Association of probable REM sleep behavior disorder with pathology and years of contact sports play in chronic traumatic encephalopathy

Abstract

Probable rapid eye movement (REM) sleep behavior disorder (pRBD) is a synucleinopathy-associated parasomnia in which loss of REM sleep muscle atonia results in motor behavior during REM sleep, including dream enactment. Traumatic brain injury is independently associated with increased risk of pRBD and Lewy body disease, and both pRBD and Lewy body disease are often observed in chronic traumatic encephalopathy (CTE). However, the frequency and pathological substrate of pRBD in CTE have not been formally studied and remain unknown. Of the total sample of 247 men, age at death of 63.1 ± 18.8 years (mean ± SD), 80 [32%] were determined by informant report to have symptoms of pRBD. These participants had played more years of contact sports (18.3 ± 11.4) than those without pRBD (15.1 ± 6.5; P = 0.02) and had an increased frequency of Lewy body disease (26/80 [33%] vs 28/167 [17%], P = 0.005). Of the 80 participants with pRBD, 54 [68%] did not have Lewy body disease; these participants were more likely to have neurofibrillary tangles and pretangles in the dorsal and median raphe (41 of 49 [84%] non-LBD participants with pRBD symptoms vs 90 of 136 [66%] non-LBD participants without pRBD symptoms, P = 0.02), brainstem nuclei with sleep regulatory function. Binary logistic regression modeling in the total study sample showed that pRBD in CTE was associated with dorsal and median raphe nuclei neurofibrillary tangles (OR = 3.96, 95% CI [1.43, 10.96], P = 0.008), Lewy body pathology (OR = 2.36, 95% CI [1.18, 4.72], P = 0.02), and years of contact sports participation (OR = 1.04, 95% CI [1.00, 1.08], P = 0.04). Overall, pRBD in CTE is associated with increased years of contact sports participation and may be attributable to Lewy body and brainstem tau pathologies.

Introduction

Repetitive head impacts (RHI) and mild traumatic brain injury (TBI) sustained through contact sport participation or exposure to military blast have been associated with the neurodegenerative tauopathy chronic traumatic encephalopathy (CTE) [6, 7, 24, 44, 46, 66]. CTE can have a heterogeneous clinical presentation, including mood and behavioral disturbances, impaired cognition, and, more variably, parkinsonism [49, 68]. In addition to CTE, RHI may increase risk of developing other neurodegenerative diseases [22, 38], and pathologies co-morbid with CTE are frequently found at autopsy [3, 44, 47]. In fact, a link between RHI and pathologies besides tau is increasingly being recognized [1, 34, 43, 65, 67], and these co-morbid pathologies may be partially responsible for the clinical heterogeneity observed following RHI [1, 3, 39, 65, 67]. Further investigation is thus necessary to improve our understanding of the neuropathologic consequences of head injury and their clinical manifestations.

Individuals who have suffered TBI or been exposed to RHI frequently report sleep dysfunction, and TBI with loss of consciousness is a risk factor for rapid eye movement (REM) sleep behavior disorder (RBD) [59]. First described decades ago [60, 61], RBD is a parasomnia afflicting an estimated 1% of the population that most prominently features loss of the muscle atonia typically present during REM sleep, resulting in motor production, often as attempted dream enactment [25, 60, 61]. REM sleep physiology is incompletely understood, but case reports of human brainstem lesions support explanatory animal models crediting its origin and regulation to the brainstem [16, 33, 40, 45, 69]. In the foremost model, mutually inhibitory GABAergic neurons in REM-off and REM-on areas gate REM sleep transitions [41]. Glutamatergic neuronal populations produce muscle atonia via projections from REM-on areas to glycinergic and GABAergic spinal interneurons, which inhibit motor neurons [41]. Further modulation of REM sleep comes via inputs from monoaminergic, cholinergic, and orexinergic loci [16, 41, 63]. Among these, the monoaminergic dorsal and median raphe nuclei are closely tethered by GABAergic regulation to the sleep-wake cycle [50, 51, 53, 63], which likely explains why antidepressants and benzodiazepines modulate RBD symptoms [20]. Indeed, disruption of brainstem circuitry is the proposed etiology of RBD [16, 45], and degeneration in these regions may underlie sleep pathology.

