Abstract
The concept of dementia with Lewy bodies (DLB) has been slowly gaining momentum since 1961 when Okazaki published case reports about two male patients who presented with dementia aged 69 and 70 and who died shortly after with severe extrapyramidal rigidity. Autopsy showed Lewy body (LB) pathology in the cerebral cortex [1]. Over the next 20 years, a total of 34 such cases were reported, all by Japanese workers who adopted the term diffuse Lewy body disease to describe the typical distribution of LB in subcortical and cortical regions [2].
During the following decade, up to 20% of all elderly demented cases reaching autopsy were found to have cortical LB pathology [3,5], and in 1990 Hansen et al. reported that 36% of patients clinically diagnosed as Alzheimer's disease (AD) had LB at autopsy, the Lewy body variant of Alzheimer's disease (LBVAD) [6]. The significance of these reports was unclear: was this a new type of dementing disorder which had not previously existed; or had the LB previously been overlooked? Re-examination of original material from a cohort of autopsy material collected during the 1960s in Newcastle upon Tyne revealed that 17% had cortical LB, a similar prevalence to that found today. It appears then that DLB is not a new disorder, it is one which has only recently been recognised, despite being the second most common form of degenerative dementia in old age, only AD itself being more common.
Lewy bodies are spherical, intracytoplasmic, eosinophilic, neuronal inclusions which have a dense hyaline core and a halo of radiating filaments comprised of abnormally truncated and phosphorylated intermediate neurofilament proteins which also contain ubiquitin and associated enzymes. They were first described by the German neuropathologist Friederich Lewy when he was working in Alzheimer's laboratory in Munich between 1910 and 1912. Sub-cortical LBs are easily seen using conventional haematoxylin and eosin staining. Cortical LBs by contrast lack the characteristic core and halo appearance of their brain-stem counterparts and were, therefore, difficult to detect until the late 1980s when the development of antiubiquitin immunocyto-chemical staining methods allowed their true prevalence to be appreciated [4]. Even more recently [7,8] alpha-synuclein antibodies have been shown to label purified LBs and the a-synuclein antibodies PER1 and PER2 give strong staining of LBs and Lewy neurites (LN).
Current thinking about LB disorders suggests that there is spectrum of disease with the clinical presentation varying according to the site of LB formation and neuronal loss [9] (Table 1). Although pure presentations are seen in clinical practice, heterogenous combinations of parkinsonism, dementia and signs of autonomic failure are most frequent.
Primary Lewy body disorders
Recommendations have recently been made as to which brain regions should be examined for LB and a simple semi-quantitative scoring system devised in which a score of 1 is given if any are seen in a given area and a score of 2 if there are more than five per field [10]. These scores are then added to generate three possible pathological categories: brain-stem predominant, limbic (or transitional) and neocortical DLB. It has not yet been established to what extent these different patterns of pathological distribution correlate with different clinical profiles. Extensive neocortical pathology is not necessary for the development of dementia or other psychiatric symptoms, all of which may occur in the presence of limbic disease alone.
A heated debate has surrounded the interpretation of the Alzheimer-type changes which are also seen in most DLB patients. High senile plaque counts are found in the vast majority and these are morphologically indistinguishable from those found in pure AD [11]. They are, however, seldom tau immunoreactive, and indeed in 80–90% of DLB cases there is no evidence of significant tau pathology, paired helical filaments or neocortical neurofibrillary tangles [12]. Whether or not DLB is considered to be a variant of AD depends upon the pathological definition of AD being used [13]. Thus, 77% of LB pathology cases with dementia had ‘plaque only’ AD, a concept derived from definitions of AD which are heavily dependent upon plaque density. By contrast, 80–90% of DLB cases fail to fulfil definitions of AD requiring suprathreshold numbers of neocortical neurofibrillary tangles [14]. The new NIA/Reagan Foundation criteria for AD appear to be making a significant shift in this direction with a proposed requirement for frequent neurofibrillary tangles equivalent to Braak stages 5 and 6 [15]. Dementia with Lewy bodies and pure AD will, according to such criteria, be pathologically distinct in the majority of cases.
