Research criteria for the diagnosis of prodromal dementia with Lewy bodies

How does prodromal DLB usually present?

One or more of the core clinical features characteristic of fully developed DLB may develop before dementia and are usually accompanied by mild cognitive complaints.⁵ Spontaneous parkinsonism often develops within the predementia stage but is not present in all patients.^3,6 RBD is a parasomnia that typically occurs years, and even decades, before the onset of dementia or parkinsonism and may presage any of the α-synucleinopathies.^4,7,e4 In patients with mild cognitive deficits who later develop dementia, both parkinsonism and RBD strongly predict a later transition to DLB rather than to AD or other dementia types.^6,8 Delirium can occur during the predementia stage of DLB⁹ as can fluctuations of cognition and arousal³ that may give rise to a diagnosis of delirium. Visual hallucinations (VHs) either spontaneous or provoked by illness or medication are more likely to occur compared with normal controls or prodromal AD.^10,11 There are also case reports of delusions, hallucinations, depression, and anxiety as presenting features of DLB.¹² Based on such observations, 3 prototypic prodromal DLB syndromes have been proposed¹³ as (1) mild cognitive impairment (MCI), (2) delirium-onset, or (3) psychiatric-onset.

MCI with Lewy bodies

The National Institute on Aging and Alzheimer's Association criteria for MCI^{14, e5} provide a standardized approach to diagnosis and form the basis for our proposed MCI with Lewy body (MCI-LB) criteria. They require a cognitive complaint from the patient or from an informant or clinician who knows them and has observed a decline. Also required are deficits in one or more cognitive domains that are greater than would be expected from normal aging, do not represent lifelong patterns of lower cognitive function, and are not associated with acute medical or neurologic insults. Although people with MCI may be less efficient or less capable at performing tasks they have performed previously, their cognitive deficits should not be sufficient to interfere with their typical daily functioning. In other words, there should be an overall preservation of their prior level of independence with minimal interference in day-to-day functional abilities, which, by definition, does not constitute a dementia.

Objective cognitive impairment is optimally based on standardized assessment with scores on neuropsychological measures that are typically 1 to 1·5 SDs below the mean for their age and education matched peers on culturally appropriate normative data (i.e., for the impaired domain(s), when available). It is emphasized that these ranges are guidelines and not cutoff scores.¹⁴ Impairment on neuropsychological tests may also be demonstrated by significant decline demonstrated on serial testing or significant decline from estimated premorbid abilities.^e5 Cognitive impairment may additionally be categorized as single or multiple domain, and as amnestic or nonamnestic, which may help to classify potential subgroups as they relate (1) to biomarkers and pathologic correlates, and (2) to differences in the rate of decline and progression to dementia, each of which may be important in the conduct of clinical trials. A further refinement to describe the cognitive profile of MCI-LB may be provided by using the MDS level II criteria for PD-MCI.^15,e6,e7

The cognitive presentation of MCI-LB

The cognitive pattern of MCI-LB is similar to DLB and typically includes disproportionate attention/executive and visual processing deficits and relatively preserved memory and object naming.^16,e8,e9 Impairment on tasks of attention, processing speed, and verbal fluency typically constitute the attention/executive deficits, and tasks of visual discrimination, assembly, and figure drawing typically constitute the visual perceptual and spatial deficits.^{5,6,17,–,19}

Diagnostic criteria for DLB emphasize that prominent or persistent memory impairment may not necessarily occur in the early stages,² but patients and carers frequently offer memory complaints as a presenting symptom.²⁰ Of 49 patients with MCI who progressed to DLB, 27% had multidomain amnestic MCI, but only 6% had single-domain amnestic MCI,³ similar to other reports.^19,21 When patients with MCI-LB have memory impairment, attention and/or visual processing deficits also tend to be present³ and may precede the memory difficulties. Patients with RBD who later developed DLB had attention/executive deficits up to 6 years before the diagnosis of dementia and memory deficits up to 2 years before the diagnosis.²² The basis for verbal memory impairment in DLB or MCI-LB may be related to slowed processing speed and deficits in working memory and retrieval that characterize the attention and executive deficits,^e10 consistent with hippocampal preservation in MCI-LB compared with MCI-AD.^23,e11 However, impaired memory in DLB may also be related to the extent of coexisting AD-related pathology,^e9 characterized by greater hippocampal atrophy on imaging ^e12 and greater CA1 hippocampal subfield pathology on autopsy.^.e13 Similar to other subtypes of MCI, a proportion of patients may revert to being cognitively normal, although they remain at a greater risk of the eventual development of dementia.^e14 Some instability of an MCI-LB diagnosis is to be expected given inherent fluctuating cognition, worsening with neuroleptics or anticholinergics, or improvement with levodopa-carbidopa or cholinesterase inhibitors.

