differing assay techniques across laboratories. Despite these multiple issues, enthusiasm runs high in the scientific community regarding CSF studies, and the search continues to discover the aberrant CSF marker that might better distinguish AD from control populations or other neuropsychiatric conditions.

An obvious starting point for any investigation with CSF in AD would be with markers of the cholinergic system. Postmortem studies uniformly reveal decreases in cholinergic indices, including choline acetyl transferase (CAT), acetylcholinesterase (AChE), and the number of cholinergic neurons (see Cholinergic Transduction, Structure and Functon of Cholonerhic Pathways in the Cerebral Cortex, Limbic System, and Thalamus of the Human Brain, Functional Heterogeneity of Centeral Cholinergic Systems, and Chapter 134 ). Furthermore, these changes have long been correlated with clinical measures of cognitive decline (11). However, the results of numerous studies of cholinergic markers in the CSF have been disappointing. In an early study, Johnson and Domino (34) reported no differences between demented subjects and healthy young controls in CSF CAT. Subsequently, the results have been mixed (35), perhaps due to methodologic differences among groups and the fact that CAT is known to be highly variable across CSF samples.

Similarly, measures of AChE have been found to be normal (61) or reduced (41). At best, the overlap between normals and Alzheimer patients is considerable, obviating the diagnostic utility of this measurement. In what is perhaps a more important measure of dynamic cholinergic change with AD, there is a report of a continuous decline in specific activity of the AChE enzyme with progression of the disease process over a 2-year period (22). Interestingly, this group had previously reported no significant changes after the first year of their study (22). When studying familial versus nonfamilial AD patients, Kumar and Giacobini (37) showed a significant decrease in CSF choline in the familial patients but no differences between the groups in CSF AChE. Butylcholinesterase (BuChE) activity has also been measured in AD and controls. As with the other cholinergic markers, there are reports of decreases (2), but most of the studies show no significant changes (4, 34). When studying the ratio of AChE to BuChE, there is still mixed opinion as to whether there is increased separation between the patient group and controls.

The CSF monoamine metabolites are the most studied substances in the CSF of AD, but the picture is still confusing. Despite the fact that there are marked decreases in total brain norepinephrine, serotonin, and dopamine at autopsy, the in vivo measures of the metabolites do not reveal consistent decreases. Most studies reveal the levels to be within normal limits, some show reductions, and one even reports an increase in norepinephrine metabolites (see Table 1). For example, Hartikainen et al. (29) showed no significant differences in CSF measures of MHPG, 5HIAA, or HVA between 27 Alzheimer patients and 34 elderly controls. On the other hand, Parnetti et al. (53) revealed lower CSF HVA in a subgroup of Alzheimer patients with a later onset of illness. Interestingly, this later-onset group also showed a significant correlation between their CSF HVA levels and several neuropsychological measures. Within the serotonergic system, Gottfries (26) has consistently reported a reduction in 5-HIAA. Concerning the noradrenergic system, Raskind et al. (59) have previously reported increases of CSF MHPG in a group of advanced AD patients (N = 9) compared to age-matched controls or more mildly affected AD patients (N = 7). However, in a separate study of histologically verified AD, the CSF MHPG was normal in spinal fluid compared to that of normal controls, but the CSF HVA and 5HIAA levels were indeed lower in patients than in controls (51) (see also Schizophrenia and Glutamate for related discussion).

CSF somatostatin levels represent a special case in AD studies, because there is remarkable consistency in the world literature. Following the initial report of a decrease in brain somatostatin in AD brains, Oram and colleagues (50a) were the first to report a substantial decrease in CSF somatostatin-like immunoreactivity (SLI) in AD patients versus neurologic controls. Since then, there have been multiple other reports of lowered CSF somatostatin levels in AD patients versus normal controls; interestingly, the CSF SLI deficit is not limited to AD, because several other neuropsychiatric conditions have been found to show reduced levels (29, 65). Such diagnostic overlap may be important, because when biopsy-proven AD patients have been compared to normal controls, only the presenile and not the senile patients have been shown to have the SLI reduction (24). Nonetheless, there appears to be a significant relationship between lumbar CSF SLI and frontal lobe SLI, at least as measured in cortical biopsies of a limited number of patients (25). In the only long-term study of CSF SLI in AD thus far published, there was no evidence of significant reductions with progression of the illness over 18 months of follow-up (3). However, when CSF SLI was measured during tetrahydroaminoacridine (THA) treatment, positive responders were noted to have increased levels post treatment, and that change was significantly correlated with neuropsychological improvement following a 4-week trial (1) (see also Neuropeptide Y and Reletated Peptides).

