Core clinical features associated with survival in patients with dementia with Lewy bodies

Abstract

Background and Objectives:

This was an analysis of clinical data from prospectively followed participants meeting criteria for probable dementia with Lewy bodies (DLB) in the Mayo Clinic Alzheimer’s Disease Research Center (ADRC) between 1998 and 2024. DLB is characterized by unique core features of visual hallucinations (VH), parkinsonism, REM sleep behavior disorder, and cognitive fluctuations with a variable disease course. DLB is associated with a poor prognosis but whether these unique DLB core clinical features influence survival is unknown. We aimed to determine whether core clinical features are associated with survival in patients with probable DLB.

Methods:

Patients followed in the Mayo Clinic ADRC between 1998–2024 underwent annual clinical assessments. Those who met clinical criteria for probable DLB were analyzed. Time-dependent Cox proportional hazard models utilizing age as the time scale determined associations between the individual and cumulative number of core clinical DLB features and survival. The prognostic significance of core features present at the time DLB criteria were met was assessed in separate models. Models were adjusted for sex and duration from the onset of cognitive symptoms to DLB diagnosis.

Results:

Of 488 probable DLB patients meeting inclusion criteria, 118 (24%) were female with a mean age of 71.9 ± 8.4 years at time of meeting probable DLB criteria. Shorter survival was associated with the development of VH (HR 3.25, 95% CI 2.46–4.29) and parkinsonism (HR 2.28, 95% CI 1.54–3.39) during the disease course, and VH at the time of DLB diagnosis (HR 1.60, 95% CI 1.18–2.16). Four total core features were also associated with shorter survival (4 core features vs. 2 core features, HR 3.58 95% CI 2.66–4.80, 4 core features vs. 3 core features, HR 2.46, 95% CI 1.86–3.25). In 191 patients (45 female (24%) with a mean age of 71.2 ± 8.6 years at probable DLB criteria) with autopsy-confirmed DLB, VH, parkinsonism and four total core features were associated with shorter survival. Sex was not associated with survival.

Discussion:

Visual hallucinations, parkinsonism, and the development of all four core features were associated with shorter survival in probable and in autopsy-confirmed DLB. These findings have important prognostic and management implications for patients with DLB and their caregivers.

Introduction

Dementia with Lewy bodies (DLB) represents the second most common cause of neurodegenerative dementia after Alzheimer’s disease (AD), accounting for up to approximately 25% of cases with dementia in prevalence studies.¹ The primary pathology of DLB is α-synuclein² with AD co-pathology identified in about 50% of patients with DLB.^3–5 A meta-analysis of 11 studies showed shorter survival from initial diagnosis in DLB compared to those with symptomatic AD.⁶ The presence of AD co-pathology confers a worse prognosis in patients with DLB as does the presence of neocortically distributed Lewy bodies compared with more restricted synuclein distribution.^{4, 7} Further, the disease course of DLB is heterogenous with rates of cognitive decline that depend on pathologic burden, and in particular, progress more rapidly in those with AD co-pathology.³

Knowledge of prognostically significant clinical features is vital for patients with DLB, and for their caregivers and clinicians. Shorter survival in DLB has been associated with greater baseline dementia severity, longer time from symptom onset to evaluation, antipsychotic use, orthostatic hypotension, smaller baseline hippocampal volumes, AD cerebrospinal fluid biomarkers, and the presence of APOE ϵ4 alleles.^8–14 The presence of complex visual hallucinations or cognitive fluctuations have been shown to increase the risk of transitioning from mild cognitive impairment to dementia over two years of follow up, but whether they influence survival is unknown.¹⁵

How core clinical features impact prognosis in DLB remains poorly understood and is complicated by the fact that the frequency of individual core clinical features varies between individuals and with types of underlying co-pathology.³ However, in patients with predominant and initial symptoms of cognitive impairment the core clinical features of DLB are largely unique to DLB thus it is conceivable they may influence the disease course and survival; either directly through falls from parkinsonism or pharmacologic therapies used to treat them such as antipsychotics for visual hallucinations. Further understanding the influence of core clinical features on survival in DLB may have important prognostic and therapeutic implications for patients with DLB and their caregivers, thus we evaluated associations between core clinical features and survival in a large clinical cohort of patients with probable DLB and in a subgroup with autopsy-confirmed DLB.