In summary, the performance pattern of MCI-LB is best characterized as single- or multidomain nonamnestic MCI, or as multidomain amnestic MCI, whereas single-domain amnestic MCI is more likely to represent MCI-AD. The MCI-LB cognitive pattern is often seen in patients with 1 or more core DLB features,⁵ although these may develop later.^3,18 Nonamnestic MCI seldom develops into AD but is associated with a greater risk of transition to DLB^10,21 with a ten-fold risk compared with amnestic MCI.³ Because a substantial subset of patients with DLB have coexisting AD-related pathology that may influence their cognitive profile, MCI-LB still should be considered an important part of the differential diagnosis in amnestic subjects.

Operationalization of MCI-LB

The scheme suggested in table 1 for the identification of MCI-LB allows a diagnosis of either possible or probable MCI-LB based on the number of qualifying clinical features or biomarkers. The terms possible and probable refer to the likelihood of underlying LB disease and not to the MCI syndrome. Structured diagnostic instruments may assist identification of the core clinical features of DLB that precede, coincide with, or follow the onset of cognitive difficulties.^24,–,26 Cognitive fluctuations may be of lesser amplitude or frequency than in more severe disease. Passage and sense of presence hallucinatory phenomena may precede the development of recurrent, well-formed and detailed VH. Clinical features supportive of DLB may occur secondary to other causes reducing their diagnostic specificity, but they can be useful indicators of underlying LB disease, particularly when they persist over time or occur in combination.

Use of the possible MCI-LB category may raise diagnostic suspicion prompting further clinical and biomarker investigation. Some possible MCI-LB combinations will likely prove better predictors than others, for example, multidomain nonamnestic MCI plus clinically well-defined RBD is anticipated to be more predictive of progression to DLB than single-domain amnestic MCI plus a history of cognitive fluctuation. The utility of different combinations of clinical features and biomarkers remains to be established.

MCI-LB and PD-MCI

Uncertainty may occur in deciding how patients exhibiting both MCI and parkinsonism are best categorized. PD-MCI will usually be the most appropriate diagnosis when PD is diagnosed before significant cognitive decline occurs. The adoption of a 1-year rule similar to that used to separate DLB and PD dementia may be helpful to distinguish some MCI-LB and PD-MCI cases if the onset and order of mild symptoms of parkinsonism and cognitive decline can be clearly established. If not, an initial diagnosis of prodromal LB disease may be preferable, recognizing that this will require revision as the full clinical picture evolves. Individuals with poor global cognitive test performance identified in community samples have been reported to be at a greater risk of developing PD over subsequent years,^e15,e16 but poor cognitive test scores are not equivalent to MCI, and this increased risk was no longer statistically significant when individuals with subtle motor signs at baseline were excluded.^e15 There are data to suggest that patients with MCI-LB may have greater cognitive impairment than those with PD-MCI,²⁷ but this is not easily operationalized to differentiate individual subjects. The cognitive deficits of PD-MCI can be heterogeneous,^e17 with disproportionately affected attention/executive function,²² visuospatial skills,^e18 and memory.²⁸ Similar to MCI-LB, there is evidence that nonamnestic MCI is more common than amnestic MCI and that multidomain amnestic MCI is more common than single-domain amnestic MCI.²⁹

Biomarkers for prodromal DLB

Direct measures of α-synuclein pathology would offer definitive diagnosis at an early stage; several are under development, but none is yet validated or available for use antemortem. Surrogate biomarkers of LB disease must therefore be used, as is the case for the diagnosis of DLB itself.² We propose biomarkers for the diagnosis of prodromal DLB, where there is either sufficient, good quality, published evidence of adequate diagnostic specificity in prodromal DLB or this can be reasonably extrapolated from fully developed DLB or related disorders. We categorize biomarkers for which only limited data are available as potential biomarkers, recognizing that such distinctions are likely to require revision as new data become available.

