Abstract
Background: Dementia is a frequent complication of idiopathic parkinsonism or PD, usually occurring later in the protracted course of the illness. The primary site of neuropathologic change in PD is the substantia nigra, but the neuropathologic and molecular basis of dementia in PD is less clear. Although Alzheimer’s pathology has been a frequent finding, recent advances in immunostaining of α-synuclein have suggested the possible importance of cortical Lewy bodies (CLBs) in the brains of demented patients with PD.
Methods: The brains of 22 demented and 20 nondemented patients with a clinical and neuropathologic diagnosis of PD were evaluated with standard neuropathologic techniques. In addition, CLBs and dystrophic neurites were identified immunohistochemically with antibodies specific for α-synuclein and ubiquitin; plaques and tangles were identified by staining with thioflavine S. Associations between dementia status and pathologic markers were tested with logistic regression.
Results: CLBs positive for α-synuclein are highly sensitive (91%) and specific (90%) neuropathologic markers of dementia in PD and slightly more sensitive than ubiquitin-positive CLBs. They are better indicators of dementia than neurofibrillary tangles, amyloid plaques, or dystrophic neurites.
Conclusion: CLBs detected by α-synuclein antibodies in patients with PD are a more sensitive and specific correlate of dementia than the presence of Alzheimer’s pathology, which was present in a minority of the cases in this series.
A close association between disturbed cognition (including dementia) and parkinsonism has been noted by clinical investigators for more than a century.1-3 Selective degeneration of the large pigmented, dopamine-producing neurons in the zona compacta of the substantia nigra (SN) in idiopathic parkinsonism or PD is considered to be the primary neuropathologic correlate of motor impairment,4 but the cellular and molecular basis of the dementia of PD is less clear.
The Lewy body (LB) is a distinctive intraneuronal inclusion of complex biochemical composition found in surviving neurons of the SN of the brain in PD and is regarded as a diagnostic hallmark that distinguishes PD from other parkinsonian disorders.5 LB in the cerebral cortex have been identified by conventional histochemical stains for decades, but unlike classic nigral LB, cortical LB (CLB) have a less distinct morphology and are poorly visualized by these methods. In the last decade, the development of newer immunohistochemical techniques for identifying LB throughout the brain has been followed by numerous reports describing demented parkinsonian patients in whom widespread LB were found in the cerebral cortex and brainstem.6,7 A dementia syndrome originally termed diffuse LB disease (DLBD) because of abundant CLB8,9 was recently renamed dementia with LB (DLB) by a consensus conference,10,11 and is now recognized to be the second most common cause of dementia after AD.12
A potential overlap between PD and AD was highlighted in the early 1990s with the proposition that a mixture of AD and LB neuropathologic changes represents a distinct subtype of AD called the LB variant of AD (LBVAD).7,13 Although the nature of this mix of neuropathologic changes is still debated,12,14 one popular concept is that LBVAD is situated midway along a neuropathologic continuum, with PD and AD at opposite poles.15 Moreover, growing evidence that CLB (with or without AD neuropathologic changes) can account for the presence of dementia in the setting of parkinsonism adds an extra dimension to the debate regarding a possible common pathogenesis for AD and PD.15,16 Notwithstanding the valuable contribution of immunostaining to a more complete understanding of the histopathology of the parkinsonian brain, the neuropathologic correlates of dementia in patients with PD are not yet definitive.
In 1997, a mutated gene on chromosome 4 that codes for the presynaptic protein α-synuclein was reported in a large Italian family with dominantly inherited, early onset PD.17 This discovery quickly led to the use of anti–α-synuclein antibodies and the demonstration that CLB detected with these antibodies are more prevalent than previously shown by ther labeling methods, thereby implicating α-synuclein as a major component of LB.18-20 Strong α-synuclein immunoreactivity also has been demonstrated in LB of the SN in PD.18
Most of the literature on clinicopathologic correlations in PD with dementia is retrospective. Few studies have looked prospectively at a population of patients with levodopa-responsive PD to correlate clinical status (including the presence or absence of dementia) with brainstem and cortical neuropathologic changes using modern immunostaining techniques. In this report we describe the neuropathologic findings in a prospective series of patients with typical PD, approximately half of whom developed dementia during the long course of the illness. Sections from diverse cortical and subcortical regions were probed with LB-sensitive antibodies to ubiquitin and α-synuclein, and with thioflavine S to detect neurofibrillary tangles (NFT) and senile plaques (SP).
Methods.
Patients.