Methods

Cohort Selection and Clinical Procedures

Participants included in our analysis met criteria for clinically probable DLB by their last evaluation (N=488), had a reliable informant, and were prospectively followed as part of the Mayo Clinic Alzheimer’s Disease Research Center (ADRC) between January 1998 and June 2024. Of the 488 participants with probable DLB, 225 individuals (46%) consented to undergo autopsy as part of their participation in the ADRC research program. Clinically probable DLB was defined using the McKeith 2017 criteria: dementia with two or more of the following four core features: probable REM sleep behavior disorder (RBD), spontaneous parkinsonism, fully formed visual hallucinations (VH), and cognitive fluctuations.¹⁶ Participants who had parkinsonism for >1 year before the onset of cognitive symptoms were excluded as they would meet criteria for Parkinson’s disease dementia.¹⁷

As part of Mayo ADRC protocol, at each visit patients underwent a comprehensive neurologic history and exam, neuropsychologic evaluation, medication assessment, and clinician interview with the patient and their informant. Systematic recording of the presence/absence and onset year/month of each core feature as well as estimated cognitive symptom onset was obtained, reaffirmed, or updated at each visit by the clinician through patient and caregiver interview and neurologic examination. The presence of cognitive fluctuations was determined through clinical interview and defined as spontaneous alterations in alertness or arousal with or without accompanying staring or clear periods of disorganized thinking with return to baseline. VH were determined through clinical interview and considered present if the participant or informant reported recurrent fully formed hallucinations. A diagnosis of probable REM sleep behavior disorder was based on clinical interview of the informant regarding the presence of recurrent dream enactment behaviors with or without dream mentation. Parkinsonism was defined as the presence of rigidity, rest tremor, bradykinesia, or postural instability and was based on neurologic examination with presence/onset determined by the neurologist at each visit. Following each visit, a consensus meeting of neurologists and neuropsychologists render a clinical diagnosis based on established criteria.¹⁶ Exposure to antipsychotic medications, dopaminergic medications, cholinesterase inhibitors or anticholinergic medications was abstracted from the research database. Data from the participants in this study have been previously published.¹⁸ Current results incorporated longitudinal follow-up data based on information obtained from the most recent evaluation.

Neuropathologic examination and subgroup identification

Brains were sampled using a standardized protocol with macroscopic and microscopic immunohistopathologic evaluations as described by previously.^{3, 18} Classification of Lewy body disease (LBD) included brainstem LBD (BLBD), AD with amygdala-only LBD (AD-ALB), transitional brainstem-limbic LBD (Limbic LBD), or diffuse brainstem-limbic-neocortical (DLBD). The distribution of neurofibrillary tangles (NFT) was classified by Braak NFT stage and characterized by no neocortical tangle deposition in stages 0 to III, initial spread into temporal neocortex in stage IV, tangle deposition in association cortex in stage V, and involvement of primary sensory and motor cortex in stage VI.

Pathologic classification of Limbic LBD (N=55) and DLBD (N=136) comprised the autopsy-confirmed DLB subgroup (n=191). Those with BLBD (N=4) were excluded because BLBD does not typically present with DLB. Similarly, AD-ALB (N=10) were excluded because it represents an advanced form of AD. The remaining groups excluded from our analysis had non-LBD pathologies and have been previously described.¹⁸

