Clinical Manifestations of Early-Onset Dementia With Lewy Bodies Compared With Late-Onset Dementia With Lewy Bodies and Early-Onset Alzheimer Disease

Key Points

Question  Do patients who develop dementia with Lewy bodies (DLB) younger than 65 years have unique clinical features?

Findings  In this case-control study of 542 individuals, early-onset DLB was often misdiagnosed; certain motor features and, to a lesser extent, neuropsychiatric features were associated with a diagnosis of early-onset DLB over early-onset Alzheimer disease dementia. Late-onset DLB had more prominent amnestic features but lower rates of depression than early-onset DLB.

Meaning  In evaluation of suspected early-onset DLB, one should assess for motor signs, apathy, depression, and determine if motor deterioration predated cognitive and behavioral changes; the amnestic features seen in late-onset DLB may be associated with Alzheimer disease copathology.

Importance  Early-onset dementia, presenting in individuals younger than 65 years, is a diagnosis with significant social and financial implications. The early-onset form of dementia with Lewy bodies (DLB) is poorly understood.

Objective  To investigate clinical features that distinguish early-onset DLB (onset and diagnosis at age <65 years) from late-onset DLB (onset at age ≥65 years) and from early-onset Alzheimer disease (AD) dementia.

Design, Setting, and Participants  This is a retrospective case-control study on patients with pathologically confirmed DLB or AD enrolled in the National Alzheimer’s Coordinating Center database from January 2005 to July 2017. The National Alzheimer’s Coordinating Center Uniform Data Set comprised deidentified data collected by Alzheimer disease centers in the United States. Of patients fulfilling criteria for all-cause dementia at enrollment (n = 1152), those who at post mortem received a pathological diagnosis of either AD (n = 848) or Lewy body disease (n = 218) were selected. Excluding 52 patients owing to missing data and 12 diagnosed with Parkinson disease dementia, remaining patients were classified by age of symptom onset into early-onset AD, early-onset DLB, and late-onset DLB subgroups. Data were analyzed from June to December 2018 and from November to December 2021.

Exposures  Demographics, cognitive, behavioral, and motor features recorded at first clinic visit and neuropathological characteristics at autopsy were analyzed by disease subgroup.

Main Outcomes and Measures  Concordance between initial etiologic diagnosis of dementia and final pathological diagnosis was assessed, as was time to death.

Results  A total of 542 individuals were categorized as having early-onset AD (n = 363; mean [SD] age, 53.0 [5.8] years; 208 [57.3%] male), early-onset DLB (n = 32; mean [SD] age, 57.9 [3.2] years; 23 [71.9%] male), and late-onset DLB (n = 147; mean [SD] age, 73.5 [5.5] years; 103 [70.1%] male). Early-onset DLB was clinically misdiagnosed in 16 individuals (50%). Features that predicted a diagnosis of early-onset DLB over early-onset AD included visual hallucinations (15 [46.9%] vs 42 [11.6%]), slowness (23 [71.9%] vs 95 [26.2%]), apathy (23 [71.9%] vs 189 [52.1%]), and motor deterioration that preceded cognitive and behavioral symptoms (7 [21.9%] vs 6 [1.7%]). Late-onset DLB had more amnestic features, but this was accounted for by a higher proportion of neocortical neuritic plaques and diffuse plaques (frequent in 79 [53.7%] vs 8 [25%]) than seen in early-onset DLB.

Conclusions and Relevance  This study found that early-onset DLB has clinical features that distinguish it from early-onset AD, whereas features of late-onset DLB are associated with a higher burden of AD copathology.

Lewy body disease is the second most common neurodegenerative cause of dementia after Alzheimer disease (AD).¹ Lewy body disease pathology is associated with 2 clinical diagnoses: dementia with Lewy bodies (DLB) and Parkinson disease dementia. Core clinical features of DLB include parkinsonism, fluctuation of consciousness, visual hallucination, and rapid eye movement sleep behavior disorder, while supportive features include neuroleptic sensitivity, postural instability, falls, and autonomic dysfunction.² Interestingly, there is likely overlap in etiology of AD dementia and DLB. Overall, 66% to 80% of postmortem brain samples from patients had some evidence of AD copathology; conversely, 40% of those diagnosed with AD dementia had a degree of Lewy body pathology.^2,3