Between 1985 and 1997, autopsy examinations were performed on 42 patients with a clinical diagnosis of PD who were cared for continually and closely by three neurologists (H.H., S.G., and M.B.S.) on the staff of the PD and Movement Disorders Center (PD & MDC) of the University of Pennsylvania. Thirty-six of the 42 patients were followed throughout the course of the illness by one neurologist (H.H.). Each patient had at least three of the four cardinal signs of PD (rest tremor, bradykinesia, rigidity, and postural instability). Except for one patient, who had tremor-dominant parkinsonism beginning at age 87 and who died at age 99, all of the patients responded strongly to levodopa/carbidopa therapy, and many developed the motor fluctuations with abnormal involuntary movements that typically occur with chronic use of levodopa. Therefore, all of the patients met the strict criteria for the clinical diagnosis of PD established by the United Kingdom Brain Bank.21 Almost all of the patients continued to use levodopa throughout the course of the illness.
Twenty-two of the 42 patients became clinically demented after the onset of motor symptoms of parkinsonism (PD with dementia [PDD]). Other parkinsonian patients in whom clear signs of an emerging dementia were evident less than 2 years after the onset of the motor signs of the illness were excluded from this study to make certain not to include any patients with a clinical diagnosis of DLB, as defined by consensus guidelines of the Consortium on DLB (i.e. motor signs of parkinsonism and significant cognitive decline occurring contemporaneously or within 1 year of one another).10 An accurate evaluation of mental status in the setting of PD is often confounded by severe motor disability, cognitive slowing (bradyphrenia), and the adverse effects of antiparkinson drugs, which frequently produce visual hallucinations, delusions, and confusion, even in nondemented people. A clinical diagnosis of dementia was made by the treating neurologist’s best global clinical impression, using criteria from the Diagnostic and Statistical Manual of Mental Disorders, 4th ed. and information supplied by the patient’s spouse or closely involved caregiver. Neuropsychological testing (Mini-Mental State Examination or more extensive test batteries) was performed in unselected patients, usually earlier in the course of the declining mental status. No attempt was made to subclassify the dementias into cortical or subcortical subgroups. In all patients, the clinical diagnosis of dementia was made before death and was not influenced by the results of the autopsy. For each patient whose advancing disability prevented him or her from returning to the PD & MDC for regular clinical evaluations, the spouse or nearest relative of the patient was interviewed by telephone to determine the cause of death and to estimate the patient’s final physical and mental status. In some of these cases, dementia developed during the interval of absence between last visit to the PD & MDC and death, often in a nursing home. Additional evaluation of cognitive status was conducted through chart reviews of all cases by two experienced clinicians (C.C. and G.G.), who were blind to the neuropathologic diagnosis and rated each patient as demented or not demented. The cause of death was not established in every case, but most patients died of aspiration pneumonia or inanition; several died of cancer or myocardial infarction. At the time of death, all 20 nondemented PD patients were in Hoehn and Yahr stages three to five of disability. All of the 22 demented patients had severe cognitive impairment and were categorized in Hoehn and Yahr stages four and five due to the combined effects of motor and mental disabilities.
Postmortem diagnosis and analysis.
All autopsies were performed at the Hospital of the University of Pennsylvania within 24 hours of death. Postmortem examination of the brain followed previously described protocols22 to identify macroscopic and microscopic evidence of disease. Briefly, this included gross examination of the whole and coronally sectioned brain. Tissue samples were obtained from multiple cortical and subcortical structures and stained for histologic examination with hematoxylin-eosin for standard diagnostic evaluation, thioflavine S for AD pathology, and Ubi-1 (Zymed Labs Inc., South San Francisco, CA) for ubiquitinated LB, dystrophic hippocampal (CA2/3) neurites (DN), and other intracellular inclusions. All staining and immunolabeling was conducted on 6-μm–thick, deparaffinized sections that had been fixed for 24 to 48 hours in ethanol (70% EtOH, 150 mm NaCl), 10% neutral buffered formalin, or Bouin’s fixative. In addition to the major pathologic changes of neurodegeneration, other findings were observed in three patients with dementia (minor état criblé; resolved microscopic infarct in right midfrontal cortex; fresh 2- to 3-week-old right parietal lobe infarct) and four patients without dementia (isolated small metastatic adenocarcinoma to right cerebellum; resolving bilateral small subdural hematomas; resolved lacunar infarct in the left putamen; mild neuron loss and gliosis in hippocampus and cerebellum consistent with prior anoxia). However, none of these incidental lesions could be implicated as meaningful contributors to mental status.