One of the more interesting recent developments in the antemortem study of AD has been the detection of altered absolute and relative amounts of amyloid derivatives in the CSF of AD patients versus older normal controls using antisera to amino acids 45–62 in the beta-amyloid precursor protein (52). Significant decreases in the larger forms (125 and 105 kD) and increases in relative amount of the smaller form (25 kD) of these derivatives have also been found in elderly controls compared to younger controls, although to a lesser extent than the changes found in AD (52). Significant decreases in the larger APP fragment thought to include the Kunitz inhibitor domain (PN-2) were also reported using a monoclonal mouse antibody which recognizes the amino-terminal epitope of all three major isoforms of APP (76); however, this method has also led to reports of quantitative increases in APP with the Kunitz-type inhibitor domain in the CSF of AD patients (75). Using a double-sandwich ELISA procedure with densitometric analysis of Western blots, a reduction of total APP, APP 695, and APP 751/770 has been reported in cases of sporadic AD (57). In a study of a family affected by presenile AD, clinical symptoms in one subject were associated with low levels of APP similar to those with sporadic disease, whereas two symptom-free family members had normal levels (23).

Another major strategy in studying CSF has been to examine the fluid for evidence of increased breakdown or "wear and tear" substances that would be likely byproducts of a degenerative process like AD. Glycosphingolipids are present in all human cells, particularly plasma membranes, and are produced during normal cell turnover; however, they have been found in higher quantities or different ratios in the CSF of AD patients (9). Examining the neuroimmune system has also led to interesting discoveries with evidence of (a) higher CSF levels of interleukin-1b in AD patients when compared to other neurological controls, including multiple sclerosis patients (13), and (b) higher SP-40,40 levels, a modulatory protein of the complement system, in the CSF of AD patients versus that of normal controls (15).

Using other approaches, the reactivity of ubiquitin, a small (8.5 kD) protein found in most cells, has been shown to be significantly higher in AD over control CSF, perhaps reflecting the extensive ubiquitinylation of paired helical filaments (PHFs) in the brains of AD patients (78). Neuronal thread protein, a 20-kD protein of unknown function which nonetheless accumulates in the brains of AD patients, has also been found in larger quantities in the CSF of AD patients than in that of both normal and other neurologic controls (17). On the other hand, levels of biopterin, peptidyl-glycine-alpha-amidating monooxygenase (PAM), and superoxidase dismutase have all been reported to be lower in the CSF of AD patients than in that of controls (12). When looking at metals and trace elements, Basun et al. (5) showed reduced levels of cadmium and calcium but increased levels of copper in the CSF of AD patients when compared to that of normal controls. Aluminum, lead, manganese, and mercury levels were measured and were not found to be statistically different amongst groups. One study reported significant increases in CSF xanthines, uric acid, and creatinine in AD patients, but the controls were younger patients with psychiatric problems and not age-matched controls (18). Finally, cyclic GMP, but not cyclic AMP, has previously been reported to be elevated in AD patients versus normal controls (74).

While AD is generally considered a central nervous system disorder, numerous biological alterations in tissues outside of the cerebral nervous system have been reported to show associations with AD. These peripheral abnormalities have been found in platelets, blood cells, skin fibroblasts, and peripheral vessels, just to name a few, and they present researchers with readily accessible tissue for further study of potential biological markers of AD. Rather than present a comprehensive review of this literature, this section serves to revisit associations that have been replicated and those for which plausible connections to the pathophysiology of AD could be formulated. A more complete review of this literature can be found elsewhere (68).

