Alterations of Striatal Dopamine Receptor Binding in Alzheimer Disease Are Associated With Lewy Body Pathology and Antemortem Psychosis FREE

ALZHEIMER DISEASE (AD) is currently estimated to affect 3 to 4 million individuals in the United States. Psychotic symptoms, delusions, and hallucinations occur in at least 30% to 40% of AD patients.¹ Patients with AD and psychotic symptoms demonstrate more severe cognitive deficits than matched AD subjects without psychosis.^{2 - 3} Similarly, AD subjects with psychotic symptoms are at risk for more rapid cognitive deterioration^{3 - 4} more rapid decline in function,⁵ and premature institutionalization.^{5 - 6}

Many individuals diagnosed ante mortem with AD will demonstrate cortical Lewy bodies (LB) at autopsy in addition to the neuropathologic findings of AD (AD with Lewy bodies, AD + LB). Alzheimer disease with Lewy bodies has been associated with excess cognitive burden, extrapyramidal symptoms, and with more frequent psychotic symptoms.⁷ Prominent reductions in postmortem measures of presynaptic dopamine have been reported in AD + LB.^{8 - 10} These reductions, however, are not as severe as seen in Parkinson disease.^{9 - 10}

A limited number of studies have examined dopamine receptor density in AD subjects who have been characterized with regard to LB. Dopamine D₁, D₂, and D₃ receptors, but not D₄ or D₅ receptors, are expressed in striatum.¹¹ Perry et al¹⁰ examined nonselective D₂/D₃ receptor binding using [³H]raclopride in AD, AD + LB, and normal control (NC) subjects. Neuroleptic-naive or neuroleptic-intolerant AD + LB subjects had reduced striatal D₂/D₃ receptor binding, while AD subjects did not differ from NC subjects. Whether the observed reductions in D₂/D₃ binding were due to a reduction in D₂ binding, in D₃ binding, or in both cannot be directly answered at this time. Because D₃ density varies along a dorsal to ventral gradient in striatum,¹² the absence of a significant interaction of striatal region with the diagnostic group in the examination of D₂/D₃ binding reported by Perry et al¹⁰ provides indirect evidence that both D₂ and D₃ binding were reduced.

Joyce et al¹³ examined D₁ and D₂ receptor binding in striatum of AD and NC subjects. Alzheimer disease was not associated with reduction in striatal D₂ receptor binding, unless the AD subjects also demonstrated extrapyramidal symptoms (EPS). D₁ binding in AD did not differ from controls, and there was no association with EPS. In contrast, D₁ binding is elevated in response to the dopamine deficit found in Parkinson disease.^{13 - 14} The hypothesis that D₁ binding is also elevated in response to the less extensive dopamine deficit seen in AD + LB has not been examined.

Although older studies reporting elevated D₂/D₃/D₄ receptor binding to be associated with schizophrenic psychosis were confounded by long-term neuroleptic exposure, recent evidence suggests the D₃ receptor may play a role in psychotic symptoms. There is a significant association of schizophrenia with homozygosity for a biallelic BalI/MscI polymorphism in the D₃ gene, although the magnitude of this association is modest.¹⁵ Similarly, we found homozygosity for this polymorphism to be associated with a modest increased rate of psychosis in subjects with AD.¹⁶ Increased striatal D₃ binding has been reported in subjects with schizophrenia who are not taking neuroleptics in comparison with NC subjects, with the largest differences in rostral-ventral striatum.¹² Neuroleptic treatment was associated with normalization of D₃ binding.¹² In contrast to schizophrenia, the association of D₃ receptor binding with psychosis in AD has not been examined.

We undertook, therefore, saturation binding studies of D₁, D₂, and D₃ receptors in neuroleptic-free NC, AD, and AD + LB subjects characterized with regard to their history of psychotic symptoms. We hypothesized that AD + LB would be associated with reduced density of both D₂ and D₃ receptors, with a corresponding increased D₁ density. We further hypothesized that increased D₃ density would be associated with a history of psychotic symptoms in both AD and AD + LB subjects.

The radioligands iodine [¹²⁵I](R)-trans-7-hydroxy-2-[N-propyl-N-(3′-iodo-2′-propenyl)amino]

tetralin ([¹²⁵I]trans-7-OH-PIPAT) and [³H]-SCH23390 were obtained from NEN Life Science Products, Boston, Mass. Ketanserin, (+)-butaclamol hydrochloride and PD-128907 were obtained from Research Biochemicals International, Natick, Mass. Ultima Gold scintillation fluid was obtained from Packard Instrument Co, Meriden, Conn. Anti-βA4 peptide was provided by the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) and purchased from Dr Henryk Wisniewski, New York State Institute for Basic Research, Staten Island. Antibodies to α-synuclein were provided by Dr Virginia M.-Y. Lee, University of Pennsylvania, Philadelphia.