The densities of CLB, NFT, and SP were graded on a semiquantitative 0 to 3+ scale in the amygdala, hippocampus, subiculum, entorhinal cortex, and four neocortices, including the midfrontal gyrus, postcentral gyrus, Wernicke’s area, and cingulate gyrus. Ratings were assigned after counting these lesions in randomly selected 100× microscopic fields as follows: for CLB and NFT: 0 = 0, 1+ = 1 per 100× field, 2+ = 2 to 4, and 3+ = ≥5; for SP: 0 = 0, 1+ = 1 to 10, 2+ = 11 to 20, and 3+ = ≥21. In addition to these pathologic markers, neuron loss and gliosis were graded in the SN on a 0 to 3+ scale (0 = none; 1+ = mild; 2+ = moderate; 3+ = severe). The presence of ubiquitinated DN in areas CA2/3 of the hippocampus was assessed and similarly graded on a 0 to 3+ scale. Global whole number ratings of CLB, NFT, and SP were calculated by averaging the densities of each of these lesions in the ventromedial temporal lobe regions, and then averaging the ventromedial temporal lobe score with those of the other cortical regions, giving equal weight to each.
The neuropathologist was blind to the cognitive status of each patient during the semiquantitative assessment of CLB, NFT, SP, and DN. A neuropathologic diagnosis of PD was made if there was substantial loss of neurons and LB in SN, a low CLB count (global rating 0 or 1+), and NFT and SP counts that were insufficient for a diagnosis of AD, according to recently revised criteria for a neuropathologic diagnosis of AD (see below and reference 43). A neuropathologic diagnosis of DLBD was made if the CLB ratings were 2+ or 3+, irrespective of whether neuropathologic changes (usually severe) were present in SN. A neuropathologic diagnosis of PD plus AD was made if criteria for both PD and AD were met, and a neuropathologic diagnosis of LBVAD was made if the patient met criteria for both DLBD and AD.
α-Synuclein immunolabeling.
We performed immunohistochemistry, using the monoclonal antibody [LB509] to α-synuclein,20 and the densities of α-synuclein immunoreactive CLB were graded in the ventromedial temporal lobe regions (i.e., hippocampus, amygdala), midfrontal gyrus, anterior cingulate gyrus, postcentral gyrus, and SN. Similar global rating scores were derived from the results of ubiquinated CLB staining. Densities of α-synuclein positive DNs were rated in a similar manner.
Statistical analysis.
Analysis of the association between individual neuropathologic markers and the presence of dementia began with the calculation of sensitivities and specificities and with simple logistic regressions. Significance was based on t-statistics from the logistic regressions. To determine whether using more than one pathologic marker increases the accuracy of prediction of clinical diagnosis, we performed a series of logistic regressions, including all possible pairs of markers. t-Statistics were used to determine whether individual markers significantly contributed to the identification of cases. The samples of demented and nondemented patients were compared on continuous variables (age at onset, age at death, and duration of disease) using t-tests for differences between the means.
Results.
Table 1 presents the characteristics of the sample. There are no significant differences between those with and those without dementia by age at onset, age at death, or duration of disease at time of death. The sex ratios of the two groups are virtually identical.
Demography of 42 patients with PD, PD with dementia (PDD), and Lewy body pathology
Comparison of α-synuclein with ubiquitin immunostaining showed that α-synuclein was more sensitive and equally specific as a marker and correlate of dementia in the brains of patients with PD (see below). Therefore, the tabulated results that follow are derived from brain tissue samples stained with α-synuclein antibodies.
Table 2 shows the distribution of patients with the clinical diagnoses of PD (n = 20) or PDD (n = 22) among the three pathologic diagnoses (PD, LBVAD, DLBD). Ninety percent of those patients classified as having PD without dementia had a pathologic diagnosis of PD (no AD neuropathologic changes and an α-synuclein CLB score of 0 or 1+). Conversely, only 10% of patients with the neuropathologic changes of PD but without cortical neuropathologic changes (i.e. CLB, NFT, SP) were judged during life to be demented. In contrast, 91% percent of patients diagnosed as demented had the cortical neuropathologic changes of dementia (CLB score of 2 to 3+, NFT or SP by Consortium to Establish a Registry for AD [CERAD] criteria), whereas 9% did not. Patients with a pathologic diagnosis of DLBD developed parkinsonism at a younger age than those with PD or LBVAD, but the differences were not significant.