Dementia with onset in individuals younger than 65 years (early-onset dementia) poses a particular diagnostic challenge for the clinician. It is a rarer condition with many possible etiologies, each with different implications for prognosis.⁴ Dementia that develops in individuals younger than 65 years leads to greater economic burden, caregiver stress, and mortality rates.^5,6 Early-onset and late-onset dementia have distinct clinical and neuroimaging findings.⁷ For instance, one-third of early-onset AD dementia may present with nonamnestic symptoms including apraxia and visuospatial dysfunction, but this is the case for only 6% of late-onset AD dementia.⁸

The mean age at onset of DLB is 75 years,⁹ and incidence increases with age.^10,11 As with AD, Lewy body pathology can also cause dementia in individuals younger than 65 years. Case reports illustrate features of rapidly progressive dementia, myoclonus, cortical visual disturbance and neuropsychiatric changes, attention deficits, myoclonus, depression, visual construction issues, and apraxia.^12-14 There are no longitudinal studies on early-onset DLB, to our knowledge. It is unclear if the core clinical consensus criteria of DLB¹⁵ are equally applicable to early-onset DLB or whether it is an entity distinct to late-onset DLB and prodromal DLB.¹⁶ Based on comparative studies of DLB and AD dementia,¹⁷ one might wonder how to distinguish early-onset DLB and early-onset AD dementia at presentation and how their prognosis compares.

This is a retrospective case-control study of patients with pathologically confirmed Lewy body disease and AD from the National Alzheimer’s Coordinating Center (NACC) database. Among individuals with dementia onset at younger than 65 years, we looked at how closely initial etiologic diagnoses agreed with these final pathological diagnoses. We identify factors that predicted a diagnosis of early-onset DLB over early-onset AD dementia based on information gathered at first clinic visit. We then compare the clinical features of early-onset DLB with late-onset DLB, with reference to differences identified on neuropathological data.

This was a retrospective case-control study based on the NACC Uniform Data Set, which comprises deidentified data provided with participants and informants’ written informed consent, collected by Alzheimer disease centers. The database is funded by the US National Institute on Aging (NIA). Data collection is in accordance with NIA policies. Database research is approved by the University of Washington institutional review board and data access complied with relevant data use agreements.¹⁶ An NACC data set used for the current study was contributed by 17 centers in the United States and included 1152 patients evaluated from January 2005 to July 2017. This study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Data on self-reported race and ethnicity were collected by the NACC database, with categories of American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or other Pacific Islander, White, or other. Patients who fulfilled clinical criteria for all-cause dementia on their first visit were included initially (Figure). Patients with more than 50% incomplete data were excluded. These were divided into 2 pathologically defined subgroups: (1) Lewy body disease: patients with Braak stage 4 and below, who either had Lewy body pathology at autopsy or were given a pathological diagnosis of DLB pathology. Patients whose only initial clinical diagnosis was Parkinson disease dementia were excluded. Patients were retrospectively classified as having early-onset DLB if onset of relevant symptoms and clinical diagnosis of dementia both occurred at younger than 65 years or late-onset DLB if symptom onset occurred at 65 years or older. (2) AD: patients who met NIA/Reagan Institute neuropathological criteria for AD pathology (either having a high NIA–Alzheimer’s Association AD neuropathologic change score or receiving a pathological diagnosis of AD, primary). Patients were classified as having early-onset AD dementia if symptom onset and clinical diagnosis of dementia both occurred at age younger than 65 years.

Based on these pathologically defined subgroups, data gathered at first clinic visit and at post mortem were then analyzed retrospectively. These included demographic, cognitive, and motor and behavioral variables, which were collected on different forms under the first visit packet in the NACC database (eTable 1 in the Supplement). Formal test batteries in the NACC database used included the Mini-Mental State Examination (MMSE), Unified Parkinson’s Disease Rating Scale, Geriatric Depression Scale (GDS), Neuropsychiatric Inventory Questionnaire (NPI-Q), Logical Memory IA and IIA, Trails A and B, and Boston Naming test. Clinical variables were measured and recorded by the clinician, while the NPI-Q was administered through informants.