Statistical Methods

Continuous variables were summarized with means and standard deviations and categorical variables were reported as frequencies with percentages. Group comparisons were made using Student’s t-tests, Welch’s t-tests, and Chi-square tests. Time-dependent Cox proportional hazard models with age as the time scale were used to determine associations between clinical features and survival. Survival was estimated from the time patients met clinical criteria for probable DLB to death.¹⁶ Time-dependent models were chosen to allow for the accumulation of core clinical features over the study period. Age was chosen as a time scale in order to remove age as a potential confounding variable. Given the disproportionate number of males with DLB, all models included sex as a covariate. Also, due to the established differences in the development of specific core features between DLB patients with and without AD co-pathology,³ the model included estimated time from cognitive symptom onset to when criteria for probable DLB criteria was met. Multivariable modeling examined the effects of probable RBD, VH, fluctuations, and parkinsonism that occurred at any point during the disease course on survival. The association between the total number of core DLB features (using 2 as the reference) on survival was evaluated in a separate model. Separate models evaluated the effect of both individual and total number of core features that were present at the time patients met criteria for DLB on survival. Right censoring was performed at the last visit for those lost to follow up or for those who died more than 2 years from the last clinical evaluation so as not to inflate the number of events due to passive accrual beyond the time expected for next clinical evaluation. Exploratory analyses were performed separately in male and female patients with DLB to determine whether clinical features impacted survival in a sex-specific manner. Lastly, time-dependent Cox proportional hazard models examined whether each core DLB feature, and the number of core features were associated with the distribution of α-synuclein (Limbic LBD vs. DLBD) and tau (Braak NFT stages 0 to III vs. stages IV to VI) pathology. Medication exposure was considered a binary yes/no variable, however, due to lack of medication initiation dates and inability to accurately assess the timing of medication initiation and relationship to core clinical feature development medication use was not included in survival models, nor were we able to assess how medications influenced core feature development (or resolution). Adjustment for multiple comparisons was not utilized to avoid type II error increase since this was a broad examination of the relationships between core clinical features of DLB and survival to serve as hypotheses generation for future studies.¹⁹

Standard protocol approval and patient consents

The study was approved by the Mayo Clinic Institutional Review Board and followed the Health Insurance Portability and Accountability Act (HIPAA) guidelines. Participants and their legally authorized representatives provided written informed consent.

Results

Clinical characterization

The demographic and clinical features are summarized in Table 1. The mean age of meeting probable DLB criteria was 71.9 ± 8.4 years probable DLB criteria. One hundred eighteen (24%) participants were female. Death occurred in 92% of the 488 patients in the cohort. Demographic data were similar between the clinically probable DLB cohort and autopsy-confirmed DLB subgroup, with males disproportionately represented. Unsurprisingly, the total number of core features increased as the disease progressed. VH were least likely to be present at the time DLB criteria were met, increasing from 35% to 73% by the last evaluation. Twenty-seven percent of the cohort was exposed to antipsychotic medications, the vast majority taking quetiapine. Antipsychotic agents were more likely in those with VH (117/355 (33%) VH vs 16/133 (12%) no VH, p < 0.001), with no differences in disease duration in antipsychotic users vs non-users (p = 0.29). Use of anticholinergic antidepressants was infrequent (9%) whereas cholinesterase inhibitor use was nearly ubiquitous (88%), each of which were unrelated to disease duration. By the last evaluation, parkinsonism was evident in 90% of 488 patients. Dopaminergic medications were used by 38% of the cohort, 90% of whom took carbidopa-levodopa, with no difference in dopaminergic medication exposure based on the presence of VH (39% VH vs. 35% no VH, p=0.44). DLB patients treated with dopaminergic medications had a longer duration of illness from cognitive symptom onset to death (9.6 ± 4.1 years) than those without dopaminergic treatment (8.8 ± 3.9 years, p = 0.023), and a longer interval from DLB diagnosis to death (7.8 ± 3.6 years vs. 6.1 ± 3.3 years, p < 0.001).