In summary, there appear to be at least three recognisable anchor points along a spectrum of neuro-degenerative disorders. Parkinson's disease (PD) is a disorder of predominantly subcortical LB neurofilament inclusions which are the most visible markers of an extensive neuritic degeneration involving a-synuclein and ubiquitin. More extensively distributed LB typify DLB, in addition to which there is usually significant beta amyloidosis and senile plaque formation falling short of AD. Alzheimer's disease, by contrast, has a combination of beta amyloidosis and neocortical neurofibrillary tangles, the latter representing dysregulation of microtubule assembly proteins, tau-related cytoskeletal abnormalities which are not found in most DLB cases.
Alzheimer's disease and DLB do share beta amyloidosis, senile plaque formation and severe depletion of acetylcholine which is even greater in DLB than AD. It is of interest in that context that they both also share an increase in apolipoprotein E ∊4 allele frequency [16], which is not seen in non-demented PD cases. Additional vascular changes are seen in up to 30% of AD and DLB cases.
A recent study of pathological burden versus clinical severity [17] examined correlations between two simple measures of cognitive ability and a range of lesion counts and neurochemical measures in midfrontal cortex of DLB cases. Dementia severity was significantly correlated with LB density, plaque density and severity of cholinergic deficit but not with neurofibrillary tangle density nor synaptophysin levels. In contrast, in AD cases it was tangle density and synaptophysin levels which were most highly correlated with clinical severity. This suggests that the dementias of DLB and AD may have different pathological but similar neurochemical substrates. Another important correlate of the extensive cholinergic deficit of DLB is with the presence of hallucinations which are a common clinical manifestation. Patients with hallucinations have significantly lower choline acetyl transferase levels than non-hallucinators [18].
Clinical features
Tables 2 and 3 summarise nine studies [19,27] which report on 190 autopsy confirmed DLB cases and compare them with 261 AD cases. Cases of DLB coming to autopsy are more likely to be men and, although mean age of onset and survival are similar, survival times in DLB are skewed with rapidly progressive illness in some individuals. Some, but not all of this reduced survival is probably attributable to neuroleptic sensitivity reactions [21].
Autopsy confirmed cases (1989–1997)
Clinical symptom rates in 190 dementia with Lewy bodies and 261 Alzheimer's disease cases (1989-1997)
Dementia is usually, but not always, the presenting feature of DLB; a minority present with psychosis in the absence of dementia, some with mood disorders or psychosis and others with orthostatic hypotension and falls. Fluctuation occurs in half to three-quarters of patients, but the range reported is very variable, probably because this is such a difficult symptom to define. Fluctuation, however, difficult to define, is not commonly seen in AD. Visual hallucinations are present in one-third to one-half of patients, although some series find them in up to 80%. Auditory hallucinations may only occur in 20% of DLB subjects but are seldom seen in AD. Depressive symptoms are common in both disorders but a 38% prevalence in DLB is significantly greater than in AD and similar to rates reported in PD. Rates for parkinsonism are highly variable, partly reflecting case ascertainment biases, some patients being collected through neurological departments which primarily receive referrals for movement disorders. Less than half of DLB cases have parkinsonism at presentation and a quarter continue to have no evidence at any point in their illness. Clinicians must, therefore, be prepared to diagnose DLB in the absence of parkinsonism; if they do not, their case detection rates will be unacceptably low. Finally, recurrent falls occur in up to a third of DLB cases, significantly more often than in AD, as does sensitivity to antipsychotic drugs, detected in 61% of all DLB cases who received antipsychotics, but in only 15% of those with AD. Two studies have examined inter-rater reliability and found agreement rates and kappa values to be acceptable for some symptoms of DLB such as delusions, hallucinations, parkinsonism and falls, but unacceptably low for other measures, particularly fluctuation [28,29].