Increased platelet membrane fluidity (PMF) in AD, as reflected by decreased fluorescence anisotropy of 1,6-diphenyl-1,3,5-hexatriene in labeled membranes, was initially reported by Zubenko et al. (84, 85) and has since been replicated by most (21, 28, 30, 56, 77), but not all, investigators (36). Using a threshold value for DPH anisotropy of 0.1920 at 37°C (the 90th percentile for neurologically intact elderly controls), increased PMF identifies a subgroup of patients with AD who have distinct clinical features. Increased PMF (DPH anisotropy less than 0.1920) appears to be associated with a group of AD patients with an earlier age of onset, more rapid progression of illness, less frequent focal abnormalities on electroencephalogram, decreased prevalence factors, and an earlier age of onset of AD in family members (for review see ref. 82).

Among demented patients, increased PMF is relatively specific for AD. This membrane abnormality is not shared by elderly patients with major depression (85), patients with multi-infarct dementia (MID) (30), or cognitively intact patients with Parkinson's disease (82). Moreover, increased PMF has been shown to be a stable characteristic over a 1-year follow-up period and to be a familial trait (92). The last finding provides strong evidence against unidentified medication exposure or a nonspecific effect of chronic illness as the source of the platelet membrane alteration. Finally, segregation analysis of 14 pedigrees revealed that at least 80% of the variance of PMF within families could be explained by the inheritance of a single major locus which has yet to be identified (14). It should be noted that while PMF appears to be under genetic control, increased PMF is not specifically associated with familial AD (21, 86). This finding suggests that the PMF locus may influence the likelihood of developing AD and may modify its clinical expression without being sufficient to cause AD.

Increased PMF does not seem to be a manifestation of a generalized cell membrane defect in AD. Instead, several lines of evidence converge to suggest that increased PMF in AD results, at least in part, from a selective abnormality of internal platelet membranes. Intact platelets exposed to DPH for a short time so that external membranes are preferentially labeled do not exhibit the increase in membrane fluidity associated with AD (92). Furthermore, ultrastructural studies have revealed an increased frequency of platelets bearing accumulations of trabeculated cisternae bounded by smooth membranes resembling smooth endoplasmic reticulum (SER) (28, 87).

Biochemical markers of SER function in platelets with increased membrane fluidity are also altered in AD. Activity of the SER marker, antimycin-A-insensitive NADH-cytochrome reductase, is decreased in platelets from AD patients (28), specifically those with increased PMF (83). Furthermore, there is evidence of disturbance in SER-mediated calcium homeostasis in AD. The percentage increase in cellular calcium following thrombin stimulation is significantly higher in platelets from AD patients than in those from matched controls. Both calcium/magnesium ATPase and calcium ATPase, SER-associated enzymes that regulate free calcium, show decreased specific activity in platelets from patients with AD (28). Overall, these results provide evidence that increased PMF in patients with AD is associated with a structural and functional impairment of internal platelet membranes that may affect protein maturation, localization, and calcium homeostasis.

Alterations in calcium homeostasis have been reported in several cell types in patients with AD, though most work has been done in cultured skin fibroblasts. Peterson and Goldman (54) reported increases in total bound calcium and intracellular calcium in fibroblasts derived from both AD and normal elderly groups, when compared to younger controls. Both findings, however, were more marked in the AD group. Cytosolic free calcium, by contrast, was decreased in fibroblasts from the AD group when compared with young controls, again with intermediate results in the normal elderly (55). These results suggest an impairment of calcium homeostasis in AD, though the observed changes may represent an exaggeration of those seen in normal aging. Caution must be used in interpreting these studies, because the actual number of AD patients involved was small.

Changes in calcium homeostasis appear to be associated with impairments in mitochondrial function and cytoskeletal organization. Mitochondrial processes such as the oxidation of glucose and glutamine and their incorporation into proteins and lipids were decreased in a fashion parallel to that of cytosolic free calcium, with cells from the AD group affected most. By comparison, markers of cytosolic and nuclear processes, such as incorporation of leucine into protein and thymidine into DNA, were depressed more by aging than by AD (54). There was decreased spreading (a measure of cytoskeletal function) of fibroblasts from AD patients which improved when these cells were treated with calcium ionophores (55).