Table 1:

Demographics and Clinical Characteristics

Variables	Entire Cohort	Autopsy cohort
N	488	191
Deceased	450 (92%)	191 (100%)
Death age	79.1 ± 7.6	77.9 ± 8.2
Male	370 (76%)	146 (76%)
Non-European Ancestry	24 (5%)	2 (1%)
Years of Follow-up	4.2 ± 3.2	4.9 ± 3.2
Core Features at time DLB criteria was met
Probable REM Sleep Behavior Disorder (RBD)	337 (69%)	133 (70%)
Parkinsonism	318 (65%)	123 (64%)
Cognitive Fluctuations	264 (54%)	113 (59%)
Visual Hallucinations (VH)	171 (35%)	68 (36%)
Three Core DLB Features	86 (18%)	39 (20%)
Four Core DLB Features	14 (3%)	8 (4%)
Core Features during disease course
Probable REM Sleep Behavior Disorder (RBD)	373 (76%)	145 (76%)
Parkinsonism	437 (90%)	173 (91%)
Cognitive Fluctuations	374 (76%)	156 (82%)
Visual Hallucinations (VH)	355 (73%)	148 (77%)
Three Core DLB Features	145 (30%)	44 (23%)
Four Core DLB Features	209 (43%)	98 (51%)
Time from
Estimated cognitive symptom onset to death, years	9.0 ± 3.8*	9.2 ± 3.6
Cognitive symptom onset to meeting criteria for DLB	2.3 ± 2.9	2.5 ± 3.3
Medication exposure
Cholinesterase inhibitor	428 (88%)	169 (89%)
Dopaminergic	186 (38%)	75 (39%)
Carbidopa-Levodopa	167 (34%)	68 (36%)
Antipsychotic	132 (27%)	54 (28%)
Quetiapine	112 (23%)	47 (25%)
Anticholinergic antidepressants	45 (9%)	16 (8%)
Neuropathology
N		191
Limbic Lewy body disease (Limbic LBD)	--	55 (29%)
Diffuse Lewy body disease (DLBD)	--	136 (71%)
Braak NFT stages 0 to III	--	89 (47%)
Braak NFT stages IV to VI	--	102 (53%)
Moderate to Frequent neuritic plaques	--	118 (62%)

Values represent n (%), mean years ± sd, DLB=Dementia with Lewy bodies, DLB criteria refers to dementia plus two or more core DLB features. NFT = neurofibrillary tangles, Limbic LBD = transitional brainstem and limbic Lewy body disease, DLBD = diffuse brainstem, limbic, and neocortical Lewy body disease.

^*Deaths occurred in 450 patients.

Clinical and neuropathologic differences between sexes

Comparisons between males and females for demographic, clinical, and neuropathologic features are provided in Table 2. At the time criteria for DLB was met, females were more likely to have VH (26% males vs. 64% females, p<0.001) and males were more likely to have RBD (77% males vs. 43% females, p < 0.001). Males and females had similar rates of parkinsonism at the time criteria for DLB was met, but by the last evaluation, males were more likely to be diagnosed with parkinsonism (92% males vs. 83% females, p = 0.014), and treated with dopaminergic medications (43% males vs. 24% females, p < 0.001). Females had a longer interval from cognitive symptom onset to when criteria for probable DLB was met compared to males (p=0.002), but sex was not associated with shorter survival (Figure 1). In the autopsy sample, there was no sex difference in the distribution of Lewy-related pathology, but AD co-pathology was more common in females than males.

Table 2:

Clinical and neuropathologic characteristics distinguished by sex

Variables	Male	Female	p-value
N	370	118	--
Deceased	336 (91%)	114 (97%)	0.047
Death age	78.6 ± 7.4	80.5 ± 7.9	0.026
Non-European Ancestry	8 (2%)	16 (14%)	< 0.001
Education	14.9 ± 3.3	13.4 ± 2.8	< 0.001
Years of follow-up	4.2 ± 3.3	3.9 ± 2.8	0.36
Core features at the time DLB criteria was met
Probable REM Sleep Behavior Disorder (RBD)	286 (77%)	51 (43%)	<0.001
Parkinsonism	247 (67%)	71 (60%)	0.22
Cognitive Fluctuations	196 (53%)	68 (58%)	0.40
Visual Hallucinations (VH)	95 (26%)	76 (64%)	<0.001
Time from
Estimated cognitive symptom onset to death	9.0 ± 3.9	9.0 ± 3.7	0.98
Cognitive symptom onset to meeting criteria for DLB	2.0 ± 2.6	3.1 ± 3.6	0.002
Medication exposure
Cholinesterase inhibitors	331 (89%)	97 (82%)	0.052
Dopaminergic	158 (43%)	28 (24%)	< 0.001
Antipsychotic	95 (26%)	37 (31%)	0.23
Anticholinergic antidepressants	32 (9%)	13 (9%)	0.18
Neuropathology
N	146	45	---
Limbic Lewy body disease	45 (31%)	10 (22%)	0.35
Diffuse Lewy body disease	101 (69%)	35 (78%)	0.35
Braak NFT stages 0 to III	80 (55%)	9 (20%)	<0.001
Braak NFT stages IV to VI	66 (45%)	36 (80%)	< 0.001
Moderate to frequent neuritic plaques	82 (56%)	36 (80%)	0.005

Values represent n (%), mean years ± sd; DLB=Dementia with Lewy bodies, DLB criteria refers to dementia plus two or more core DLB features, NFT = neurofibrillary tangles, Limbic Lewy body disease = transitional brainstem and limbic Lewy body disease, Diffuse Lewy body disease = brainstem, limbic and neocortical Lewy body disease.

Figure 1: Core DLB features associated with survival in probable and autopsy-confirmed DLB.

DLB=dementia with Lewy bodies, HR=hazard ratio, CI=confidence interval.

Core clinical features and association with survival

In the clinical cohort, the core DLB features associated with lower rates of survival were VH (HR 3.25 95% CI 2.46–4.29), parkinsonism (HR 2.28 95% CI 1.54–3.38), and the number of core DLB features during the disease course (4 vs 2: HR 3.57 95% CI 2.66–4.8; 4 vs 3: HR 2.46 95% CI 1.86–3.25) (Figures 1A, 1B). Fluctuations (HR 1.40 95% CI 1.05–1.87) and a longer duration from the onset of cognitive symptoms to the time when DLB criteria was met (HR 1.06 95% CI 1.02–1.10) showed a statistical difference, but the hazard ratios render the risk of death to be clinically negligible. Sex and RBD were not associated with survival (Figure 1A, 1B). These results were also demonstrated in autopsy-confirmed DLB (Figures 1C, 1D) and in separate analyses of males and females (Table 3).

Table 3:

Sex-specific associations between core clinical features and survival in clinically probable DLB

	Male DLB Participants (N = 370)			Female DLB Participants (N=118)
Core Features Model	HR	95% CI	p-value	HR	95% CI	p-value
Time from cognitive onset to DLB	1.06	1.00–1.12	0.04	1.05	0.99–1.12	0.13
Probable REM Sleep Behavior Disorder (RBD)	1.12	0.77–1.63	0.56	1.32	0.75–2.33	0.34
Parkinsonism	2.13	1.34–3.38	0.001	2.82	1.29–6.20	0.01
Visual Hallucinations (VH)	3.57	2.62–4.86	<0.001	2.02	1.00–4.06	0.05
Fluctuations	1.40	0.99–1.98	0.055	1.35	0.74–2.46	0.32
Number of Core Features Model	HR	95% CI	p-value	HR	95% CI	p-value
Time from cognitive onset to DLB	1.07	1.02–1.13	0.01	1.07	1.00–1.14	0.04
3 vs 2 core DLB features	1.55	1.05–2.29	0.03	1.34	0.70–2.54	0.37
4 vs 2 core DLB features	3.97	2.78–5.66	<0.001	2.65	1.46–4.82	0.001
4 vs 3 core DLB features	2.56	1.88–3.49	<0.001	1.99	1.04–3.77	0.04

HR=Hazard Ratio; CI=confidence interval. Events were 232 for 370 men and 70 events for 118 women.