REM sleep behavior disorder is often associated with synucleinopathies [10, 13, 21, 30], and its onset usually presages the development of other synucleinopathy-associable symptoms, such as parkinsonism and dementia [15, 17, 55, 57, 58, 64]. TBI has also been associated with increased risk of Parkinson’s disease (PD) and Lewy body pathology [1, 19, 22, 23, 31], and our group found that years of contact sports participation increased risk of neocortical Lewy body disease (LBD) with LBD, in turn, associated with increased odds for having dementia and parkinsonism [1]. Enhanced Lewy pathology in contact sport athletes may thus likewise be partially responsible for the RBD symptoms that can be observed in CTE [47]. We hypothesized that RBD would be frequent in CTE in association with years of exposure to RHI (using years of contact sports participation as a proxy), and that RBD in CTE would be associated with LBD and with tau pathology within the raphe nuclei of the brainstem.

Materials and Methods

Participants

Autopsy participants with a history of RHI exposure through contact sports participation and neuropathologically diagnosed with CTE (n = 363) were drawn from the Veterans Affairs-Boston University-Concussion Legacy Foundation (VA-BU-CLF) brain bank of the Understanding Neurologic Injury and Traumatic Encephalopathy (UNITE) Study, inclusion criteria into which include a history of contact sports participation, military service, or domestic violence [48]. For the present study, participants lacking informant-based data from our adapted version of the Mayo Sleep Questionnaire (MSQ), used to screen for RBD [11, 12], were excluded (n = 102), as were those with a co-morbid neuropathological diagnosis of progressive supranuclear palsy (n = 5), motor neuron disease (n = 8), or corticobasal degeneration (n = 1) to minimize error introduced by clinical misdiagnosis and informant recall [2, 35, 55], yielding a total sample of 247 male contact sport participants with CTE. Participants either clinically diagnosed with RBD during life (n = 6) and/or noted during next-of-kin MSQ administration to act out their dreams while sleeping (n = 78) were grouped as probable RBD (pRBD). Most brain donations were from peri- or postmortem next-of-kin-contact; others were referred by medical examiners, recruited by the CLF, or agreed during life to donation via the Brain Donation Registry. Written consent for brain donation and research participation was provided by donor next-of-kin, and institutional review board approval was granted from the Boston University Medical Center (BUMC) and the Edith Nourse Rogers Memorial Veterans Hospital, Bedford, MA. The BUMC IRB also granted approval for postmortem clinical record review, neuropathological evaluation, and clinical interviews with donor family members.

Informant clinical assessment

Informant-based clinical assessments were performed as described [3, 48]. RHI exposure history, including exposure type (e.g., contact sport, military service, domestic violence) and duration (e.g., years of contact sport participation); history of cognitive, mood, and behavioral changes; and antemortem clinical diagnoses were assessed via postmortem informant interviews, online surveys, and medical record review. Interviews were conducted blinded to the results of neuropathological examination by clinicians trained to assess for RHI exposure and neurodegenerative diseases. pRBD was determined by abbreviating the MSQ to specifically question the informant about RBD symptoms in the decedent. Principally, the question “Did [decedent] ever appear to act out his/her dreams while sleeping (punched or flailed arms in the air, shouted, or screamed)?” classified pRBD status, and informants were also asked whether the decedent had been diagnosed with RBD and/or obstructive sleep apnea (OSA). Informants were uniformly questioned and prompted to state if these symptoms or diagnoses were ever present or made, respectively, during the decedent’s life. The presence of antemortem dementia—based on Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition, Text Revision) criteria—and parkinsonism were determined by a panel of neurologists, neuropsychologists, psychiatrists, and/or neurosurgeons after reviewing medical and study records.

Neuropathological assessment

Neuropathological processing included comprehensive screening for neurodegenerative conditions adhering to procedures established for the VA-BU-CLF brain bank [48, 72]. CTE was neuropathologically diagnosed using the National Institutes of Health consensus criteria, which include abnormal perivascular accumulations of hyperphosphorylated tau (p-tau) in neurons, astrocytes, and cell processes concentrated at the sulcal depths [42], and staged based on regional p-tau involvement [44]. Neuronal p-tau was required for diagnosis and to distinguish from aging-related tau astrogliopathy [36]. In addition to regional evaluation to diagnose CTE [42], we assessed the dorsal and median raphe nuclei of the pons and midbrain for the presence or absence of intraneuronal p-tau in the form of neurofibrillary tangles (NFTs) or pretangles using AT8 immunohistochemistry. For standardization, the dorsal and median raphe nuclei for all cases was defined in brainstem sections taken at the level of the locus coeruleus and bordered by the medial longitudinal fasciculus and pontine white matter including pontocerebellar tracts ventrally (Fig. 1a). The National Institute of Aging-Reagan criteria were used to diagnose Alzheimer disease (AD), which included intermediate or high probability [28, 52]. Diffuse and neuritic plaques were evaluated using 4G8 and Bielschowsky silver staining, respectively.

