The Psychotic Phenomenon in Probable Alzheimer's Disease

P;t4>sychotic symptoms (e.g., delusions, hallucinations) are common in AD patients and impose a considerable burden on caregivers and clinicians involved in the patients' care.^1–8 These symptoms are associated with premorbid low educational levels^9,10 and are important predictors of institutionalization and cognitive and functional decline in AD patients.^11–15 Because the presence of psychotic symptoms can alter the course of the disease and have significant social and medical implications, researchers have explored the clinical, neuropathological, and radiological correlates of delusions, hallucinations, or both in AD patients in order to understand their pathophysiological mechanisms.

One recent approach to the study of cerebral neurophysiology of psychosis in AD has been the use of functional neuroimaging techniques, although the results have been inconsistent (Table 1). Some single-photon emission computed tomography studies in AD patients with delusions have found bilateral temporal lobe hypoperfusion,¹⁶ while others found a right temporal¹⁷ or frontal¹⁸ lobe hypoperfusion. Kotrla et al.¹⁹ reported frontal lobe hypoperfusion in patients with delusions and parietal lobe hypoperfusion in those with hallucinations.

Positron emission tomography (PET) is a sensitive instrument to measure cerebral metabolism in AD, even in its early stages.²⁰ Grady et al.²¹ found that patients with psychosis had hypometabolism of the frontal lobes, in addition to temporoparietal dysfunction. Sultzer et al.²² found that delusions, hallucinations, and suspiciousness correlated with frontal hypometabolism. More recently, Hirono et al.²³ found that regional cerebral glucose metabolism was increased in the left temporal lobe and decreased in the left occipital lobe in AD patients with a history of delusions.

One of the difficulties with functional neuroimaging studies of psychotic symptoms in AD is that these symptoms rarely occur in isolation. For example, patients with psychosis have more episodes of verbal and physical aggression^14,24–26 and more behavioral disturbances^14,26 than those without psychosis. The presence of these other psychiatric symptoms can alter the results of neuroimaging studies. Furthermore, patients may present with either delusions or hallucinations (or both), which may have separate clinical and neuroimaging correlates in AD. For example, patients with visual hallucinations may have dysfunction in the right posterior cortex, whereas patients with auditory hallucinations may have greater left posterior cortex dysfunction.

The frontal and temporal lobes appear to be involved in the etiology of both delusions and hallucinations. However, it is not clear whether frontal lobe dysfunction is critical in the etiology of the psychotic phenomenon. It is possible that previous studies detected an abnormality related to the presence of delusions (which are more frequent than hallucinations) but that the functional abnormality associated with hallucinations remains undetected. We report here a series of 4 AD patients with psychosis who underwent PET studies of cerebral blood flow. The purpose of the study was to determine whether there is a critical brain region that serves as a common denominator for the development of the psychotic phenomenon in AD.

METHODS

We reviewed the psychiatric characteristics of 9 patients with Probable AD²⁷ who were participating in a longitudinal study of dementia at the University of Pittsburgh. Each patient underwent an extensive evaluation, including medical, neurological, psychiatric, social work/nursing, and neuropsychological examinations, as previously described.^5,28 All patients were fluent speakers of English before the onset of dementia. None of these patients had significant signs of cerebrovascular disease, and none met the exclusion criteria of Hachinski Ischemic Score²⁹ greater than 4 or evidence of possible infarction in neuroimaging studies. The clinical diagnosis was reached after a careful review at a consensus conference by two neurologists, two psychiatrists, and two neuropsychologists.

Demographic characteristics of the subjects are shown in Table 2. All patients had more than 12 years of education. Although our goal was to closely match patients by age, educational level, and Mini-Mental State Examination (MMSE) score,³⁰ 1 patient with delusions was older and 1 patient with hallucinations was more cognitively impaired than other psychotic and nonpsychotic patients.

First, the relative regional cerebral blood flow (rel-CBF) of the 9 AD patients and 9 nondemented elderly control subjects was examined to determine specific differences between groups. None of the nondemented control subjects had experienced any psychiatric illness or was taking psychiatric medication. Second, the rel-CBF values of 2 patients with delusions and 2 patients with hallucinations were compared and contrasted with the mean rel-CBF of 5 AD patients without those symptoms. All of the 4 patients with psychosis experienced psychotic symptoms within 24 hours of the PET scan session, and none of them was taking psychiatric medication. None of the 5 AD patients without psychosis had experienced symptoms of major depression, mood lability, or apathy, and none of them was taking psychiatric medication.

Psychiatric Assessment

The psychiatric evaluations were conducted by psychiatrists using a semistructured interview.³¹ Delusions were defined in accordance with the DSM-IV criteria³² and were distinguished from confabulations, disorientation, and amnesia by requiring that the false beliefs be persistent despite evidence to the contrary. Hallucinations were accepted as present if the patient had spontaneously reported a sensory perception with no concomitant external perception. None of these patients had a past history of major depression, a previous psychotic illness, or a history of substance abuse.