When evaluating core features present at the time DLB criteria were met, only VH present at the time criteria for DLB was met (HR 1.60 95% CI 1.18–2.16) was associated with shorter survival, and only in the clinically probable DLB cohort (Table 4). The remainder of individual core clinical features, and the number of core features present at DLB diagnosis were not associated with survival in either the clinical cohort or in the autopsy-confirmed DLB subgroup (Table 4).

Table 4:

Associations between core DLB features present at time DLB criteria were met and survival

	Clinically Probable DLB (N=488)			Autopsy-Confirmed DLB (N=191)
Core Features Model	HR	95% CI	p-value	HR	95% CI	p-value
Male sex	0.87	0.63–1.19	0.38	0.96	0.64–1.44	0.83
Time from cognitive onset to DLB	1.03	0.99–1.08	0.098	1.04	0.99–1.09	0.17
Probable RBD at time DLB criteria was met	1.25	0.91–1.73	0.17	0.97	0.68–1.37	0.89
Parkinsonism at time DLB criteria was met	1.13	0.84–1.51	0.42	1.08	0.75–1.56	0.67
VH at time DLB criteria was met	1.60	1.18–2.16	0.003	1.26	0.86–1.86	0.24
Fluctuations at time DLB criteria was met	1.20	0.91–1.60	0.19	0.97	0.68–1.37	0.86
Number of Core Features Model	HR	95% CI	p-value	HR	95% CI	p-value
Male sex	0.98	0.74–1.30	0.87	1.04	0.71–1.51	0.85
Time from cognitive onset to DLB	1.04	1.00–1.08	0.081	1.04	1.00–1.09	0.06
3 vs 2 core DLB features when DLB criteria was met	1.34	1.00–1.80	0.053	1.13	0.79–1.63	0.51
4 vs 2 core DLB features when DLB criteria was met	1.40	0.78–2.83	0.23	0.97	0.45–2.14	0.96
4 vs 3 core DLB features when DLB criteria was met	1.07	0.54–2.10	0.76	0.87	0.38–1.96	0.73

HR=Hazard Ratio; CI=confidence interval; DLB = Dementia with Lewy bodies, RBD=REM Sleep Behavior Disorder, VH = visual hallucinations Events were 302 for 488 clinically probable DLB.

Core features and the distribution of α-synuclein and tau pathology

In autopsy-confirmed DLB, 71% had DLBD, 29% had Limbic LBD, and 53% had a Braak NFT stage of IV to VI (38% with stage V and VI only) (Table 1). There were no associations between individual or cumulative number of core features and the distribution of synuclein (limbic LBD vs DLBD) or absence vs. presence of neocortical tau pathology.

Discussion

In a series of 488 patients with clinically probable DLB and a subgroup of 191 autopsy-confirmed DLB, the presence of VH and parkinsonism during the disease course, and VH at the time of DLB diagnosis, were independently associated with survival. The total number of core features were also associated with worse prognosis and those with 4 core DLB features had the shortest survival.

The development of VH during the disease course was associated with shorter survival in clinically diagnosed and autopsy-confirmed DLB (Figure 1). Survival was also shorter in the clinical cohort when VH were present at the time of DLB diagnosis. VH were just as likely to occur in Limbic Lewy body disease as in diffuse Lewy body disease, and was unrelated to the presence or distribution of tau pathology, consistent with prior studies.^{7, 20}As such, the shorter survival specifically associated with VH cannot be attributed to widespread or greater overall synuclein or tau pathologic burden acknowledging there may be more widespread neuronal dysfunction that was not assessed by these neuropathologic measures which could potentially explain the relationship between shorter survival and VH.^{4, 10} Given that VH can occur when α-synuclein does not extend beyond the limbic regions, and greater Lewy body density in limbic regions has been associated with the earlier emergence of VH, this highlights the importance of the limbic circuitry in the development of VH in DLB acknowledging complex networks and other brain regions such as the occipital cortex also play a role in the development of VH in patients with DLB.^{20, 21} VH have also been associated with greater cholinergic cell loss and denervation in the nucleus basalis of Meynert, which heavily projects to limbic regions but also to the occipital cortex, and occurs in the early stages of DLB.^{22, 23} VH are also associated with greater cognitive impairment, which has also been associated with shorter survival, thus VH may be a surrogate marker of more severe cognitive dysfunction.²⁴ It is plausible that VH is a manifestation of more severe cholinergic dysfunction and limbic pathology in DLB which may play a central role in survival.²⁵ This is supported by evidence of improved VH as well as reduced rates of cognitive decline and death in DLB patients treated with cholinesterase inhibitors.^26–29 For instance, treatment with cholinesterase inhibitors is associated a decline of −0.39 points per year on the Mini Mental Status Exam 95% CI (−0.96 – 0.18) compared with −2.50 points per year (95% CI −4.28 - −0.73) in non user 5 years after DLB diagnosis and a HR of 0.66 (95% CI 0.46–0.94) of death in the first year after DLB diagnosis. The majority of the patients in this cohort were prescribed cholinesterase inhibitors which limits the ability to assess the significance of cholinesterase inhibitors in the current study. Studies evaluating whether prevention of VH confers a survival benefit in DLB and mechanisms underlying this association are needed.