Fig. 1.

Tau pathology in probable REM sleep behavior disorder in CTE. Extensive tau pathology is present in the median raphe nucleus in a participant with CTE and probable REM sleep behavior disorder. a,b Luxol fast blue hematoxylin and eosin stain of the dorsal and median raphe nucleus shown at 20x magnification bordered by the medial longitudinal fasciculus (mlf) and pontine white matter including pontocerebellar tracts (a, dorsal raphe: small white circle, median raphe: large white circle) and at 100x magnification (b). c α-synuclein immunostaining demonstrates an absence of Lewy pathology within the raphe nucleus. d Abundant tau pathology with neurofibrillary tangles (arrows) and tau-positive neurites and dots (arrowheads) as shown with AT8 immunostaining. Scale bar a, 500μm; b-d, l00μm

Lewy body disease was diagnosed using a modified version of the proposed unified staging system for Lewy body disorders (USSLB) that harmonizes our semi-quantitative severity scoring [1, 4, 52]. Briefly, using α-synuclein immunohistochemistry we quantified the number of Lewy bodies per neuroanatomical region of the olfactory bulb and tract, medulla, midbrain, amygdala, entorhinal cortex, anterior cingulate gyrus, and dorsolateral frontal cortex. LBD was classified as Olfactory bulb-only, Brainstem-Predominant, Limbic-Predominant, Brainstem and Limbic, or Neocortical [4].

Immunohistochemistry

Immunohistochemistry was performed as detailed [67]. Human tissue fixed in periodatelysine-paraformaldehyde was blocked, paraffin-embedded, and sectioned at 1μm. Antigen retrieval for α-synuclein and β-amyloid was performed with formic acid treatment for two minutes. Sections were incubated with primary antibodies overnight at 4°C. Antibodies used were α-synuclein (rabbit polyclonal; Chemicon, Temecula, CA; 1:15,000), phosphorylated PHF-tau (AT8; Pierce Endogen, Rockford, IL; 1:2000), and Aβ (4G8; BioLegend, San Diego, CA; 1:100,000). Sections were treated with biotinylated secondary antibody after three PBS washes and labeled with a 3-amino-9-ethylcarbazol HRP substrate kit (Vector Laboratories, Burlingame, CA), counterstained with Gill’s hematoxylin (Vector Laboratories H-3401), and coverslipped with Permount mounting medium (Thermo Scientific, Rockford, IL).

APOE genotyping

Apolipoprotein E (APOE) genotyping was performed as described [67]. DNA was extracted from brain tissue using a Qiagen QIAamp DNA extraction kit (Qiagen, Valencia, CA) and analyzed for two single nucleotide polymorphisms (National Center for Biotechnology Information SNPs rs429358 and rs7412) using TaqMan assays (Applied Biosystems, Foster City, CA).

Statistical analysis

Statistical analyses were performed with SPSS v25.0 (IBM Corp, Armonk, NY) and Prism v5 (GraphPad Software, La Jolla, CA). Initially, means were compared between groups with and without pRBD on clinical, exposure, and neuropathological characteristics using two-sided Student’s t-test and Mann-Whitney U-test for continuous and ordinal variables, respectively, and chi-square test for proportions to evaluate dichotomous variables. Subsets of the CTE population with and without LBD were isolated to identify non-LBD pathologies significantly associated with pRBD. For a final model, we employed a summary binary logistic regression with pRBD as the dependent variable to evaluate the robustness of these predictors, adjusting for age at death, in the total sample of CTE participants. Eight participants lacked data for raphe nuclei NFTs (five with and three without pRBD), and six (two with and four without pRBD) lacked data for years of play; these were excluded from the analysis, but a tipping point analysis was performed to comprehensively evaluate all possible outcomes from missing raphe nuclei NFT data [18]. Statistical significance throughout was set as α of 0.05.