Pathologic diagnoses in 42 patients with clinical PD and PD with dementia (PDD)
The four principal cortical lesions in this sample were α-synuclein–positive CLB and DN, and thioflavine S–stained NFT and SP. Table 3 shows that 91% of the patients with PDD had a CLB score of 2 to 3+ (sensitivity), whereas 90% of the PD patients without dementia had a CLB score of <2+ (specificity). Ubiquitin staining for CLB was equally specific (90%) but much less sensitive (68% for a score of 2+ or higher).
Cortical Lewy body score in PD with and without dementia
When compared with the three other markers, α-synuclein CLB stands out as the best histopathologic correlate of dementia, as indicated by the sensitivity and specificity measures in table 4. A score of 2 or greater for any of the four measures is highly sensitive (less than 5% of the patients with dementia had a score of 0 or 1+ for any marker), but the specificity of any marker is only 55% (i.e., 45% of those without dementia had a score of 2+ or greater). Sensitivity and specificity of the α-synuclein marker for CLB increased with age. In the 21 patients 75 years and older, sensitivity and specificity were 100%, but only 85% and 81% in patients younger than 75. Logistic regression analysis of the power of a combination of markers to detect dementia shows no improvement over CLB alone as the best histopathologic correlate of dementia. The predictive values of NFT, SP, or DN were not significant when added to a logistic regression that already included CLB.
Pathologic markers of dementia in PD: sensitivity and specificity of clinical correlations
SN was severely affected in all cases (2+ and 3+ loss of pigmented neurons), irrespective of mental or motor status. As a result, no clinicopathologic correlations could be made.
Comprehensive assessment of the validity of four commonly used histopathologic markers of cortical neuropathologic changes in the setting of PD shows that an α-synuclein CLB score of 2+ or greater is the single best correlate of dementia (OR 31.5; 95% CI 4.2 to 150.4).
Discussion.
The earliest observers were impressed by the high frequency of dementia in the setting of parkinsonism. Frederich Lewy reported in 1923 that 54 of the 70 parkinsonian patients in his clinical series were demented.1 Severe loss of cognitive function and the frequent occurrence of dementia, usually later in the course of PD, have become more widely recognized in the last two decades for several reasons. First, the introduction of levodopa in the late 1960s dramatically changed the natural history of PD by greatly improving motor function and extending the life span of the typical parkinsonian patient who responded well to levodopa.23 Consequently, improved longevity equated with a greater probability of developing dementia. Second, parkinsonian patients with improved motor function were more easily testable with increasingly sophisticated neuropsychological techniques. Third, reports of the frequent occurrence of AD in brains of patients who died with the diagnosis of PD24,25 began to appear shortly after levodopa pharmacotherapy had become standard, but few investigators had reported the association of these two pathologies before levodopa. Fourth, the possibility that levodopa’s potential for generating membrane-toxic oxygen free radicals has stirred interest and concern that dementia might be one of the many adverse clinical consequences of this drug-induced molecular process.26 However, clinical evidence to support a causal connection between levodopa and dementia is lacking.
Defining the pathologic basis of dementia in PD has become more complicated in the last decade as immunostaining has made CLB easier to identify. Varying numbers of CLB have been found in the cortex of the majority of patients dying with PD, with or without dementia.27-30 The creation of LBVAD7,13 as a separate diagnostic category underscored the importance of immunostaining as a histopathologic tool as well as the common overlap between DLB and AD in demented patients, with or without parkinsonism.
The finding that α-synuclein is present in LB has rapidly led to the application of α-synuclein immunostaining to the descriptive anatomy of a variety of degenerative neurologic disorders, including PD, multisystem atrophy, and Down’s syndrome.18,31-33 Our results comparing α-synuclein to ubiquitin staining show that α-synuclein is a more sensitive marker for CLB in PD with an approximately equal specificity. This is consistent with a recent report that compared the two immunostaining techniques in a study of DLB.34 Moreover, there is a direct relationship between the number of α-synuclein CLB identified by our semiquantitative method and the presence of dementia. α-Synuclein is also a more rational marker of LB than ubiquitin because it is a principal component of LB and because of the genetic link of α-synuclein to several families with autosomal dominant PD.17,34,35 Thus, α-synuclein may be an intrinsic molecular component of the abnormal pathway that accounts for the onset and progression of PD.