The recent consensus criteria for the clinical diagnosis of DLB are shown in Table 4 [10]. Emphasis is placed upon the particular characteristics of the dementia syndrome, with attentional deficits and prominent fronto-subcortical and visuospatial dysfunction. Fluctuation is no longer essential for diagnosis, although it is frequently present. It seems probable that the fluctuating attentional deficit is linked to dysregulation of central cholinergic mechanisms controlling level of consciousness. The important hallucinatory symptoms are specified as being visual, recurrent and detailed, usually occurring most days of the week, typically colourful, three dimensional images of animals and children. Insight into the unreal nature of these hallucinations is usually absent while they occur but is gained after the event. Spontaneous parkinsonism not attributable to medication is a key symptom. If two of these three symptoms are present a diagnosis of probable DLB is made, if only one is present a diagnosis of possible DLB is allowed.
Consensus criteria for the clinical diagnosis of probable and possible dementia with Lewy bodies (DLB)
The sensitivity and specificity of the consensus clinical criteria against autopsy findings have been examined in several studies. Mega et al. [28] correctly identified 75% of DLB cases with a specificity of 0.79, by retrospectively applying the criteria to the charts of autopsy confirmed cases. Four raters of varying experience blindly applied an earlier version of the consensus criteria to the case notes of 50 cases of dementia with a mix of pathological diagnoses. The specificity of a diagnosis of DLB was uniformly high (0.90-0.97), but sensitivity of case detection was more variable (0.55-0.90), the least experienced rater being most liable to underdiagnose DLB [29]. This tendency to clinical underdiagnosis was again noted during the Second International Workshop on Dementia with Lewy Bodies which was held in July 1998 during the 6th International Conference on Alzheimer's Disease and Related Disorders in Amsterdam [30]
The results of a prospective validation study in Newcastle upon Tyne, funded by the UK Medical Research Council, were reported for the first time at the Workshop [31]. Within a sample of 50 hospital referred demented cases followed to autopsy, the sensitivity and specificity for a clinical diagnosis of probable DLB were 0.83 and 0.91, respectively. These high specificity figures were consistent with most other studies presented indicating that the DLB clinical criteria are appropriate for confirmation of diagnosis (few false positives) but may be of limited value in screening for DLB cases (low sensitivity). Clinical underdiagnosis of DLB remains a major problem, case detection rates being as low as 22% in a recent paper reporting a community based dementia study with autopsy confirmation of diagnosis [32].
Differential diagnosis of dementia with Lewy bodies
There are four main categories of disorder which should be considered in the differential diagnosis of DLB. These are as follows.
Other dementia syndromes
Sixty-five percent of autopsy confirmed DLB cases meet the NINCDS-ADRDA clinical criteria for probable or possible AD [33], and this is the most frequent clinical misdiagnosis of DLB patients presenting with a primary dementia syndrome. This suggests DLB should routinely be excluded when making the diagnosis of AD. Up to a third of DLB cases are additionally misclassified as vascular dementia by the Hachinski Ischaemic Index by virtue of items such as fluctuating nature and course of illness. Pyramidal and focal neurological signs are however, usually absent. The development of myoclonus in patients with a rapidly progressive form of DLB may lead the clinician to suspect Creutzfeldt-Jacob disease [10].
Other causes of delirium
In patients with intermittent delirium, appropriate examination and laboratory tests should be performed during the acute phase to maximise the chances of detecting infective, metabolic, inflammatory or other aetiological factors. Pharmacological causes are particularly common in elderly patients. Although the presence of any of these features makes a diagnosis of DLB less likely, comorbidity is not unusual in elderly patients and the diagnosis should not be excluded simply on this basis.
Other neurological syndromes
In patients with a prior diagnosis of PD, the onset of visual hallucinations and fluctuating cognitive impairment may be attributed to side effects of antiparkinsonian medications, and this must be tested by dose reduction or withdrawal. Other atypical parkinsonian syndromes associated with poor levodopa response, cognitive impairment and postural instability include progressive supranuclear palsy and multisystem atrophy. Syncopal episodes in DLB are often incorrectly attributed to transient ischaemic attacks, despite an absence of focal neurological signs. Recurrent disturbances in consciousness accompanied by complex visual hallucinations may suggest complex partial seizures (temporal lobe epilepsy) and vivid dreaming with violent movements during sleep may meet criteria for REM sleep behaviour disorder. Both of these conditions have been reported as uncommon presenting symptoms of autopsy confirmed DLB.