Results

Probable RBD is frequent in CTE in association with years of contact sports participation

Of the total sample of 247 male CTE participants, 80 [32%] had pRBD symptoms during life. Comparison of clinical and neuropathological characteristics between participants with (n = 80) and without (n = 167) pRBD demonstrated similar frequencies of dementia, parkinsonism, sleep apnea, and APOE ε4 presence in each group, as well as equivalent mean ages of onset of cognitive decline and death (Table 1). pRBD participants experienced the onset of reported sleep symptoms at a significantly younger age (50.3 ± 19.2 years, mean ± SD) than the onset of reported cognitive symptoms (56.7 ± 16.0; 95% CI [−12.11, −0.62], P = 0.03).

Table 1.

Clinical characteristics and history of contact sport play in CTE participants with and without probable RBD

Characteristic	No pRBD	pRBD	P value
Sample size (n)	167a	80b	-
Age, mean (SD), y
Reported cognitive symptom onset	52.5 (19.2)	56.7 (16.0)	0.10
Death	61.7 (20.0)	66.0 (16.1)	0.10
Duration of cognitive impairment, mean (SD), y	9.9 (7.3)	9.8 (9.7)	0.92
Dementia, n (%)	104 (63.0)	58 (72.5)	0.14
Parkinsonism, n (%)	52 (31.1)	33 (41.3)	0.12
Sleep apnea, n (%)	38 (23.2)	26 (32.9)	0.11
APOE ε4 presence, n (%)	47 (36.2)	28 (35.0)	0.65
Contact sport exposure
Age of first exposure, mean (SD), y	11.5 (3.9)	11.3 (3.0)	0.74
Total years of play, mean (SD), y	15.1 (6.5)	18.3 (11.4)	0.02
Main sport played, n (%)
Football	148 (88.6)	70 (87.5)	0.96
Hockey	6 (3.6)	4 (5.0)
Boxing	6 (3.6)	3 (3.75)
Otherc	7 (4.2)	3 (3.75)

CTE chronic traumatic encephalopathy; pRBD probable REM sleep behavior disorder

^aOf 167 participants without pRBD, sample sizes were reduced to 166 individuals for dementia, 165 for sleep apnea, and 130 for APOE ε4 because of missing data.

^bOf 80 participants with pRBD, the sample size was reduced to 79 individuals for sleep apnea because of missing data

^cOther includes kickboxing, mixed martial arts, rugby, soccer, and competitive skiing

Years of RHI exposure has been associated with CTE and LBD, and these pathologies in turn predict dementia and, for LBD, parkinsonism [1]. Here we found that CTE participants with pRBD had played contact sports for more years (18.3 ± 11.4) than participants without pRBD (15.1 ± 6.5; 95% CI [0.42, 5.95], P = 0.02), but the age of first exposure to contact sports and the primary sport type were equivalent between groups (Table 1).

Probable RBD in CTE is associated with Lewy body disease and brainstem tau

REM sleep behavior disorder is highly predictive of α-synuclein pathology [14, 55], and RHI exposure is associated with LBD in CTE [1]. CTE participants with pRBD were more likely to have LBD (26/80 [33%]) than those without pRBD (28/167 [17%], P = 0.005), but USSLB LBD stage and severity were equivalent to the CTE participant group without pRBD (Table 2). No differences in CTE stage, AD severity, or vascular pathology were observed (Table 2).

Table 2.

Neuropathological characteristics of CTE participants with and without probable RBD