The etiologic diagnoses of dementia made by clinicians at first visit were also analyzed. The NACC recommended these diagnoses to be guided by consensus criteria for probable DLB,¹⁵ and NIA–Alzheimer’s Association criteria for probable AD dementia.¹⁸ A diagnosis was taken into account whether it was deemed a primary or contributing diagnosis for dementia.

The κ statistic was used to assess concordance between initial clinical and final pathological diagnoses of AD dementia and DLB. This is adjusted for agreement by chance and interpreted as such¹⁹: slight agreement, 0.21 to 0.40; fair agreement, 0.41 to 0.60; and moderate agreement, 0.61 to 0.80.

Mean (SD) were reported for continuous variables, while frequency and proportion were reported for categorical data. Two pairwise comparisons were performed for every variable: between early-onset AD and early-onset DLB and between early-onset DLB and late-onset DLB. Each pair was compared using a t test for continuous variables and Fisher exact tests for categorical variables. Univariate comparisons were adjusted for either MMSE score (early-onset DLB vs AD) or mood medication usage (early-onset DLB vs late-onset DLB). Missing values were not imputed; complete-case analysis was used.

Logistical regression was then performed to select factors that best distinguished early-onset DLB from early-onset AD. After variables with more than 40% missing values were excluded, all predictors with P value of .10 or less at univariate analysis were used to build a multivariable model. Reduced model selection was performed using a backward stepdown by applying the Akaike information criterion. All data analysis was performed in R version 3.5.3 (R Foundation), with 2-sided significance level set at .05. Data were analyzed from June to December 2018 and from November to December 2021.

After inclusion and exclusion criteria were applied, 363 patients with early-onset AD (mean [SD] age, 53.0 [5.8] years; 208 [57.3%] male), 32 patients with early-onset DLB (mean [SD] age, 57.9 [3.2] years; 23 [71.9%] male), and 147 patients with late-onset DLB remained (mean [SD] age, 73.5 [5.5] years; 103 [70.1%] male) (Figure). A tabulation of missing values is provided (eTable 1 in the Supplement).

Among patients with early-onset dementia, the initial clinical diagnosis concurred with the pathologic diagnosis in 79.9% (290 of 363) for pathologically confirmed AD but only 50% (16 of 32) of patients with pathologically confirmed DLB (eTable 2 in the Supplement). Based on κ statistics, there was only slight agreement in AD diagnoses and moderate agreement in DLB diagnoses. Interestingly, 12 of 32 patients (37.5%) with pathologically confirmed DLB and early-onset dementia received a clinical diagnosis of AD at their first clinic visit. This suggests that we can improve our ability to distinguish early-onset AD dementia and DLB clinically at presentation.

Patients with early-onset DLB in this cohort were older than those with early-onset AD at symptom onset and at first clinic visit but had shorter time to death and higher MMSE scores than patients with AD (Table 1).

Even after correcting for baseline MMSE and age differences between these 2 groups, early-onset DLB was more likely to present with hallucinations, delusions, apathy, rapid eye movement sleep behavior disorder, and motor symptoms including altered gait, tremors, slowing, and increased falls than those with AD (Table 2). They were more likely than patients with AD to report deterioration starting in the motor domain, rather than the cognitive or behavioral domains. Patients with early-onset DLB also had higher GDS scores, although the frequency of reporting low mood was similar in both groups. Although memory impairment was judged by clinicians to be present in more than 85% of patients with early-onset DLB at first visit, those with early-onset AD patients more likely to report cognitive changes as the first sign of deterioration and to do worse on Logical Memory tests but not Trails or Boston Naming (Table 2). A comparison of other neuropsychological test battery results available in the NACC data set is also provided (eTable 3 in the Supplement).

A multivariate regression was then performed (Table 3) using all variables that distinguished early-onset DLB and AD at a significance level of P < .10 on univariate analysis, excluding variables with more than 40% missing values. This analysis identified 5 factors, combinations of which most specifically predicted a diagnosis of early-onset DLB over early-onset AD. These factors included slowness, visual hallucinations, apathy (on NPI-Q), absence of agitation (on NPI-Q), and changes being seen first in the motor domain (predating cognitive and behavioral changes).