An alternative model of pathogenesis hypothesizes the presence of impaired mitochondrial function in AD, with resulting impairments of calcium homeostasis (8). Changes in mitochondrial oxidative functions have been reported in cultured skin fibroblasts; Blass et al. (8) have postulated that uncoupling of mitochondrial oxidative phosphorylation may lead to the accumulation of the abnormal cytoskeletal elements seen in AD. The mechanism linking mitochondrial uncoupling to the development of abnormal cytoskeletal elements, however, remains speculative and may be related to the effects of mitochondrial function on calcium homeostasis (8).

Mammalian olfactory receptor cells are located in the epithelium of the upper nasal cavity beneath the cribriform plate. Each of these neuronal cells extends a dendritic process that terminates in a ciliated olfactory vesicle at the mucosal surface of the nasal epithelium, while projecting an axonal process that terminates on the olfactory bulb. Because of their intimate connection to the central nervous system, olfactory neurons are a promising model that may provide insights into the molecular pathology of neuropsychiatric disorders including AD. Indeed, pathological changes in morphology, distribution, and immunoreactivity of olfactory neurons have been reported in biopsy samples of olfactory epithelium obtained at autopsy from patients with AD (69).

Olfactory neurons are continuously generated from stem cells located within the epithelial layer and appear to be unique among neurons in their ability to regenerate throughout life. Recent success has been reported in the establishment of primary cultures of olfactory epithelium from adult human cadavers and living subjects (80). Cells from cloned cell lines have been reported to exhibit neuronal properties including neuron-specific enolase, olfactory marker protein, neurofilaments, and growth-associated protein 43. In addition, the cells express non-neuronal molecules, suggesting properties of neuroblasts or stem cells. The clonal cultures have been reported to contain 5–10% of cells sufficiently differentiated to manifest odorant-dependent biochemical responses to submicromolar concentrations of odorants.

Olfactory neuroblasts, as well as several other cell lines associated with the central nervous system, have been reported to contain detectable levels of carboxyl terminal degradation products (CTDs) of the amyloid precursor protein. The levels of these potentially amyloidgenic peptides are significantly elevated by incubation of cells in the presence of low concentrations of chloroquine, an inhibitor of lyosomal function, and leupeptin, a protease inhibitor (81). In contrast, incubation in the presence of monensin, an endosomal specific inhibitor, was not accompanied by a significant change in the baseline level of these degradation products (81). Overall, these results suggest that lysosomes or some closely related internal membrane compartment play an important role in the degradation of APP CTDs.

Wolozin et al. (81) also reported considerable variation in the basal levels of APP CTDs across cell lines. Cell types associated with the pathology of AD, such as olfactory neuroblasts and cortical vascular endothelial cells, had higher levels of CTDs than did lymphoblasts and melanoma cells. These differences became even greater after incubation in the presence of inhibitors of lysosomal function. Consistent with previous evidence, this observation suggests that the degradation of APP may be highly tissue-specific. This finding has obvious significance for studies of amyloidogenesis and may have important implications for the pathophysiology of AD.

Alterations in neurotransmitter and neurohormone function are thought to be involved in the physiology, if not the etiology, of neuropsychiatric illnesses such as AD. Because of these changes, and the modulatory effects that neurochemicals have on each other, it has been hypothesized that studies of peripherally obtained neuroendocrine measures will give clues about neurochemical alterations in the brain. Besides neuroendocrine measures, cognitive, behavioral, and physiologic responses can also be studied in response to drug challenges in an attempt to characterize patient subgroups and allow for better diagnostic or prognostic accuracy during subsequent clinical treatment.

Because of the central importance of the cholinergic deficit in AD, there has been a significant emphasis on probes of the cholinergic system. Table 2 lists studies published since 1985, in which a cholinergic agent has been used acutely in AD patients to assess effects on cognition, behavior, physiologic, and/or neuroendocrine measures. While a number of studies have shown a positive effect on memory function or activation level in these patients, many others were negative. Few studies have compared neuroendocrine responses after cholinergic stimulation between AD and age-matched normal control subjects. Challenges with the cholinesterase inhibitor, physostigmine, showed that AD subjects had blunted plasma arginine vasopressin, b-endorphin, and epinephrine responses as compared with controls (60). These data suggested that deterioration in AD results in decreased responsiveness of neuroendocrine systems regulated by central cholinergic systems.