Proposed biomarkers

Potential biomarkers

These are biomarkers consistent with underlying LB disease, which may help the diagnostic evaluation, but for which there is still insufficient evidence of diagnostic sensitivity and specificity in early disease. This will change as new evidence emerges, at which point these, or other candidates, may become considered as indicative of prodromal DLB.

Other biomarkers

There is currently no α-synuclein radioligand with sufficient evidence supporting its utility for imaging DLB or any other α-synucleinopathy, nor have any diagnostically applicable biofluid, peripheral tissue, or genotypic biomarkers been established.² CSF measures cannot reliably discriminate between DLB and AD,³⁹ although the measurement of CSF α-synuclein aggregates using seeding aggregation assays such as protein misfolding cyclic amplification and real-time quaking-induced conversion is providing encouraging preliminary results and may be extended to more accessible biofluids including saliva, plasma, serum, or urine.^e28 Combining CSF markers may be useful; for example, addition of CSF phospho-tau to α-synuclein significantly improved the differential diagnosis between MCI-LB and MCI-AD increasing sensitivity and specificity both > than 95%.^e29 Other direct or indirect biomarker candidates that may occur early include identification of phosphorylated α-synuclein deposits in peripheral nervous tissue from skin biopsy,⁴⁰ gait analysis,^e30 and abnormalities in color vision discrimination.^41,42

The figure shows a framework for considering a wider range of biomarkers in the setting of prodromal DLB determining the extent to which AD-related pathology is contributing to the dementia syndrome and influencing clinical trajectories in these patients.^{43,44,e31,e32} Aβ42 declines earlier in AD than DLB, whereas Aβ40 levels increase alone in AD. A low Aβ42/Aβ40 ratio may therefore differentiate patients with prodromal AD from patients with prodromal DLB better than individual measures.⁴³ There is also evidence showing different frequencies and topographies of abnormal uptake on amyloid and tau PET imaging in DLB vs AD.⁴⁵

• Download figure
• Open in new tab
• Download powerpoint

Figure Multimodal biomarkers for prodromal DLB

Reduced dopamine transporter uptake in basal ganglia demonstrated by SPECT or PET; PSG-confirmed REM sleep without atonia and MIBG myocardial scintigraphy are proposed as biomarkers of prodromal DLB. Multimodality biomarkers are also important in characterizing patients with prodromal DLB, in whom the pathologic mechanisms of cognitive impairment include both LB and AD pathologies and who may have abnormalities associated with both. Medial temporal lobe uptake on tau PET, medial temporal lobe atrophy on MRI, and high levels of uptake on amyloid PET and positive CSF biomarkers of AD-related pathology may characterize patients with prodromal DLB with significant AD pathology. It is expected that biomarker abnormalities detected in the earliest stages will remain abnormal as disease progresses, with additional biomarkers showing changes later during the progression from prodromal to dementia phase DLB. DLB = dementia with Lewy bodies; MIBG = meta-iodobenzylguanidine.

Delirium-onset DLB

Cognitive fluctuations, or pronounced variations in alertness and arousal, are a core feature of DLB, apparent as marked fluctuations in cognitive ability or function, detectable by observation or informant report, and measurable using sensitive cognitive and electrophysiologic measures.^e33 The occurrence of delirium (sometimes referred to as acute confusional state) as an early presenting feature of DLB has been described in numerous case reports and series,^e34 including in those with no apparent previous history of cognitive impairment.^46,e35 Whether this represents a greater vulnerability to delirium in subjects with DLB, or the misdiagnosis of severe fluctuations with clouding of consciousness as delirium, or a combination of both is unclear.