Characteristic	No pRBD	pRBD	P value
Sample size (n)	167a	80b	-
LBD present, +/− (%)	28/139 (16.8)	26/54 (32.5)	0.005
USSLB LBD stage, n (%)
I. Olfactory bulb-only	4 (14.3)	4 (15.4)	0.23
IIa. Brainstem Predominant	13 (46.4)	14 (53.9)
IIb. Limbic Predominant	4 (14.3)	1 (3.8)
III. Brainstem and Limbic	7 (25.0)	4 (15.4)
IV. Neocortical	0 (0)	3 (11.5)
LBD severityc, mean (SD), 0-3
Olfactory bulb	1.4 (0.9)	1.4 (1.0)	0.92
Dorsal Motor Nucleus	1.3 (1.2)	1.2 (1.0)	0.65
Substantia nigra	1.0 (1.0)	1.4 (1.1)	0.23
Amygdala	0.9 (1.1)	0.8 (1.0)	0.72
Entorhinal Cortex	0.8 (1.0)	0.8 (1.0)	0.90
Anterior Cingulate Cortex	0.5 (0.9)	0.7 (1.0)	0.36
Dorsolateral Frontal Cortex	0.1 (0.4)	0.5 (0.9)	0.11
Lewy neurites	1.0 (1.1)	1.0 (1.0)	0.98
Substantia nigra
Neuron loss, mean (SD), 0-3	1.1 (0.9)	1.3 (1.0)	0.10
Tau NFTs, yes/no (% yes)	119/45 (72.6)	64/16 (80.0)	0.21
CTE stage, n (%)
Stage I	36 (21.7)	10 (12.5)	0.37
Stage II	31 (18.7)	15 (18.8)
Stage III	56 (33.7)	31 (38.8)
Stage IV	43 (25.9)	24 (30.0)
Braak stage (AD), n (%)
Stage 0	39 (24.2)	11 (14.7)	0.22
Stage I-II	35 (21.7)	16 (21.3)
Stage III-IV	57 (35.5)	36 (48.0)
Stage V-VI	30 (18.6)	12 (16.0)
Thal phase, n (%)
Phase 0	71 (42.5)	32 (40.0)	0.94
Phase 1-2	21 (12.6)	12 (15.0)
Phase 3	21 (12.6)	11 (13.8)
Phase 4-5	53 (31.8)	25 (31.3)
Neuritic plaques, n (%)
Absent	108 (65.1)	45 (56.3)	0.39
Sparse	35 (21.1)	25 (31.3)
Moderate	14(8.4)	6(7.5)
Frequent	9 (5.4)	4 (5.0)
AD pathology (NIA), n (%)
Not	90 (54.2)	40 (50.0)	0.43
Low	22(13.3)	10(12.5)
Intermediate	33 (19.8)	23 (28.7)
High	21 (12.7)	7 (8.8)
Diffuse plaques, n (%)
Absent	72 (43.1)	32 (40.0)	0.89
Sparse	29 (17.3)	16 (20.0)
Moderate	27 (16.2)	15 (18.7)
Frequent	39 (23.4)	17 (21.3)
Vascular pathology
Microinfarcts, mean (SD)	0.02 (0.2)	0.03 (0.2)	0.76
Arteriolosclerosis grade, n (%)
0/1	74 (44.9)	35 (44.3)	0.94
2/3	91 (55.1)	44 (55.7)
Atherosclerosis grade, n (%)
0/1	125 (80.1)	59 (77.6)	0.66
2/3	31 (19.9)	17 (22.4)
Cerebral amyloid angiopathy grade, n (%)
0/1	132 (80.0)	62 (77.5)	0.65
2/3	33 (20.0)	18 (22.5)

pRBD probable REM sleep behavior disorder; LBD Lewy body disease; USSLB unified staging system for Lewy body disorders; NFT neurofibrillary tangle; CTE chronic traumatic encephalopathy; AD Alzheimer disease

^aOf the 167 participants without pRBD, sample sizes were reduced to 164 for substantia nigra NFTs, 166 for CTE stage, 161 for Braak stage, 166 for neuritic plaques and Alzheimer’s disease pathology, 165 for arteriolosclerosis, 156 for atherosclerosis, and 165 for cerebral amyloid angiopathy because of missing data.

^bOf the 80 participants with pRBD, sample sizes were reduced to 75 for Braak stage, 79 for arteriolosclerosis, and 76 for atherosclerosis because of missing data.

^cAnalysis of regional Lewy body severity only included CTE participants neuropathologically diagnosed with LBD

Despite the increased frequency of LBD in the pRBD group, 68% of pRBD participants did not have Lewy pathology, including 50% of those diagnosed with RBD during life. To identify non-LBD neuropathology associated with pRBD, we excluded participants with LBD to conduct a subgroup analysis of non-LBD participants with and without pRBD. Clinically, no significant differences were observed between groups in ages of cognitive decline or death, dementia, parkinsonism, OSA, or APOE ε4 status (data not shown). pRBD participants had a similar age of first exposure to contact sports as participants without pRBD but had played contact sports for significantly more years (18.0 ± 9.9 vs 15.0 ± 6.1, 95% CI [0.59, 5.29], P = 0.02; Table 3).

Table 3.