The nature of any relationship between PD and AD continues to be controversial. The significance of data supporting an etiologic link between the dementia occurring in PD and the pathology of AD15,27,36,37 is often disputed by advocates of data showing that the connection is weak or merely coincidental.14,38 Some of the inconsistency may be due to inadequacies of older neuropathologic criteria for a diagnosis of AD, which emphasized SP rather than NFT.39,40 Because pathologists and investigators of aging have long recognized that SP may be found in abundance in the brains of elderly subjects without cognitive impairment,41,42 the newer Reagan criteria emphasize the quantitative importance of both SP and NFT in the neuropathologic diagnosis of AD.43 Moreover, recent clinicopathologic data favor neurofibrillary neuropathologic changes as better markers of dementia in AD than SP37,44; we also found that NFT found in the brains of PD patients are a reliable correlate of dementia in those rare cases in which they occur in abundance, whereas SP are not.
Some investigators have explored the possibility that CLB, like SP and NFT, correlate quantitatively with the degree of dementia in PD, although consensus is as deficient for this point of view as it is for the relative importance of plaques or tangles in AD. For example, two studies of postmortem tissue from patients with typical PD using antibodies to ubiquitin found that CLB were present in all patients, irrespective of cognitive status,27,29 whereas two other studies noted a definite correlation between the total number of CLB and cognitive impairment.28,45 Our results, showing a correlation between prevalence of CLB and dementia in PD using antibodies to α-synuclein as a marker, clearly agree with the second pair of reports and are in line with the results of another report46 showing that concentrations of CLB correlated significantly better with severity of dementia in patients with LBVAD than the presence of lesions of AD. However, perhaps as an indication that the etiologic role of CLB in DLB remains unresolved, one important observer has noted that a threshold for CLB density that distinguishes between demented and nondemented patients in PD remains undefined.47
Several studies of DLB, especially those that focus on LBVAD, have concluded that a combination of the cortical pathologies of AD and LB, rather than either one alone, is necessary to exceed the threshold of clinical expression of dementia. This view is supported by the frequent finding of an abundance of SP without NFT in patients whose brains contain CLB and AD neuropathologic changes, as defined by the National Institute on Aging and CERAD. The concept that Alzheimer pathology in LBVAD is a “plaque only” disorder48 suggests that LBVAD may occupy a distinct place on a spectrum of neurodegenerative diseases of the CNS and may not be simply a co-occurrence of PD and AD in the same individual, as some investigators have concluded.33,49 Yet many experts still adhere to the view that dementia in PD is more strongly influenced by the presence of AD neuropathologic changes than any other factor,26,27,36 because of the consistent finding that approximately half of all patients with PD and dementia who are autopsied have AD neuropathologic changes or no distinctive dementia-related neuropathologic changes. However, most histopathologic studies of dementia in PD were published in the era before immunostaining was available to identify CLB.26,36 As a result, these reports clearly underestimated the frequency and importance of CLB and, conversely, emphasized AD neuropathologic changes to the exclusion of CLB in PD. Our results, using sensitive and specific immunohistochemical labeling, show that the dementia of PD is less closely associated with Alzheimer neuropathologic changes than was previously thought and is more likely caused by an unexplained spread of LB neuropathologic changes from brainstem to a variety of locations in the cerebral cortex. The cause of propagation of CLB throughout the cerebral hemispheres in patients with PD who become demented is unknown.
DN in the CA2-3 region of the hippocampus, another lesion reported to be associated with DLB,50 were commonly observed in our sample but were not a strong correlate of dementia.
Our results support three conclusions related to the morbid anatomy of the dementia associated with PD. First, LB neuropathologic changes, as demonstrated by α-synuclein immunostaining, rather than the neuropathologic changes of AD, are the most important pathologic correlate of dementia occurring during the natural course of idiopathic PD. LBVAD was an infrequent finding in our series of patients, and there were no cases of PD with the neuropathologic changes of AD without the presence of CLB. Second, we were unable to confirm a significant increase of DN as a DLB marker that distinguishes demented from nondemented subjects. Third, a high count of NFT, although an infrequent finding, was a strong predictor of dementia, whereas a high count of SP was not. Finally, our data also show that a clinical diagnosis of PD, as measured against the “gold standard” of pathologic diagnosis, is highly accurate if patients are evaluated by accepted clinical criteria and are observed longitudinally by the same treating neurologist.26,51
Acknowledgments
Supported by National Institutes of Health Grants P01-AG09215, AG10124, and MH55199, and a grant from the Alzheimer’s Association.
Acknowledgment
The authors thank Ms. Danielle Lavalla and Ms. Theresa Schuck for expert technical assistance.
- Received July 28, 1999.
- Accepted in final form February 16, 2000.
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