The cholinergic system deficits in AD led to the hypothesis that patients with AD should be more sensitive than normal controls to the memory-impairing and other central effects of the muscarinic antagonist scopolamine. A centrally mediated "functional hypersensitivity" was indeed documented in AD patients as compared with normals, on a number of behavioral and cognitive measures (66). Similar increases in sensitivity have also been noted in other elderly neuropsychiatric populations including Parkinson's patients and Korsakoff's patients but not elderly depressives (64). At least five muscarinic receptor subtypes have been subsequently identified by molecular genetic studies. It is hoped that more receptor-subtype selective agonist and antagonists will become available for use as probes to help discern their physiologic roles.

The serotonin (5-HT) agonist m-chlorophenylpiperazine (m-CPP) is thought to be relatively selective for 5-HT1 receptors (38). One study examined the effects of m-CPP in AD patients, and it found that they had increased behavioral responsivity and cognitive sensitivity to intravenous infusion of the drug as compared with elderly control subjects (38). AD patients became anxious and restless, experienced psychomotor activation and perceptual abnormalities, and had a greater performance decrement on some measures of memory as compared with controls. Neuroendocrine responses (plasma levels of cortisol and prolactin) did not differ between the two groups (38). The increased behavioral sensitivity to m-CPP in AD patients was hypothesized to be secondary to decreased inhibition on 5-HT1 receptors from the loss of 5-HT2 receptors, or possibly from decreased inhibition from damaged cholinergic neurons on 5-HT systems. This hyperresponsiveness has been postulated to reflect a contribution of 5-HT systems to some of the behavioral disturbances (anxiety, depression, agitation, sleeplessness) which occur in AD (38) (Serotonin and Behavior: A General Hyothesis and Indoleamines: The Role of Serotonin in Clinical Disorders).

The benzodiazepine class of drugs interacts with the benzodiazepine–gamma-aminobutyric acid (GABA) receptor complex to enhance GABA activity, which is thought to lead to their anxiolytic action. Few studies of this neurotransmitter system have been done in AD. In one utilizing lorazepam as a probe, AD patients experienced more attentional impairment as compared with controls after drug administration (67). It was hypothesized that one possible reason for the difference between the groups was altered benzodiazepine sensitivity in the AD patients, secondary to decreased benzodiazepine receptor numbers.

AD and major depression can have overlapping clinical symptoms (e.g., depressed mood, sleep disturbance, psychomotor retardation, anxiety, agitation, and cognitive impairment). There also have been reports of overlapping biological markers, including nonsuppression of serum cortisol after a dose of dexamethasone (usually 1.0 mg, taken orally). The dexamethasone suppression test (DST) has been widely studied in samples of patients with AD, both with and without symptoms of major depression (see ref. 43 for review). The consensus has been that the rate of nonsuppression of cortisol during the test in nondepressed patients with AD is similar to that found in patients with major depressive disorder (about 50%) (43). Therefore, the DST is not useful in distinguishing dementia from depression. Positive relationships between post-dexamethasone cortisol levels and measures of dementia severity have been reported in some, though not all, studies (43).

Serum adrenocorticotropic hormone (ACTH) levels following administration of dexamethasone were reported to be elevated in two studies of AD patients (39). This change may be an indication of less effective feedback inhibition of cortisol, which could be caused by down-regulation of glucocorticoid receptors in the brain. Two studies have reported that, like patients with major depression, AD patients had a blunted ACTH response and a normal cortisol response to CRH infusion; this would be consistent with hypersecretion of CRH, and it implies a pathological process at the level of the hypothalamus or above. Two other studies found no difference in response to CRH between AD patients and controls (31) (The Role of Acetylcholine in Mood Disorders).