Prior studies have reported increased mortality risk with antipsychotic use in all types of dementia,^{14, 30} and a greater risk of neuroleptic sensitivity in DLB patients when exposed to dopamine D2 receptor blocking medications.¹⁶ About 27% of our clinical cohort was exposed to antipsychotics, the most common being quetiapine, with only 2% of the cohort being exposed to typical antipsychotics with predominant D2 receptor blockade. DLB patients treated with antipsychotic agents in our cohort had no difference in their disease duration than their antipsychotic-naïve counterparts. Although polypharmacy in management of VH has been associated with increased mortality in individuals with dementia,³¹ only 22/355 (6%) of those with VH were exposed to more than one antipsychotic in our cohort. These data suggest that antipsychotic use is unlikely to account for the shorter survival associated with VH in our cohort. While prior reports show increased mortality in antipsychotic users, our data suggests its possible that the VH could be the driving mediator of the shorter survival that has been attributed to antipsychotics as prior studies have not directly accounted for the presence of VH in analysis.^{14, 30} We are unable to further elaborate on this potential relationship based on our data but additional studies are needed to better undestand the association between VH, antipsychotic use and survival in DLB. Overall, our results highlight the unique prognostic importance of screening for VH in patients with suspected DLB, and the need to examine the potential mediating factors.

Parkinsonism was also associated with shorter survival in both the clinically probable DLB cohort and the autopsy-confirmed DLB subgroup. Parkinsonism can lead to falls and greater dependence and complications arising from these could explain the worse prognosis. In Parkinson’s disease, more severe baseline motor dysfunction has been associated with increased risk of mortality, but this has not been evaluated in DLB.³² Only 38% of our cohort were treated with dopaminergic medications, suggesting the severity of parkinsonism was mild although we did not have data on parkinsonism severity, such as the UPDRS/MDS-UPDRS part III motor data available for this analysis. Alternatively, given 90% of the cohort had parkinsonism during the disease course but only 38% were treated with dopaminergic medications it is possible that undertreating parkinsonism itself could have contributed to shorter survival. It is possible that parkinsonism was under-treated out of concern that it might worsen neuropsychiatric symptoms. However, dopaminergic medication use did not differ between those with and without VH, making this an unlikely explanation. DLB patients who took dopaminergic medication had a longer duration of disease compared to those without treatment. This may represent lead time bias because those with parkinsonism deemed sufficient to warrant treatment may have begun dopaminergic intervention earlier than those with milder parkinsonism, which may have influenced this finding. An alternative explanation could be that clinician perception of the safety of initiating dopaminergic medications in particular patients may reflect individuals with a less aggressive disease. Despite the lack of differences in dopaminergic medication use between those with and without VH in our cohort, limiting dopaminergic medication use in patients with VH represents a potential practical limitation in clinical practice which could influence survival. Ultimately, whether treatment of parkinsonism with dopaminergic agents in DLB improves prognosis with and without VH is worth pursuing in future studies.