Positron Emission Tomography

Subjects were scanned in a Siemens 951R/31 PET scanner. At the end of the transmission scan, 40–50 mCi of [¹⁵O]-water was injected as an intravenous bolus, and data acquisition began 30 seconds after the injection. The actual rel-CBF from the [¹⁵O]-water time activity data was calculated by using a standard compartment model with corrections for dispersion and timing delays in the blood activity curve.³³ The rel-CBF was examined in nine regions of interest (ROIs): orbitofrontal, frontal dorsolateral, anterior cingulate, medial temporal, superior temporal, parietal, and occipital cortices, and basal ganglia and thalami. The rel-CBF data for each ROI were normalized to the whole brain activity by computing an average of all of the ROIs and dividing each ROI by this denominator. The patients' magnetic resonance (MR) scans and a neuroanatomy atlas³⁴ were used to identify the ROI. The MR images were not digitally co-registered to the PET images, but were used to visually guide the identification of the ROI.

RESULTS

Nondemented control subjects showed higher rel-CBF in the left dorsolateral prefrontal cortex, superior temporal, and parietal cortices bilaterally than AD patients. The Mini-Mental State Examination³⁰ and the Mattis Dementia Rating Scale³⁵ scores were higher in nondemented control subjects than in AD patients (Table 3).

Three principal findings emerged from this study. First, all of the AD patients with psychotic symptoms had significantly lower rel-CBF in the left medial temporal and dorsolateral frontal cortices than AD patients without psychotic symptoms. As shown in Table 4, the rel-CBF values for the psychotic patients in these brains regions were all below the 95% confidence interval established by the nonpsychotic patients. Second, the 2 patients with hallucinations had, in addition, significantly lower rel-CBF in the right parietal cortex relative to the nonpsychotic AD patients. Third, the 2 patients with aggression (1 with hallucinations and 1 with delusions) also had decreased rel-CBF in the right orbital and dorsolateral frontal cortices and the right cingulate cortex.

One of the patients with delusions had increased rel-CBF in the right thalamus, the other in the left thalamus. One of the patients with hallucinations had increased rel-CBF in the right basal ganglia and right thalamus, the other in the right anterior cingulate gyrus and right orbitofrontal cortex and in the basal ganglia, bilaterally. These results are most likely related to the method of normalizing the data, which accentuated the relatively normal rel-CBF of these regions.

DISCUSSION

Although most neuroimaging studies have found either a temporal^16,17,22 or a frontal^18,19 lobe abnormality associated with delusions in AD, our findings suggest that both a frontal and a temporal lobe dysfunction are necessary for the development of psychotic symptoms in general, including delusions. Our findings further suggest that for the development of hallucinations, AD patients must also have an abnormality in the functions of the right parietal cortex.

These findings converge with neuropsychological and neuropathological studies that found temporal and frontal lobe abnormalities in AD patients with psychosis. Neuropsychological studies have shown a greater frontal lobe dysfunction in AD patients with delusions³⁶ and a greater temporal–parietal lobe dysfunction in AD patients with psychosis (e.g., delusions and visual and auditory hallucinations)¹⁴ than in AD patients without these features. Neuropathological studies have found more neurofibrillary tangles in the superior middle frontal lobe and more senile (neuritic) plaques in the temporal prosubicular area in AD patients with psychosis than in those without psychosis.³⁷

Aggressive behavior was associated with frontal lobe dysfunction, especially on the right. Because AD patients with aggressive behavior have been shown to have a greater frontal lobe dysfunction than those without aggression,^21,38,39 our finding would suggest there is an even greater vulnerability of the frontal association areas in AD patients with both psychosis and aggressive behavior. Moreover, as mentioned above, patients with psychotic symptoms have a higher frequency of aggressive behavior than those without psychosis.^14,24–26 These associations may affect neuroimaging studies of psychosis in AD; that is, including a greater number of patients with both aggressive behavior and psychosis may increase the probability of finding hypofrontality associated with psychosis.

Clearly, with a small (albeit well characterized) sample of AD patients with hallucinations and delusions, the current findings are only suggestive. However, the association between parietal and frontal–temporal blood flow reductions in the etiology of delusions or hallucinations is consistent with a variety of behavioral neurological data. The question of whether a combined frontal-temporal system dysfunction is a precondition for the development of the psychotic phenomenon in AD bears careful examination. The results of such a study would have implications not only for the etiology of psychotic phenomena, but also for their possible pharmacological management.

ACKNOWLEDGMENTS

This study was funded by Grants AG03705 and AG05133 from the National Institute on Aging and by Grants MH57078, MH49936, and MH01621 from the National Institute of Mental Health. J.T.B. is a recipient of the Research Scientist Development Award (Level II) (K02-MH01077).

TABLE 1. Functional neuroimaging studies of psychosis in Alzheimer's disease

Enlarge table

TABLE 2. Demographic and neuropsychiatric characteristics of Alzheimer's disease patients with and without psychosis

Enlarge table

TABLE 3. Relative regional cerebral blood flow in Alzheimer's disease patients and nondemented control subjects

Enlarge table

TABLE 4. Relative regional cerebral blood flow (rel-CBF) in Alzheimer's disease patients with and without psychosis

Enlarge table