Associations with probable RBD in CTE participants without Lewy pathology

Characteristic	No pRBD	pRBD	P value
Sample size (n)	139a	54b	-
Age, mean (SD), y
Cognitive symptom onset	50.1 (19.8)	52.7 (16.3)	0.41
Death	58.9 (20.5)	62.0 (16.9)	0.33
Contact sport exposure, mean (SD), y
Age of First Exposure	11.0 (3.9)	11.2 (3.0)	0.75
Total Years of Play	15.0 (6.1)	18.0 (9.9)	0.02
Presence of NFTs, yes/no (% yes)
Dorsal/median raphe nuclei	90/46 (66.2)	41/8 (83.7)	0.02
Locus coeruleus	119/14 (89.5)	47/4 (92.2)	0.58
Subcoeruleus	75/57 (56.8)	29/21 (58.0)	0.89
Basis pontis	42/96 (30.4)	15/35 (30.0)	0.95
Substantia nigra	95/41 (69.9)	41/13 (75.9)	0.40

LBD Lewy body disease; pRBD probable REM sleep behavior disorder; NFT neurofibrillary tangle; CTE chronic traumatic encephalopathy

^aOf the 139 participants without pRBD, sample sizes were reduced to 136 for dorsal/median raphe nuclei NFTs, 133 for locus coeruleus NFTs, 132 for subcoeruleus NFTs, 138 for basis pontis NFTs, and 136 for substantia nigra NFTs because of missing data.

^bOf the 54 participants with pRBD, sample sizes were reduced to 49 for dorsal/median raphe nuclei NFTs, 51 for locus coeruleus NFTs, 50 for subcoeruleus, and 50 for basis pontis NFTs because of missing data.

Because pRBD pathophysiology has been proposed to originate, in part, in the brainstem [9, 16, 45], we investigated non-LBD pathology in select sleep-relevant brainstem centers. CTE participants with pRBD were more likely (41/49 [84%]) to have p-tau NFTs in the dorsal and median raphe nuclei compared to participants without pRBD (90/136 [66%], P = 0.02; Table 3; Fig. 1b-d; Supplementary Figure 1, online resource). No significant differences emerged between groups in the presence of NFTs in the basis pontis, locus coeruleus, subcoeruleus, or substantia nigra (Table 3), and no differences were observed in CTE stage, AD, or vascular pathology (data not shown). A complementary analysis within pRBD participants with and without LBD showed that those without Lewy bodies had earlier ages of pRBD onset, reported cognitive symptom onset, and death compared to those with Lewy bodies (P’s <0.001; Supplementary Table 1, online resource).

Analysis of probable RBD in the total population of CTE participants

Given the associations between increased frequency of LBD in CTE participants with pRBD and increased frequency of NFTs in the dorsal and median raphe nuclei in the subset of pRBD participants without LBD, we sought to model pathological associations with pRBD in the total population of CTE participants. Variables significantly associated with pRBD were included in a summary binary logistic regression model that included all CTE participants (N = 247). p-tau pathology and LBD were each dichotomized as present versus absent.

Probable RBD was associated with dorsal and median raphe nuclei NFTs (OR = 3.96, 95% CI [1.43, 10.96], P = 0.008), Lewy body pathology (OR = 2.36, 95% CI [1.18, 4.72], P = 0.02), and years of contact sports participation (OR = 1.04, 95% CI [1.00, 1.08], P = 0.04), adjusting for age at death (Table 4). A separate binary logistic regression in the total population showed dorsal and median raphe nuclei NFTs were significantly associated with moderate-severe neuronal loss (OR = 1.72, 95% CI [1.18, 2.53], P = 0.005). Sensitivity analysis restricted to CTE participants whose primary sport had been US football preserved the prediction of pRBD by dorsal and median raphe nuclei NFTs (OR = 3.24, 95% CI [1.12, 9.42], P = 0.03) and Lewy body pathology (OR = 2.77, 95% CI [1.33, 5.74], P = 0.006), but not years of participation (OR = 1.03, 95% CI [0.98, 1.08], P = 0.20).

Table 4.