Shorter survival was associated with a greater number of cumulative core DLB features, with four core features conferring the worst prognosis. The total number of core clinical features implies a greater symptom burden and may represent a surrogate measure of functional impairment. On the other hand, more core features may reflect widespread neuronal dysfunction/degeneration and a more advanced stage of DLB. We did not find an association between the conventional neuropathologic assessments of the distribution of α-synuclein pathology (i.e. neocortical LBD vs. limbic LBD) or tau burden with the total number of core clinical features. This could argue against more widespread neocortical pathology driving the association between the number of core clinical features, which do largely arise from brainstem and limbic dysfunction. However, it is also possible that there is more widespread neuronal dysfunction that is not currently detectable by present pathologic assessments. As novel neuropathologic technologies advance, additional neuropathologic studies will be needed to better understand the association between the total number of core features, neuropathologic features and survival in DLB.

Cognitive fluctuations were only weakly associated with survival in the larger clinical cohort and were unrelated to survival in the autopsy-confirmed DLB subgroup. The reasons for this are unclear but may be attributable to the challenges inherent in assessing fluctuations and their onset. Probable RBD was not associated with survival which is not surprising because RBD is a prodromal feature that can occur well before the onset of cognitive symptoms or the other core features in DLB.^{18, 33}

Sex differences in DLB remain under studied.³⁴ There was no sex difference in the distribution of α-synuclein, but females were more likely to have AD co-pathology and had a longer duration between cognitive symptom onset and when criteria was met for probable DLB.¹⁸ Our results showed no sex effect on rates of survival nor unique sex-specific survival associations with individual or total core features (Figure 1, Table 3). While VH were more common in females and parkinsonism was more common in males, both males and females had shorter survival in the context of VH, parkinsonism, and 4 core DLB features (Table 3). This is in contrast with Parkinson’s disease which demonstrates unique sex-effects, with male PD patients having increased mortality risk which may be in part due to a lower risk of dementia in women with PD and overall supports unique disease processes in PD and DLB despite shared neuropathologic findings.^{35, 36} Despite the lack of sex differences in our cohort, this data provides important prognostic information for providers caring for individuals with DLB.

Our study has several limitations. Our cohort is derived from a tertiary care referral center, potentially limiting generalizability to community-based patients with DLB. On the other hand, our study identified prognostically significant clinical features that do not rely on significant resources to ascertain; thus our results may be applicable to lower resourced settings. Advances in medical therapies over the 26 years of participant enrollment and follow-up may have influenced patient survival and confounded our results. The onset of each core clinical feature relied on informant report and may involve recall bias. Due to the limitations of time-dependent modeling, we could not incorporate dementia severity. However, a unique strength of time-dependent modeling is that it accommodates the accumulation of features over time and removes age as a potential confound with inclusion of age as the timescale. Due to the lack of medication initiation dates, we were unable to examine the effect of medication use on survival in our models. Furthermore, we did not have measures of parkinsonism severity so we could not assess whether this may have influenced survival and should be evaluated in future studies. Cerebrospinal fluid biomarkers for α-synuclein were not available to confirm underlying pathology in our clinical cohort. Finally, the diagnosis of RBD was made clinically and not confirmed with polysomnography and no measures of objective autonomic function or systematic assessment of autonomic features were performed to determine whether different aspects of autonomic dysfunction impact survival.

Conclusion

In patients with clinically probable and autopsy-proven DLB, the presence of VH, particularly early VH, parkinsonism, and cumulatively more core clinical features were associated with shorter survival. These data highlight future areas of study that may improve outcomes in probable DLB, such as the investigation of agents for VH through modulation of cholinergic neurotransmission and early treatment of parkinsonism. Additional studies are needed to elucidate mechanisms underlying these associations, ideally with the inclusion of neuroimaging or neurophysiologic biomarker data to further refine these findings.

DATA AVAILABILITY

Anonymized data will be shared by request from qualified investigators in accordance with the Mayo Alzheimer Disease Research Center data-sharing protocol.