Subjects were identified through the brain bank of the Alzheimer's Disease Research Center at the University of Pittsburgh. All studies were approved by the institutional review board of the University of Pittsburgh. Of the 27 AD and AD + LB subjects, 25 underwent complete neurologic, neuropsychologic, and psychiatric diagnostic evaluations at initial and annual evaluations as part of their participation in the Alzheimer's Disease Research Center. Details of these assessment protocols have been reported elsewhere.^{5 ,16 - 17} In addition, all available records were reviewed for evidence of delusions and hallucinations by 1 of 2 raters (R.A.S., M.T.H.). Interrater reliability for the classification of psychosis was established in 20 subjects and was high (κ = 0.70). Delusions and hallucinations were defined as previously described.^{16 - 17} No patient had a history of schizophrenia, schizoaffective disorder, or other idiopathic psychosis. Extrapyramidal symptoms were defined by the presence of bradykinesia, cogwheel rigidity, or resting tremor. Extrapyramidal symptom ratings were unavailable for 3 AD + LB subjects. Current medications were recorded at all Alzheimer's Disease Research Center visits. In addition, psychotropic medications used in the past (initial visit) or since the last evaluation (annual visits) were recorded. Medications used at the time of death and during the 3 months prior to death were reviewed, and the interval since last neuroleptic use was recorded.

Demographic and clinical information for the 8 NC, 13 AD, and 14 AD + LB subjects are presented in Table 1. All subjects were either neuroleptic naive (n = 30) or neuroleptic free for at least 1 month (n = 5). Lifetime duration of neuroleptic treatment did not differ between groups (exact multinomial, P = .80). Normal control subjects were significantly younger (mean [SD], 66 [14] years, F₂ = 5.5, P<.01), with nearly significantly longer mean [SD] postmortem intervals (6 [3] days, F₂ = 3.1, P = .06), and more often men (100%, exact multinomial P<.001) compared with AD subjects. There were no significant differences in these variables between the AD and AD + LB groups or in Braak score,¹⁸ a measure of neurofibrillary pathology (F₁ = .03, P = .90).

At the time of brain removal, postmortem interval was recorded and the brain was divided in the midsagittal plane. The right hemibrain was coronally sectioned at 1.0-cm intervals. The majority of the right nucleus accumbens in each case was dissected and divided into 3 frozen samples, 1 of which was used for this study. The head of the right caudate was sampled from the same section. All samples were stored at −80°C until assayed. The left hemibrain was fixed in 10% formalin, then sectioned coronally. The tissue sections examined, and the histologic and immunocytochemical methods used, followed CERAD protocols and have been described in detail elsewhere.¹⁷ Neuritic plaques, diffuse plaques, and neurofibrillary tangles were semiquantitatively scored and diagnoses established using CERAD criteria.¹⁹ A diagnosis of AD + LB further required multiple α-synuclein–positive LB in limbic and/or neocortical areas. Conversely, a diagnosis of AD indicated that no α-synuclein–positive LB were present in any area, including the substantia nigra.

For all assays, unwashed tissue was used to preserve the original tissue density of dopamine receptors.²⁰ Caudate and nucleus accumbens samples were ground into powder over liquid nitrogen. At onset of the study, samples were weighed, thawed, then homogenized and suspended in buffer the morning of each assay day. Later the homogenization step was performed immediately prior to addition to the assay tubes. This methodologic change was treated as a covariate in all statistical analyses. Total and nonspecific binding at each concentration of ligand were determined in triplicate.

