Report of the second dementia with Lewy body international workshop

Diagnostic accuracy.

Only two published studies have reported sensitivity and specificity of the consensus clinical criteria against autopsy findings. Mega et al.4 correctly identified 75% of DLB cases, with a specificity of 79%, by retrospectively applying the criteria to the charts of autopsy confirmed cases. Holmes et al.5 found a specificity of 100% but a sensitivity of only 22% in a community-based dementia study with autopsy confirmation. This high specificity was confirmed in five out of six new studies presented in the workshop, indicating that the consensus criteria are appropriate for confirmation of diagnosis (few false positives) but may be of limited value in screening for DLB cases (high false negative rate). Clinical underdiagnosis of DLB remains a problem in all but a few highly specialized centers, AD being the most frequent misdiagnosis of autopsy-confirmed DLB cases.6 A false positive diagnosis of vascular dementia is less frequent and when made may be associated with concomitant cerebrovascular disease.7 Up to 30% of DLB cases are now reported as having additional microvascular pathology or ischemic lesions at autopsy.5,8

Clinical assessment.

It was resolved that the clinical consensus criteria, in their current format, have not yet been in use long enough to be adequately assessed and it was therefore recommended that no modifications should be made at present. Research efforts should be focused on increasing case detection rates by developing more sensitive assessment tools, and these must be shown to be reliable across a range of clinical settings.

With respect to the core clinical features, visual hallucinations and extrapyramidal features are symptoms that can be assessed with adequate inter-rater reliability,9,10 but the precise characteristics of the extrapyramidal movement disorder in DLB need to be more fully investigated and contrasted with the motor manifestations of PD.11

Because cognitive fluctuations are so characteristic of DLB patients and occur in the majority, further efforts need to be made to characterize these in terms of their time course, persistence, and magnitude. No operationalized systems for assessing cognitive fluctuations have been devised, although the original consensus paper3 gives a detailed clinical description of some of the salient characteristics. The current gold standard for determining fluctuation is “expert opinion,” but inter-rater reliability is considered to generally low, with kappa values of only 0.25 to 0.36.4 A variety of complementary techniques were proposed to improve this situation, including 1) the use of questionnaires and diary keeping by a reliable informant and 2) the use of specific psychometric procedures that can discriminate DLB and AD,12,13 including computer-based tasks that are sensitive to the attentional dysfunction characteristic of DLB.14

Additional clinical features.

Two areas of clinical symptomatology that were not considered in the original guidelines may be supportive of a diagnosis of DLB. REM sleep behavior disorder appears to be common,15 probably reflecting another facet of the fundamental disturbance in central regulation of consciousness in DLB. Sleep questionnaires and polysomnography need to be further investigated for their utility in differential diagnosis and guidelines developed for the optimal clinical management of sleep disorder. Depressive symptoms are reported in 33 to 50% of DLB cases,16,17 a rate higher than in AD and similar to PD. There are no data on response to antidepressant treatments in DLB.

Pathologic assessment and classification.

In many brains of patients with DLB (55%) there is accompanying AD pathology in the form of neocortical diffuse plaques, neuritic plaques that generally lack tau-positive neurites, and modest numbers of medial temporal lobe neurofibrillary tangles (Braak Stages III/IV).18 In 15% there are severe AD changes and 30% have no more AD pathology than age-matched controls. The 1996 consensus paper recommended that the precise terminology used by individual groups in relation to AD pathology be considered less important in practice than the need to establish a common protocol for assessing and evaluating Alzheimer-type pathologic lesions in DLB (see reference 19 for review). The Consortium to Establish a Registry for Alzheimer’s Disease protocol continues to be recommended in the consensus criteria. The newly adopted Reagan Institute criteria for pathologic diagnosis of AD20 has moved toward a requirement for extensive neocortical neurofibrillary tangle and tau pathology, and the majority of DLB cases will therefore not, by such criteria, be considered to have concomitant AD. This conclusion is at variance with some of the data on genetic influences in DLB (see following).

Clinicopathologic correlation.

The precise pathologic correlates of dementia in DLB are not well defined. It has been suggested that a “cerebral type” of LB disease21 exists, in which LBs occur in the cerebral cortex earlier than in the brainstem nuclei, explaining why dementia frequently precedes parkinsonism in DLB. Even in severely demented cases, cortical LB density is low, compared, for example, with plaque and tangle counts in AD. Samuel et al.22 reported that midfrontal neocortical LB numbers, modified Braak stage of neurofibrillary tangle burden in the entorhinal cortex, midfrontal neocortical neuritic plaque density, and loss of the cholinergic enzyme choline acetyltransferase were all correlated with global dementia severity assessed by Mini-Mental State Examination score. Another study found no such correlates except among measures of cortical cholinergic deficits, severity of cognitive dysfunction, and presence of visual hallucinations.23 The majority of cases (69%) in a series of 29 autopsy-confirmed cases reported at the workshop fulfilled consensus criteria for neocortical type DLB, 24% for limbic DLB and only 7% brainstem DLB.8 The distribution of LBs and associated neuronal loss did not appear to bear a simple relationship with the clinical profiles of these cases.