Overall, 20 of 32 patients (63%) with early-onset DLB possessed at least 3 of these 5 factors at first clinic visit, compared with 57 of 363 patients (16%) with early-onset AD (eFigure in the Supplement). The accurately diagnosed patients with DLB included 6 patients with early-onset DLB whose initial clinical diagnosis did not concur with their pathological diagnosis. Among 16 patients with early-onset DLB who were initially misdiagnosed, 8 showed motor slowing, 11 were not agitated, and 11 were apathetic (data not shown).

Patients with early-onset and late-onset DLB were largely similar, except for a few notable features. Although baseline MMSE scores were similar (Table 1), late-onset DLB was characterized by more significant memory impairment and worse performance on neuropsychological test batteries than early-onset DLB (Table 2, with more complete neuropsychological testing data in eTable 3 in the Supplement).

Patients with early-onset DLB also more frequently took mood-control medications than patients with late-onset DLB (Table 1). Consistent with this, patients with early-onset DLB were more frequently reporting depressed mood and also higher GDS scores than patients with late-onset DLB (Table 2). This association persisted even after correcting for prior use of mood medications.

Postmortem data from early and late-onset DLB subgroups were compared. There was no differential distribution of Lewy bodies across limbic and neocortical regions (eTable 4 in the Supplement). However, there was a higher density of neocortical neuritic plaques and diffuse plaques in late-onset DLB brain samples, changes that are more classically seen in AD (Table 4). After limiting analysis to patients with DLB in Braak stages 0 to 2, effectively excluding those with significant AD copathology, reported memory impairment and scores on selected neuropsychological tests became no different between patients with early- and late-onset DLB, yet differences in GDS scores remained (eTable 5 in the Supplement). This strongly suggested that the amnestic features seen in the late-onset DLB group were driven by AD copathology.

Early-onset DLB was often misdiagnosed as AD dementia in this study, but we identify clear distinguishing characteristics that clinicians could focus on when confronted with an early-onset dementia of undifferentiated etiology. Patients with early-onset DLB showed more psychotic features, cognitive fluctuations, motor changes, and apathy than patients with early-onset AD. There was a predominance of motor features, more so than neuropsychiatric features. This has important implications, as it suggests a thorough motor examination is critical when assessing early-onset dementia. The limited range of neuropsychological tests used in this study could not consistently distinguish early-onset DLB from early-onset AD dementia. However, patients with late-onset DLB had more amnestic deficits than those with early onset, which we attribute to a higher burden of AD copathology.

The clinical features of early-onset DLB here are consistent both with those set forth in the 2017 DLB consortium criteria¹⁵ and with those of prodromal DLB described elsewhere.²⁰ Visual hallucinations and parkinsonism are core features in consortium criteria, while apathy is a supportive feature. We highlight here that a focus on motor slowing, hallucinations, and the temporal sequence of motor changes relative to cognitive or behavioral changes improves our ability to discriminate early-onset DLB from early-onset AD. However, it was surprising that early-onset DLB was associated with lower agitation, whereas higher agitation levels are more commonly reported in DLB.²¹

We also propose the importance of assessing for depression through different modalities as part of the workup of early-onset dementia. While subgroups were indistinguishable on clinician evaluation of low mood, patients with early-onset DLB had higher GDS scores. Early-onset DLB was also linked to more frequent prescriptions of mood medication compared with late-onset DLB; this is aligned with studies showing that depression predicts an earlier age of onset in DLB but not in other dementias,²² perhaps highlighting a role for serotonergic pathways.²³

The findings of our comparison between early- and late-onset DLB are also consistent with current literature. Concomitant AD pathology is seen in up to 85% of DLB cases.³ Individuals with mixed AD-DLB pathology are usually older and show more marked cognitive decline.^24,25 Braak stage 3 or higher copathology is associated with specific amnestic and nonamnestic features in DLB, compared with individuals with DLB without AD copathology.²⁶ Conversely early-onset DLB shows more divergence from the AD dementia phenotype, given the lower burden of AD copathology. In our data, comparing patients with early- and late-onset DLB in lower Braak stages (eTable 5 in the Supplement), we show that AD copathology may account for the differences in amnestic features more so than the difference in GDS scores (nonamnestic features). We are unable to determine from this data set if AD copathology was present at disease onset or accumulated over time and whether vascular copathology contributed to the late-onset DLB phenotype. These should be explored in future studies.