Other psychiatric disorders
If a patient spontaneously develops parkinsonian features or cognitive decline, or shows excessive sensitivity to neuroleptic medication, in the course of late onset delusional disorder, depressive psychosis or mania, DLB should be considered.
Investigations
As with any patient presenting with cognitive impairment, full history, mental state and physical examination are essential steps towards making a firm clinical diagnosis. As with suspected cases of AD, the level and extent of laboratory investigations will vary according to the clinical picture, associated comorbidity and physical examination. However, the particular associations of DLB with fluctuations in attention and cognition and visual hallucinations, both very commonly associated with a variety of other organic disorders, means that investigation of suspected cases of DLB will, of necessity, require very careful laboratory evaluation. This will usually include routine haematology and biochemistry, erythrocyte sedimentation rate (ESR) and/or C-reactive protein (CRP), thyroid function tests, B12 and folate, syphilis serology and urinalysis. A chest X-ray may also be considered routine in view of the high incidence of lung carcinomas in the elderly, especially in smokers. As with the diagnosis of AD, neuroimaging investigations will often be helpful both in terms of excluding other intracranial disorders (including cerebrovascular disease) that may be responsible for the cognitive impairment and in providing supportive features for the diagnosis. Increasingly, some form of structural imaging is becoming essential to rigorously apply diagnostic criteria such as the NINCDS/ADRDA criteria for AD, the NINCS/ADRDA criteria for vascular dementia and the consensus criteria for DLB.
Structural imaging changes in dementia with Lewy bodies
Few studies have investigated computerised tomography or magnetic resonance imaging changes in DLB. In a longitudinal study of AD subjects who came to postmortem, Förstl et al. [34] reported more pronounced frontal lobe atrophy on CT in eight subjects who had LB pathology at postmortem compared to pure AD cases. However, using MRI, Harvey et al. [35] found no differences in frontal lobe volumes between AD and DLB subjects, a finding replicated in a different and larger cohort by Barber et al. [36]. While further studies are awaited, frontal lobe atrophy does not seem to be a particular feature of DLB. Similarly, DLB does not seem to differ from AD in terms of degree of ventricular enlargement or presence of white matter changes on MRI [37].
The strong association between AD and atrophy of the medial temporal lobe, whether assessed by a linear measurement of medial temporal lobe width on CT [38] or visual or volumetric ratings of hippocampal atrophy on MRI [39,40] led to investigation of whether similar changes were associated with DLB. Jobst et al. [38] found medial temporal lobe atrophy of similar magnitude to AD in two of their four cases of DLB. However, using MRI both case reports and controlled studies have shown DLB to be associated with relative preservation of temporal lobe structures compared to AD [35,36,41,42]. Using visual ratings, Barber et al. [36] found 38% of DLB subjects compared with no AD subjects had a normal rating of temporal lobe atrophy suggesting that, at least in some cases, relative preservation of the hippocampus and medial temporal lobe may support a diagnosis of DLB. The reason for this variability in temporal lobe atrophy in DLB is unknown, though based on very limited autopsy examination of four cases Harvey et al. [35] suggested that temporal lobe atrophy on MRI may be a marker of concurrent AD pathology in DLB. However, although cross-sectional imaging may be helpful in some cases, it will clearly not be a diagnostic marker. It has yet to be determined whether accurate longitudinal assessment of regional volume change on MRI may improve accuracy of diagnosis, as may be the case for AD [43]
In summary, the limited evidence available suggests that structural imaging in DLB will reveal similar generalised atrophic changes to AD in most cases, although approximately 40% of DLB subjects will show preservation of medial temporal lobe structures.