D₁ assays were conducted as described by Hall et al²¹ with minor modifications. Total volume was 500 µL, with a final concentration of caudate tissue of 2.0 mg wet weight per milliliter. Incubation buffer was 50-mmol/L tris-hydrochloride (pH 7.4), 120-mmol/L sodium chloride, 1-mmol/L magnesium chloride, 2-mmol/L calcium chloride, 5-mmol/L potassium chloride, and 0.1% ascorbic acid. Ketanserin, 200 nmol/L, was added to block binding to serotonin_2A receptors. The ligand was [³H]-SCH23390, 0.25 to 8.0 nmol/L. Nonspecific binding was defined by 1-µmol/L (+)-butaclamol. D₂ assays were conducted in a total volume of 500 µL, with a final concentration of caudate tissue of 2 mg wet weight per milliliter. Incubation buffer was 50-mmol/L tris-hydrochloride (pH 7.4), 1-mmol/L EDTA, 120-mmol/L sodium chloride, 2-mmol/L magnesium chloride, 1.5-mmol/L calcium chloride, and 300-µmol/L guanosine triphosphate. The ligand was [³H]raclopride, 0.07 to 40.0 nmol/L. To block D₃ binding, 50 µL of 5.0-µmol/L PD128907 (final concentration, 0.50 µmol/L) was added to each tube. Nonspecific binding was defined by 1-µmol/L domperidone. Both D₁ and D₂ assays were incubated at room temperature for 60 minutes and terminated by placing tubes in an ice water bath. Samples were harvested onto filter strips (Whatman GF/B; Whatman Inc, Newton, Mass), which had been soaked in 0.5% poly(ethylenimine) solution, then washed with 12-mL ice cold 50-mmol/L tris-hydrochloride solution (pH 7.4). Filters were placed into scintillation vials with 4.0-mL scintillation fluid and counted for 5 minutes in a scintillation counter (Beckman LS-5801; Beckman, Fullerton, Calif).

D₃ assays were conducted adapting the methods Gurevich et al¹² and Pugsley et al²² for selective labeling of D₃ receptors, and used tissue from nucleus accumbens. Total volume was 100 µL, with a final tissue concentration of 10 mg wet weight per milliliter. Incubation buffer was 50-mmol/L tris-hydrochloride (pH 7.4), 0.50-mmol/L EDTA, 120-mmol/L sodium chloride, 1.0-mmol/L magnesium chloride, 1.5-mmol/L calcium chloride, and 300-µmol/L guanosine triphosphate. The ligand was [¹²⁵I]trans-7-OH-PIPAT, 0.125 to 8.0 nmol/L. Nonspecific binding was defined by 0.50-µmol/L PD128907. Under these conditions, binding to receptors other than D₃ (eg, D₂, serotonin_1A, or sigma sites) is eliminated.^{22 - 23} Assays were incubated at room temperature for 90 minutes, then terminated and harvested as described above. Filters were counted for 1 minute in a gamma counter (Titertek).

B_max and K_d were determined by nonlinear regression analysis of specific binding using GraphPad Prism software.²⁴ All binding studies were best fit by a 1-site model, which was confirmed by both statistical comparison of fits and by visual inspection of Scatchard transformations. Statistical analyses used Statistical Product and Service Solutions for Windows 9.0.0 (SPSS Inc, Chicago, Ill). K_d was transformed to pK_d (−logK_d) for all analyses. B_max and pK_d for each receptor were the dependent variables. The association of B_max and pK_d with diagnostic group was tested by analysis of covariance, with age, postmortem interval, and sex entered as covariates. A covariate indicating the tissue homogenization method (see above) was entered in all analyses. Tests of the associations of B_max and pK_d with psychosis and EPS used 2-way analysis of covariance with diagnosis and psychosis (or EPS) presence-absence entered as cofactors. Age, postmortem interval, and sex were not included in tests restricted to AD and AD + LB subjects, as these groups were matched on these variables.

The association of receptor densities and affinities with neuropathologic group is presented in Figure 1. There was a significant difference among groups in D₁ density (F₂ = 3.5, P = .04) and D₁ affinity (F₂ = 3.5, P = .04). Mean (SEM) covariate-adjusted D₁ densities were 34% higher in AD + LB vs AD subjects, 21.7 (1.5) vs 16.2 (1.6), respectively (P = .01). When D₁ affinity was included as a covariate, the association of D₁ density with diagnostic group was reduced but continued to demonstrate a significant trend (F₂ = 3.2, P = .06). D₂ density also demonstrated a trend toward a significant difference among groups (F₂ = 3.2, P = .06). Mean covariate-adjusted D₂ densities were 30% lower in AD + LB than AD subjects, 11.9 (2.0) and 18.8 (2.0), respectively (P = .02). D₂ affinity differed significantly among groups (F₂ = 4.1, P = .03), with significantly higher affinity in AD and AD + LB vs NC subjects (P = .03 and P = .009, respectively). After entering D₂ affinity as a covariate, the association of D₂ density with diagnostic group was significant (F₂ = 4.2, P = .03). D₃ density did not differ among groups (F₂ = 2.2, P = .10). However, D₃ affinity was significantly different among groups (F₂ = 4.3, P = .03). Post hoc comparisons revealed a significant difference between the AD + LB and NC groups (P = .02). After inclusion of D₃ affinity as a covariate, there remained no significant association of D₃ density with diagnostic group (F₂ = 2.0, P = .20).