Associations with probable RBD in the total population of CTE participants

Characteristic	Odds Ratio (95% CI)	P value
Dorsal/median raphe nuclei tau pathology (NFTs)	3.96 (1.43 – 10.96)	0.008
Lewy body pathology	2.36 (1.18 – 4.72)	0.02
Contact sports participation (per year)	1.04 (1.00 – 1.08)	0.04

Binary logistic regression adjusting for age at death. Both tau and Lewy body pathologies were dichotomized as present versus absent. RBD REM sleep behavior disorder; CTE chronic traumatic encephalopathy; NFTs neurofibrillary tangles

Discussion

REM sleep behavior disorder affects an estimated ~1% of the general population and is associated with antecedent TBI [25, 59]. In a group of deceased contact sports athletes with CTE, pRBD was frequent (>30%), and distinct neuropathological and clinical characteristics distinguished participants with and without pRBD. The number of years of RHI exposure sustained through contact sports participation was significantly higher in participants with pRBD. Although pRBD in CTE was associated with increased LBD frequency, over two-thirds (68%) of pRBD participants did not have LBD. These participants reported sleep and cognitive symptoms at younger ages than participants with LBD and had a younger age at death. Neuropathologically, dorsal and median raphe nuclei p-tau pathology was more frequent in non-LBD participants with pRBD. Overall, in the total sample of CTE participants, pRBD was significantly predicted by the presence of dorsal and median raphe nuclei NFTs (OR = 3.96), the presence of Lewy body pathology (OR = 2.36), and years of contact sports play (OR = 1.04 per year of play).

REM sleep behavior disorder is associated with neurodegenerative diseases, particularly synucleinopathies, and although RBD was documented in ~1% of a middle-to-older aged community-dwelling population [25], pRBD was observed in ~17% of a brain bank cohort enriched for aging and neurodegenerative diseases [62]. RBD has likewise been associated with antecedent TBI and other risk factors for PD and with increased LBD severity in PD patients [55, 59]. TBI with loss of consciousness has been associated with increased risk of cortical Lewy bodies and with incident PD and the progression of parkinsonian signs [19, 23], and our group has shown that RHI sustained through years of contact sports play is associated with increased neocortical LBD as well as parkinsonism and dementia [1]. We observed a similar relationship in the present study such that CTE participants with pRBD were found to have participated in contact sports for significantly more years than those who did not develop pRBD, and pRBD participants had increased LBD frequency, suggesting that pRBD in CTE is partially explained by neuropathology common to other pRBD contexts. But our summary model demonstrated increased risk of pRBD from contact sports participation (proxying for RHI) independently, one explanation for which may be that the physical strain and stress exerted by RHI on the interconnected brainstem nuclei causes sufficient neuronal nuclear and axonal damage to produce RBD symptoms [8]. Future studies that isolate the contribution of RHI to pRBD independent of CTE will further inform our understanding of pRBD pathogenesis.

Despite the increased frequency of LBD observed in CTE participants with pRBD, a surprising majority of these participants did not have LBD. We also did not observe increased Lewy body severity per neuroanatomical region, a finding that differs from that of a study of pRBD in PD which found that, compared to PD participants without pRBD, those with pRBD had more severe LBD in every region examined except the frontal cortex [55]. This suggests that α-synuclein deposition alone is insufficient to account for pRBD in CTE, and degenerative pathology of other pathways is likely involved. pRBD has been associated, albeit less frequently, with neurodegenerative disorders other than synucleinopathy [9, 16, 62], and we add to this literature our observation that most CTE participants with pRBD do not have LBD, indicating neurodegenerative pathology other than α-synuclein has underappreciated contribution to the pRBD phenotype. Future investigation will be important to detect additional pathology (e.g., vascular, neuroinflammatory) that may underlie pRBD in CTE, and the factors that influence development of disparate underlying pathologies, such as genetic contribution [37], likewise remain to be explored.

Although much of pRBD pathophysiology is yet to be described, the disorder is proposed to result in part from lesions of the brainstem, particularly those that involve sleep-relevant nuclei and circuits that regulate REM sleep [16, 41, 45, 50, 51, 70]. Two nuclei thought to be particularly salient in sleep regulation are the dorsal and median raphe [50, 51, 70]. Although animal models have been the primary modality used to demonstrate the direct effect of lesioning these regions on sleep regulation [70], studies of human patients with dementia with Lewy bodies or PD or who have sustained midbrain/pontine lesions have documented neuronal degeneration in the dorsal and median raphe that is thought to have clinical consequence [5, 26, 27, 32, 33, 40, 54, 69]. Our analysis of select sleep-relevant centers in the brainstem of CTE participants without LBD revealed that significantly more of those who developed pRBD had p-tau NFTs in the dorsal and median raphe nuclei. It should be noted that several non-LBD CTE participants with and without pRBD were missing data on this parameter, and possible combinations of frequencies could have altered the magnitude of the difference between subgroups, statistically altering our conclusion. However, our tipping point analysis ameliorated this concern by showing that the current findings would be unlikely to be overturned by the missing data given the >80% (83.7%) frequency of dorsal and median raphe nuclei NFTs in participants for whom this data is available. In fact, evaluation of the expected outcome without missing data was projected to be the same as the outcome we actually observed with the available data (Supplementary Figure 1, online resource).