Many studies have reported associations between thyroid dysfunction and psychiatric symptoms, especially depressed mood. Clinically, patients with AD and major depression often have overlapping clinical syndromes, so it was hypothesized that a blunted TSH response to TRH may occur in some AD patients as it does in depressed patients. Most studies report no significant difference in the TSH response in AD patients as compared with normals, although two report blunted responses (45, 73). Some studies have reported that patients with a blunted TSH response tended to have higher levels of free T4, indicating normal feedback inhibition at the pituitary (45). Hypothalamic neuropathology has been documented in AD, but most studies have not found a significant change in TRH concentration in AD brain. It is therefore unknown why some of these patients have increased levels of T4; the mechanism may involve decreased feedback or increased stimulation of thyroid hormones by some other substance, such as norepinephrine. A few studies have examined the relationship between degree of dementia severity and TSH response; results have generally been negative (6, 20, 45). Measures of depression and TSH response do not appear to be significantly correlated (45).

The prolactin response in AD patients, usually to TRH, has been studied by a number of investigators. Dopamine has a strong inhibitory effect on prolactin, and serotonin and TRH stimulate its release. Deficits in the dopaminergic and serotoninergic systems have been reported in AD, so alterations in prolactin response may be expected. Over half the studies of prolactin response in AD found no significant difference from controls, one reported a blunted response, and two reported an enhanced response as compared with controls (20, 73). Two studies looked for correlations in response to measures of dementia severity but found none (6, 20) (Thyrotopin-Releasing Hormome: Focus on Basic Neurobiology).

Acetylcholine has a stimulatory effect on growth hormone (GH), so alterations in it have been hypothesized in AD given the cholinergic deficit. Also, decreased levels of the peptide somatostatin, a major inhibitory modulator of GH, in both CSF and brain of AD patients is well-documented (44). The adrenergic and GABAergic systems also have regulatory actions on GH secretion, and alterations have been reported in those systems in AD. GH responses to a number of substances, including GH-releasing hormone, TRH, and cholinergic agents, have been explored in AD patients. Over half of them show no difference between AD patients and controls, a few reported a blunted response, and one an enhanced response as compared with controls (16, 73).

The understanding of mechanisms of actions of drugs and hormones has recently moved beyond receptors, to the level of second messengers and genes. Evidence has also accumulated that AD is a systemic disorder, because many abnormalities have been identified in tissues other than brain (68). Some of these abnormalities therefore should be amenable to exploration with pharmacologic probes. In the brain, as well as in fibroblasts of patients with AD, alterations in the phosphoinositide second messenger system, including the enzyme protein kinase C (PKC), have been reported (62). There is evidence that PKC is involved in memory mechanisms, and increasing evidence suggests that lithium affects PKC function. In AD, chronic lithium resulted in significantly less membrane-associated (activated) PKC when compared to controls in three of the four isoenzymes examined (46), suggesting possible differential regulation in the platelets of AD subjects. Such aberrations in peripheral cells indicate that PKC may be involved early in the pathophysiology of the illness, and is not simply a secondary response to neuronal loss (62). Studies pursuing other possible alterations in phosphoinositide system components such as guanine-nucleotide-binding (G) proteins and membrane phospholipids in this paradigm are underway. In fact, future pharmacologic challenges may even link examinations of second messenger systems with dynamic brain imaging techniques.

Depression and psychosis often complicate the course of primary dementia, exaggerating the functional impairment suffered by patients and increasing the burden experienced by their caregivers (91). These behavioral abnormalities appear to occur independently in demented patients and, once they emerge, to be chronic or recurrent. Available evidence suggests that these syndromes of major depression and psychosis are less responsive to treatment when they occur in the context of primary dementia than when they develop in cognitively intact elders (91). Both syndromes are associated with poorer outcomes among demented patients who develop them. The emergence of major depression appears to be associated with an increased mortality rate (88, 93), but has no effect on the rate of cognitive decline (40). Conversely, psychosis appears to be associated with a faster rate of cognitive decline (19) without affecting mortality (19).

Autopsy studies of the neuropathologic and neurochemical correlates of major depression and psychosis in primary dementia have been guided by preexisting studies suggesting anatomic substrates for the idiopathic forms of these behavioral syndromes. The catecholamine hypothesis of affective disorders was attractive in this regard, because the cell bodies of the majority of the catecholaminergic neurons in the brain are localized in relatively discrete brainstem nuclei. In contrast, chronic idiopathic psychotic disorders are associated with decreases in blood flow glucose metabolism, cell loss, and cytoarchitectural changes in the neocortex and allocortex (see ref. 82 for review).