Recognition that DLB may first present as delirium is important because most guidelines for behavioral disturbances in delirium recommend antipsychotics as first-line pharmacologic treatment. Delirium or marked fluctuations in consciousness were reported by 43% of caregivers before DLB diagnosis,⁴⁷ and previous episodes of delirium were much more frequent in patients with DLB compared with patients with AD (25% vs 7%), with 1 in 4 of those with DLB having repeated delirium.¹¹ In the majority of these, potential provoking causes were present. A similarly higher incidence of delirium was found for subjects with DLB compared with subjects with AD in the year before diagnosis (17·6 v 3·2/100 person-years),⁹ and admissions for delirium were more common in those already diagnosed with DLB than for other dementias. Prolonged delirium may also raise the index of suspicion for DLB.^e36 In an FDG-PET imaging study of prolonged delirium, 32% of subjects were found to have a DLB like pattern, a high proportion in regard both to the known prevalence of the disease and to that of their matched cognitively impaired control group (4%).^e37

The similarities between the marked fluctuations in DLB and disturbances of attention and consciousness in delirium have been noted and reviewed,⁴⁶ but there is as yet little understanding of a shared neurobiological basis. There have been few pathologic studies, and although in an epidemiologic-based autopsy cohort, there was no link between delirium and LB pathology at autopsy,^e38 a significantly increased risk of postoperative delirium in those with peripheral α-synuclein pathology has been described.⁴⁸ Delirium has been suggested to be more common in those subjects with DLB with later age at onset,^e39 but further evidence is needed.

In summary (table 2), it is clear that DLB can present with delirium, and in patients diagnosed with delirium, a careful search for other DLB features should be made with a low threshold for undertaking DLB biomarker examinations, especially in those with recurrent, unexplained, or prolonged delirium. The extent to which delirium presentations of DLB have the biomarker abnormalities associated with established DLB or other prodromal DLB presentations is unclear, although the PET imaging study cited above supports this,^e37 as does a report of reduced DAT uptake in a patient with acute, unexplained delirium.⁴⁹ The link between delirium and DLB is an important area for future research, to clarify the relationship between the 2, and to establish which factors associated with delirium should raise the index of suspicion for underlying prodromal DLB.

Psychiatric-onset DLB

Early clinicopathologic studies suggested that DLB may present as a primary psychiatric disorder,^50,51 but subsequent focus on the cognitive and motor aspects of DLB has limited the documentation of such cases outside of a few centers.^{12,32,52,–,54,e36} Late-onset major depressive disorder and late-onset psychosis are the most frequently reported presentations differing markedly from the construct of mild behavioral impairment^e40 and sometimes sufficiently severe to require hospitalization. Symptoms include hallucinations in visual and in other modalities and systematized delusions including Capgras syndrome,⁵⁵ apathy, anxiety, and depression. Psychiatric-onset DLB cases are not easily differentiated from non-LB late-onset psychosis cases on the basis of primary psychiatric phenomenology or neuropsychological profile alone.⁵⁴ Psychomotor retardation such as slowed speech, thinking, and body movements can resemble the bradykinesia of parkinsonism. The occurrence of rest tremor or rigidity is more helpful than bradykinesia to suspect prodromal DLB in patients with depressive disorder,⁵³ but psychotropic-induced parkinsonism may complicate diagnosis. Atypical clinical features may prove valuable pointers to underlying LB pathology, particularly the presence of recurrent VH¹² when these occur before cognitive impairment becomes evident. As with all α-synucleinopathies, RBD may be a useful indicator, although a relationship between antidepressant usage and subsequent RBD onset is a potential confounder.^e41 Although the primary psychiatric manifestations are often accompanied by mild cognitive deficits,⁵⁴ cognitive evaluation and interpretation of performance can be difficult when psychiatric symptoms are prominent. The frequency of cognitive fluctuations in psychiatric-onset LB cases has not been determined.