There were no significant associations of density or affinity of any of the receptors with EPS in AD subjects. Similarly, neither D₁ density and affinity nor D₂ density and affinity were associated with psychosis in the AD subjects. In contrast, D₃ density was significantly elevated among subjects with psychosis (F₁ = 5.8, P = .03, Figure 2). Controlling for the presence of psychosis, there was also a significant reduction of D₃ density in the AD + LB compared with AD subjects (F₁ = 6.9, P = .02). In contrast, D₃ affinity was not associated with psychosis (F₁ = 1.2, P = .30), although a significant increase in D₃ affinity was present in AD + LB compared with AD subjects (F₁ = 11.0, P = .006). There was also a significant interaction of psychosis and diagnosis (F₁ = 10.5, P = .007). The associations of D₃ density with psychosis and diagnostic group were unaltered when D₃ affinity was entered as a covariate (F₁ = 5.7, P = .03 and F₁ = 5.0, P = .04, respectively). Mean [SD] covariate-adjusted D₃ densities were increased by 72% in subjects with psychosis (6.9 [1.0]) compared with those without psychosis (4.0 [0.7]). In contrast, mean covariate-adjusted D₃ densities were decreased by 52% in AD + LB subjects vs AD subjects, 3.5 (1.0) and 7.3 (1.1), respectively.

We identified 5 subjects who had been excluded from the above analyses due to recent neuroleptic use. The clinical characteristics of these patients are presented in Table 2. All neuroleptic-treated subjects were diagnosed as having AD + LB. They did not differ significantly in age, sex, race, Braak score, or history of psychosis from the 9 AD + LB subjects not taking neuroleptics in whom D₃ assays were conducted (Table 1). D₃ density and affinity for the AD + LB subjects treated with and without neuroleptics are presented in Figure 3. There was no effect of neuroleptic treatment on D₃ density (F₁ = 2.0, P = 0.2). However, D₃ affinity was significantly reduced in the neuroleptic-treated subjects (F₁ = 5.8, P = .03). In 1 subject treated with neuroleptics, neuroleptic use was 16 days prior to death. Exclusion of this subject had almost no effect on mean [SD] D₃ affinity in the neuroleptic-treated group (pK_d = 8.63 [0.3]), and the difference between groups in D₃ affinity continued to show a trend toward significance (P = .08).

Consistent with our hypotheses, the presence of LB in AD subjects was associated with increased density of caudate D₁ receptors, reduced density of nucleus accumbens D₃ receptors, and a trend toward reduced density of caudate D₂ receptors. Alzheimer disease with LB was also associated with an increased affinity of all 3 dopamine receptors, although the changes in receptor density were unaffected by the observed changes in affinity. In contrast to the association of altered receptor binding with AD + LB, D₃ receptor density was elevated in subjects with a history of psychosis, independent of the presence or absence of Lewy bodies.

The observed alterations in D₁ and D₃ density may result, in part, from a postsynaptic response to sustained striatal dopamine deficits, which have been observed in AD + LB.^{8 - 10 ,25 - 26} Although we did not measure striatal dopamine concentration directly, the observed increase in receptor affinities in AD + LB is consistent with reduced dopamine concentration in caudate and nucleus accumbens of these subjects.²⁷ In the prototype dopamine deficiency disorder, Parkinson disease, caudate D₁ density is typically up-regulated in untreated subjects by 40% to 50%.¹⁴ The effect of dopamine deficiency on the D₃ receptor is not well established, although recent evidence indicates that D₃ binding is down-regulated by about 40% to 50% in caudate and nucleus accumbens of patients with Parkinson disease.²⁸

In contrast to D₁ and D₃, the trend toward reduction in D₂ density is not consistent with a postsynaptic response to dopamine depletion. In Parkinson disease, caudate D₂ density is typically up-regulated in untreated subjects by 15% to 80%.^{14 ,27} The extent of dopamine deficiency in AD + LB is not as severe as is seen in Parkinson disease.⁹ Thus, it is possible that D₂ receptors are not up-regulated in response to moderate concentration decreases, possibly due to lower affinity for dopamine than D₁ and D₃.²⁹ Absent up-regulation, however, does not explain the observed reduction in D₂ density. We have described numerous α-synuclein–positive neurites in striatum of AD + LB.³⁰ Whether this pathologic process preferentially affects D₂ pathways is not known.