In a study of 37 demented patients who had participated in a longitudinal clinical study prior to death, Zubenko and Moossy (88) found that the emergence of major depression was associated with degeneration of the locus coeruleus and substantia nigra. However, patients who developed major depression did not differ from those who did not with respect to other clinical features or neuropathologic indices of global severity including cortical dementia of senile plaques (SPs) or neurofibrillary tangles (NFTs), reflecting the specificity of the brainstem findings. A logistic regression model that included the degenerative features of both the locus coeruleus and substantia nigra predicted the emergence of major depression with significantly greater accuracy than models employing the characteristics of either nucleus alone. This observation suggests an interaction of both noradrenergic and dopaminergic systems in the pathogenesis of major depression in primary dementia. Zweig et al. (93) have also reported an association of major depression with degeneration of the locus coeruleus and dorsal raphe nuclei in AD.

In a neurochemical analysis of the same 37 cases, Zubenko et al. (89) reported that the demented patients with major depression exhibited a 10-fold or greater reduction in the concentration of norepinephrine in the cortex, along with the relative preservation of choline acetyltransferase (ChAT) activity in subcortical regions, compared to demented patients without depression. The level of serotonin and its metabolite 5-HIAA showed a trend toward reduction in all cortical and subcortical regions examined. A paradoxical elevation in dopamine levels was observed in one region of the hippocampus in the depressed patients, but no consistent pattern emerged across brain regions. These findings did not appear to be related to medication exposure.

Zubenko et al. (90) reported an analogous study of 27 autopsy-confirmed cases of AD, with or without psychosis. In contrast to major depression, psychosis was associated with (a) increased cortical densities of SPs and NFTs and (b) a relative preservation of norepinephrine in the substantia nigra with trends in this direction for the remaining brain regions examined. Like major depression, psychosis was associated with a significant reduction of serotonin in the hippocampus that was accompanied by trends toward reduction of this amine and its metabolite 5-HIAA in the remaining regions. In their morphometric study of brainstem aminergic nuclei, Zweig et al. (93) did not observe a significant change in the numbers of neurons in either the locus coeruleus or dorsal raphe nuclei in AD patients who developed hallucinations or delusions.

The significance of these findings is manifold. Overall, these results provide the first direct evidence addressing the neurochemical and anatomic substrates of major depression and are largely consistent with the aminergic hypotheses of depression (42). Whether the relative preservation of ChAT activity in depressed, demented patients implicates cholinergic hyperactivity specifically in the pathogenesis of major depression (33) or, instead, reflects the inability to express or detect major depression in severely demented patients cannot be determined from these studies. It is noteworthy that the neuropathologic and neurochemical correlates of major depression and psychosis in dementia differ qualitatively from each other and from those associated with primary dementia alone. This observation reflects the specificity of the findings and may be related to the clinical observation that depression and psychosis appear to emerge independently of each other. The diffuse, but modest, reduction in the brain levels of serotonin and its metabolite 5-HIAA in both behavioral complications may be related to the apparent nonspecific therapeutic effect of trazodone in the management of agitated patients with dementia (32). Moreover, the degenerative nature of these changes may explain the clinical observation that major depression and psychosis are less responsive to treatment in the context of dementia than when they emerge in cognitively intact patients (91).

It is tempting to speculate that the biological correlates of major depression and psychosis in dementia may explain the negative prognoses associated with each of these complications of dementia (91). If the brainstem degenerative changes in depressed, demented patients extend beyond the resident aminergic nuclei, it may not be surprising that this behavioral complication is associated with a higher mortality rate. Moreover, greater cognitive impairment would seem a predictable consequence of the exaggerated cortical degeneration in demented patients with delusions or hallucinations. In summary, continued clinicopathologic and neurochemical studies of the behavioral complications of dementia may provide additional insights into the nature of AD as a multifocal brain disorder, may better define the relationship of depression to cognitive impairment in late life, and may suggest novel pharmacologic interventions. This area has been reviewed in detail elsewhere (47, 91).

published 2000