Functional imaging changes in dementia with Lewy bodies
Single photon emission tomography (SPET) using blood flow markers such as Tc-HMPAO has been extensively investigated in dementia. In AD, the classic appearances are of posterior bilateral symmetric temporo-parietal hypoperfusion [44,38]; which contrast with the frontal hypoperfusion characteristically seen in frontal lobe dementia [45]. Vascular dementia is associated with a mottled uneven patchy appearance reflecting the variable anatomical localisation of vascular disease [46]. In PD, decreased blood flow in basal ganglia occurs and, when associated with dementia, biparietal changes similar to those seen in AD are reported [47,48]. Similarly, in the few SPET studies of DLB similar patterns of blood flow changes to AD have been found, though Donnemiller et al. [49] also found a subtle difference in perfusion patterns with a greater degree of occipital hypoperfusion in DLB compared to AD.
The more powerful, although still research based, use of SPET involves specific ligands for different neurochemical systems. Ligands have been developed for pre- and postsynaptic dopaminergic and cholinergic systems. Donnemiller et al. [49] found significant differences between DLB and AD in beta-CIT binding (a ligand for the dopamine transporter), a difference that would be predicted from known neurochemical differences between AD and DLB. Reduced D2 receptor density in basal ganglia using IBZM has also been reported in DLB [50]. Using a marker of the choline transporter, significant differences between AD subjects and controls as well as between PD subjects with and without dementia have been found [51]. In summary, current evidence suggests blood flow SPET will show similar appearances in DLB to those seen in AD though, as with AD, SPET may still be useful in distinguishing DLB from frontal lobe dementia or vascular dementia. However, new chemical imaging techniques, though not yet clinically available, show great promise in differentiating DLB from other disorders and are an exciting area of current research.
Treatment
Activity of the cholinergic enzyme choline acetyltransferase (ChAt) is lower in DLB than AD, particularly in temporal and parietal cortex [18] Clouding of consciousness, confusion and visual hallucinations are recognised effects of anticholinergic drug toxicity and the summative effects of subcortical and cortical cholinergic dysfunction probably play a major role in the spontaneous generation of similar fluctuating symptoms in DLB. Reductions in ChAt are correlated with severity of cognitive impairment [17] and hallucinations may be related to hypocholinergic and (relatively) hypermonoaminergic neocortical neurotransmitter function [18]. Levodopa responsiveness is less predictable in DLB than in PD [52].
There have been several reports that patients who respond well to cholinesterase inhibitor treatments are more likely to have DLB than AD at autopsy [53,54]. This is consistent with the neurochemical profile of DLB and the fact that post synaptic cortical muscarinic receptors are functionally intact. Case reports suggest that cholinesterase inhibitors can reduce psychotic symptoms in DLB [55] and placebo controlled studies are in progress.
The most important practice point in the management of a patient with DLB is caution in (or preferably avoidance of) the use of neuroleptic medications, which are the mainstay of antipsychotic treatment in other patient groups. Severe neuroleptic sensitivity reactions [20,21] can precipitate irreversible parkinsonism, further impair consciousness level, and induce autonomic disturbances reminiscent of neuroleptic malignant syndrome. They occur in 40–50% of neuroleptic treated DLB cases and are associated with a 2–3 fold increased mortality. Acute D2 receptor blockade is thought to mediate these effects, and despite some promising initial reports, atypical and novel antipsychotics such as risperidone and olanzapine seem to be as likely to cause neuroleptic sensitivity reactions as older drugs. Until safe and effective medications become available, there is no doubt that the mainstay of clinical management is to educate patients and carers about the nature of their symptoms and suggest strategies to cope with them. The clinician must ascertain which symptoms are most troublesome for the sufferer and explain the risks and benefits associated with changes in medication [56]. In these circumstances, where the clinician is walking a therapeutic tightrope between parkinsonism and psychosis, the best outcome is invariably a compromise between a relatively mobile but psychotic patient and a non-psychotic but immobile individual. The patient and his or her carers may only be able to decide which is the lesser of these evils after experiencing both states.