We also found that having accounted for the effect of LB presence on D₃ receptor binding, there was a significant increase in D₃ density in subjects with a history of psychosis. The increase in D₃ density in psychosis was present in both AD and AD + LB subjects and unaffected by D₃ affinity. Overall, subjects with psychosis had a 72% elevation of mean D₃ density compared with subjects without psychosis. The magnitude of this elevation is consistent with the report of a 55% to 110% increase in D₃ binding, measured by autoradiography, in ventral striatum of subjects with schizophrenic psychosis who were not taking neuroleptics.¹² Guided by our studies of the association of psychosis with genetic variation in the D₃ receptor, we had hypothesized that D₃ plays a permissive role in the onset of psychotic symptoms in subjects with an underlying neurodevelopmental or neurodegenerative condition.^{16 ,31} The findings of the present study, taken together with those of Gurevich et al,¹² provide support for this hypothesis in 3 disorders: schizophrenia, AD, and AD + LB. The magnitude of the association of psychosis with striatal D₃ density is substantially greater than that of the association of psychosis with homozygosity for the BalI polymorphism in the D3 gene. This suggests that the contribution of the BalI polymorphism to the determination of D3 density in brain is modest. Consistent with this interpretation, we did not find a significant association between D₃ density and genotype in these subjects (data not shown).

Whether elevated D₃ density in psychosis is a primary deficit or reflects a pathologic response to other disease-related processes cannot be determined. However, the elevations in D₃ density do not appear to be an artifact of neuroleptic treatment. In contrast to findings for the D₂ receptor in striatum, striatal D₃ receptor binding is not up-regulated by antipsychotic treatment in animals.³² Similarly, we found no effect of recent neuroleptic treatment on D₃ density in our subjects. Finally, there was no corresponding up-regulation of D₂ receptors in our psychotic subjects, indicating the specificity of our observation.

In contrast to the absence of neuroleptic effects on D₃ density, we found that D₃ affinity was reduced by neuroleptic treatment. This latter finding may explain an apparent discrepancy between the present study and that by Gurevich et al.¹² Using receptor autoradiography, they found that neuroleptic treatment at the time of death led to "down-regulation" of D₃ binding. Because autoradiographic binding is dependent on both receptor density and affinity, the reduced D₃ binding interpreted by Gurevich et al could have resulted from an affinity reduction. The mean D₃K_d in our subjects taking neuroleptics was 2.4 times higher than in subjects not taking neuroleptics (3.1 nmol/L and 1.3 nmol/L, respectively). This difference would yield an approximate 100% difference in D₃ binding at the 0.30-nmol/L ligand concentration used by Gurevich et al, a value close to the difference they observed in ventral striatum. Thus, the most parsimonious explanation of the effect of neuroleptic treatment on D₃ receptors in both AD and schizophrenia is that it acts to increase D₃ receptor apparent affinity. Whether this occurs through simple competitive inhibition or through an alternate mechanism (eg, posttranslational modification) remains unknown.

An important limitation of the present study is that most AD and AD + LB subjects had advanced dementia at the time of autopsy. We cannot conclude that the observed dopamine receptor changes are present in earlier stages of illness when the clinical profiles of AD and AD + LB diverge. Another limitation of our study is the inexact matching of the NC subjects, which may have reduced the magnitude of detectable difference between groups because of the need to enter covariates into the analyses. We did not find an association of decreased D₂ density with EPS in our subjects. This may have resulted from limited power due to the small number of subjects with EPS, rather than a lack of true association.^{10 ,13} Strengths of the present study include the ante mortem behavioral characterization of the subjects, examination of multiple receptor populations in the same subjects, and the use of saturation binding methods that provide separate information about affinity and density.

In conclusion, this study provides evidence for a distinct pattern of expression of striatal D₁, D₂, and D₃ receptors in AD + LB. Future studies would benefit from inclusion of a Parkinson disease comparison group. Further studies examining the D₃ receptor as a target of neuroleptic action are indicated.

Accepted for publication June 12, 2000.

This work was supported in part by grants MH01153 and AG15133 from the National Institutes of Health and grant M1998-003 from the John F. and Nancy A. Emmerling fund of The Pittsburgh Foundation.

Corresponding author and reprints: Robert A. Sweet, MD, Western Psychiatric Institute and Clinic, 3811 O'Hara St, Pittsburgh, PA 15213 (e-mail: Sweetra@msx.upmc.edu).