Initial reports suggest that ¹²³I-MIBG scintigraphy may be helpful in psychiatric-onset DLB.^52,53,56 Eighteen of 35 patients with a first onset of major depressive disorder >50 years and with bradykinesia developed a clinical diagnosis of DLB after 6 years of follow-up. All 18 had an abnormal ventilatory response to hypercapnia (VRH), indicative of severe autonomic dysfunction, whereas none of the 17 patients with a normal VRH converted to DLB within the study period. For the converters, the most common presentation was with psychotic and melancholic features simultaneously. The frequency of hypersensitivity to antipsychotics, antidepressants, and antianxiety drugs was higher in converters than in nonconverters.⁵³ Further studies need to confirm these findings and to determine the value of other DLB biomarkers in psychiatric-onset cases.^32,52,56

In summary, it is not yet clear how to identify patients with prominent late-onset psychiatric symptoms who may have underlying LB disease and subsequently progress to DLB. It is premature to try to construct formal criteria for psychiatric-onset DLB, but clinicians in mental health and other settings need to be aware that this possibility exists, not least because of the risk of severe antipsychotic sensitivity reactions with increased morbidity and mortality. The small available literature including reports such as those cited above does provide some useful guidance, summarized in table 3, but requires replication and further investigation.

Idiopathic REM sleep behavior disorder

RBD is characterized by abnormal dream enactment behavior during REM sleep, accompanied by loss of muscle atonia during REM sleep (REM sleep without atonia or RSWA) on PSG. For the purpose of a prodromal DLB diagnosis, a clinical history of RBD is obtained in the same way as for DLB,² that is, a clear complaint of dream enactment by a bedpartner or other witness, with no evidence of sleep disorders that may mimic RBD on clinical interview. Conditions known to mimic or masquerade as RBD include non-REM parasomnias (sleep walking, sleep talking, or other behaviors that emerge from non-REM sleep stages), obstructive sleep apnea, confusional arousals, or nocturnal seizures. In the absence of PSG, the risk of a false-positive clinical diagnosis of RBD is reduced by using optimized sleep questionnaires, which can have sensitivity and specificity over 90%.^e42

RBD may be considered as idiopathic until it is associated with another ongoing neurologic condition, the most frequent being an α-synucleinopathy (PD, DLB, or MSA). In a recent large study, 73.5% of patients with RBD converted to an overt neurodegenerative syndrome after 12-year follow-up, 56.5% developing parkinsonism, and 43.5% dementia as their first disease manifestation. Among many predictive markers tested, only the cognitive variables differed at baseline between those converting to primary dementia vs parkinsonism.⁷ Other multicenter studies confirm that the only clinical feature that predicts dementia vs parkinsonism first is neuropsychological performance; those with RBD who exhibit cognitive changes are more likely to evolve first into MCI or DLB.^22,e4 Thus, although RBD may be regarded an early manifestation of α-synucleinopathy, it is not possible to clearly distinguish whether an individual presenting with RBD will develop dementia first, that is, evolve into a primary DLB diagnosis, or parkinsonism first, that is., a primary PD/MSA diagnosis. Any MCI subtype associated with PSG-confirmed RBD is highly likely to represent underlying LB disease. Although idiopathic RBD is a useful model to study the early stages of LB disease progression, it is not necessarily representative of the whole spectrum of patients with PD and DLB, a significant minority of whom do not have RBD.⁴²

Autonomic dysfunction/anosmia and other nonspecific prodromal LB disease symptoms

Patients with DLB frequently report a history of autonomic dysfunction at first clinic attendance, with 25%–50% complaining of one of more among constipation, orthostatic dizziness, urinary incontinence, erectile dysfunction, increased sweating, or increased saliva.^56,57 In a cohort of patients with pure autonomic failure, 34% converted to an overt α-synucleinopathy over 6 years of follow-up; this was to DLB (52%), PD (24%), or MSA (24%).⁵⁸ However, because there are many other causes for autonomic dysfunction in older people, these symptoms, either alone or in combination, have low positive predictive value.⁵⁹ The situation is similar for hyposmia, which also frequently occurs early in prodromal DLB and PD,⁴² but which can have many other causes in this age group.