This study would have benefited from a larger and newer data set with greater ethnic diversity. The small size of the early-onset DLB subgroup likely subjected it to greater bias from missing data than the other 2 subgroups. This work also lacked a statistical correction for multiple comparisons. Although cognitive fluctuations and rapid eye movement sleep behavior disorder are integral to DLB consensus criteria, they had to be excluded from our multivariate analysis owing to missing data (eTable 1 in the Supplement), which likely limited the generalizability and accuracy of our predictions. Future work would involve applying the results of our multivariate analysis to a validation data set, without which the model remains largely descriptive. It would also be meaningful to do a longitudinal analysis of clinical features and diagnoses to determine if these changed over serial evaluations.

The evaluation of certain variables could also be improved. Rapid eye movement sleep disorder was assessed with a single question posed to clinicians and patients. Motor function evaluated by the Unified Parkinson’s Disease Rating Scale alone was not objectively measured but taken by report alone; we might instead consider using the Unified Parkinson’s Disease Rating Scale motor subscale for a more quantitative analysis. The use of NACC data has limitations for the study of DLB, as data collection was primarily focused on the AD spectrum, particularly in earlier database versions, and in the selection of neuropsychological tests available. Future study in DLB should include tests more specific to executive function, visuoperceptual abilities, and construction.²⁷

In this study, early-onset DLB was best distinguished from early-onset AD dementia by features largely found in DLB consortium criteria. Compared with late-onset DLB, early-onset DLB has a lower burden of AD copathology and saw an association with features of early depression. Further research into the neuropathological basis of these differences is warranted. Nevertheless, memory impairment was still the most common symptom in DLB regardless of age at onset. Thus, when approaching a new patient with early-onset dementia, careful history-taking to determine if motor changes preceded cognitive and behavioral impairment, asking caregivers about apathy and agitation levels, and examining for motor slowing might improve the accuracy of a DLB diagnosis. Finally, patients with early-onset DLB experienced a shorter time from dementia diagnosis to death than patients with early-onset AD dementia. This may reflect delays in presentation, delays in diagnosis, higher symptom burden, or worse prognosis in patients with early-onset DLB. We hope this study challenges clinicians to better understand the characteristics and needs of this disease subgroup.

Accepted for Publication: March 12, 2022.

Published Online: May 23, 2022. doi:10.1001/jamaneurol.2022.1133

Corresponding Author: Simon Kang Seng Ting, MD, Singapore General Hospital, Outram Road, Singapore 169608, Singapore (simon.ting.k.s@singhealth.com.sg).

Author Contributions: Drs Ting and Sim had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Hameed, Ting.

Acquisition, analysis, or interpretation of data: Sim, Li, Ting.

Drafting of the manuscript: Sim, Ting.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Sim, Li, Ting.

Administrative, technical, or material support: Sim, Ting.

Supervision: Ting.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by Singhealth Foundation (grant NRS 15/001) and NNI Centre (grant NCG CS02). The National Alzheimer's Coordinating Center database is funded by the National Institute on Aging/National Institutes of Health (grant U24 AG072122). National Alzheimer's Coordinating Center data are contributed by the National Institute on Aging–funded Alzheimer disease research centers (grants P50 AG005131, P50 AG005133, P50 AG005134, P50 AG005136, P50 AG005138, P50 AG005142, P50 AG005146, P50 AG005681, P30 AG008017, P30 AG008051, P50 AG008702, P30 AG010124, P30 AG010129, P30 AG010133, P30 AG010161, P30 AG012300, P30 AG013846, P30 AG013854, P50 AG016573, P50 AG016574, P30 AG019610, P50 AG023501, P50 AG025688, P30 AG028383, P50 AG033514, P30 AG035982, P50 AG047266, P50 AG047270, P50 AG047366, P30 AG049638, P30 AG053760, P30 AG066546, P20 AG068024, P20 AG068053, P20 AG068077, P20 AG068082, P30 AG072958, and P30 AG072959.

Role of the Funder/Sponsor: Singhealth Foundation and the NNI Centre had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.