At the time that the original recommendations for pathologic examination were made,3 conventional hematoxylin & eosin staining for LBs was regarded as acceptable for routine diagnostic purposes, although antiubiquitin immunocytochemical staining was acknowledged as significantly more sensitive. Since the first workshop, α-synuclein antibodies have been shown to label purified LBs24 and both LBs and Lewy neurites (LNs) in situ. LNs constitute an important component of the pathology of DLB, consisting of abnormal filaments similar to those found in LBs and with a similar immunohistochemical staining profile. Double staining for α-synuclein and ubiquitin identifies both antigens in most LBs and LNs. Ubiquitin antibodies frequently only stain the halo of brainstem-type LBs whereas α-synuclein antibodies stain both the halo and the core. Some LBs and LNs are immunoreactive for α-synuclein but not for ubiquitin, with α-synuclein–positive neurites outnumbering those immunoreactive for ubiquitin. These findings are consistent with the polymeric assembly of α-synuclein filaments leading to LB formation and subsequent ubiquitination and accumulation of neurofilaments.25 α-Synuclein immunohistochemistry is thus potentially the most sensitive and specific technique for detecting and quantifying the clinically and pathologically relevant lesions in DLB. Opinion at the workshop was divided as to whether the pathologic guideline should be modified at this stage to take account of these immunocytochemical developments. It was concluded that α-synuclein–based methods are likely to be useful in research laboratories, particularly with respect to clinicopathologic correlative studies. However, for classification purposes in research and clinical practice, ubiquitin immunocytochemistry should remain the first method of choice for the detection of LB, particularly because this is only now being widely adopted. Most research groups were considering using both antibodies in parallel to judge their relative merits and provide new data for the next workshop. The structural similarities between the apolipoproteins and the synucleins may prove important in understanding the cellular functions of these proteins and their role in the neurodegenerative mechanisms underlying these overlapping conditions.

Genetics.

Although recent studies suggest the genetic overlap between clinical AD and PD is limited, cortical LBs are found in most cases of familial AD, increasing in density with longer disease duration. Certain APP mutations precipitate LB formation although parkinsonism is not a prominent clinical feature. It was also noted that LBs occur in some older patients with trisomy 21. Thus similar pathogenic mechanisms may contribute to the formation of AD and LB pathology. There are numerous reports that AD and DLB patients share an increased prevalence of the apolipoprotein ε4 allele on chromosome 19.26 ε4 Is also increased in PD patients with dementia, but not in those without.27 The cytochrome P459 (CYP2D6) allele on chromosome 22 is increased in PD compared with AD and controls but early reports of an increase in DLB26 have not been consistently reproduced. Mutations in the presenilin-1 gene are not associated with DLB and mutations in the α-synuclein gene on chromosome 4, which are a rare cause of autosomal dominant PD, do not appear to precipitate dementia.

Neuroimaging.

Functional brain imaging using PET ¹⁸F-deoxyglucose and SPECT ⁹⁹Tc^m-HMPAO have not been shown to be of particular value in the differential diagnosis of patients with DLB and AD. ¹²³I-FP-CIT, a fluoroaylkyl analogue of cocaine, binds to the presynaptic dopamine transporter system, mapping the presynaptic dopaminergic terminals of the nigrostriatal pathway in vivo. Preliminary data were presented at the workshop suggesting that ¹²³I-FP-CIT SPECT may be useful in separating DLB from AD, the latter disorder generally showing no nigrostriatal degeneration, particularly in the early stages.

Although structural brain imaging has not yet contributed significantly to the antemortem assessment of patients with DLB, an MRI study28 has reported that medial temporal lobe and hippocampal atrophy are less prominent compared with AD. Periventricular lesions and white matter hyperintensities may be present both in DLB and AD.29

Clinical trials.

It was agreed that larger, randomized clinical trials should be conducted to establish the optimum treatment of DLB patients. The extent of levodopa responsiveness in DLB needs to be compared to PD, and the antipsychotic, behavioral, and cognitive effects of cholinergic therapies evaluated. Case selection and recruitment remains a potential problem because of the tendency of existing clinical criteria to result in underdiagnosis of DLB. Novel trial designs will be required, differing from those that have been used to evaluate anti-AD effects. As trials are established, appropriate primary and secondary outcome measures will need to be defined—specific aspects of cognition and relevant neuropsychiatric features are the most likely primary treatment targets. Appropriate measures of attention, psychosis, and behavioral disturbance and motor disability will need to be validated in this population. Existing measures in activities of daily living may be confounded by the synergistic effects of mental and motor disability. Clinical global assessment tools will probably need only minor modification.