Further strengthening the association between dorsal and median raphe nuclei NFTs and pRBD in CTE, p-tau within the raphe nuclei conferred higher odds of pRBD in the total sample of CTE participants than did Lewy body pathology. Thus, despite the strong historical association of pRBD with synucleinopathy, sleep dysfunction following sustained exposure to RHI likely reflects p-tau in sleep-relevant brainstem regions, not α-synuclein. The equivalent CTE stage observed between those with and without pRBD is likely due to early brainstem involvement in CTE, as early-stage CTE often features p-tau in the locus coeruleus and substantia nigra [44]. pRBD, even in the absence of Lewy pathology, was associated with longer duration of contact sports participation, and, to our knowledge, this is the first study presenting a link between pRBD, contact sports participation, and brainstem tau pathology.

Limitations

There were several limitations inherent to the sample and methodologies of this study. Obtaining postmortem tissue of the athletes under study is principally via self-selection or next-of-kin referral, necessary approaches that unavoidably introduce a well-recognized, autopsy-based selection bias that may hinder generalizability. However, inverse probability weighting recently demonstrated that the relationship between years of contact sports participation and CTE pathology was not significantly affected by study selection [46]. Separately, although expert clinicians conferred clinical diagnoses using research-quality guidelines, the informant-based information obtained via retrospective review could have been subject to recall bias. We sought to minimize limitations by restricting our analyses to CTE participants, all of whom were neuropathologically diagnosed by the same expert neuropathologists utilizing standard research protocols, a restriction intended to reduce the inherent variability of the study population and any residual influence of selection bias.

Because this study involves secondary data analyses of the larger UNITE study, polysomnography (PSG) data were unavailable for our brain donors. We instead abridged the more practical MSQ—one of several instruments developed for pRBD screening, particularly in research settings—by extracting questions about pRBD symptoms. The full-length version could not be administered to both the patient and informant due to the nature of this autopsy-based study. Studies validating the MSQ showed high sensitivity (SN) and specificity (SP) of the principal question on dream enactment, with SN = 98% and SP = 74% in an aging cohort enriched for neurodegenerative diseases with similar demographics to our study population [12] and SN = 100% and SP = 95% in an older-aged, mostly male community-based cohort [11]. As the MSQ has yet to be validated in the setting of RHI, measurement error is an important limitation of this study. We expect misclassification of pRBD to be low given the robustness of the MSQ’s explicit questioning at detecting the specific symptomatology of RBD [11, 12, 56], but whether informants were bedpartners is unclear, limiting reliability of their reporting accuracy of brain donors’ nighttime behaviors.

Future clinicopathologic investigation in additional cohorts will further clarify our understanding of RBD symptoms in CTE. The retrospective data available for the present study precluded the possibility of quantitatively associating the severity of the clinical manifestations with pathological severity, so studies that employ either quantitative rating scales to assess RBD symptoms or PSG to quantify REM sleep without atonia will be an important future direction. It is also unclear whether particular neuronal subtypes within the raphe nuclei are predisposed to involvement by p-tau, but prospective studies utilizing unbiased, random sampling of the neuronally heterogeneous dorsal and median raphe nuclei will be important to accurately determine whether particular neuronal subtypes are differentially affected. In addition, although the central REM sleep generator is thought to localize to the midbrain/pons [63], animal modeling has suggested that disrupting circuitry in the medulla may also be associated with RBD symptoms [71], so studies of the medulla in human participants with a history of head injury will be important to clarify whether additional neuropathology contributes to RBD symptoms in CTE.

Antidepressants and benzodiazepines may affect the presentation of RBD symptoms [20], and because their use was unknown, we cannot exclude the possibility that these medications influenced the results. Likewise, untreated OSA can produce symptoms similar to those of RBD but that resolve with successful treatment [29]; we have carefully denoted diagnosed OSA throughout the present study, but undiagnosed or inadequately treated OSA may have gone undetected or was not captured by informant report. Informant-reported OSA could thus conceivably have introduced a low occurrence of pRBD misclassification. Such classification error, however, is unlikely to have affected our findings because OSA was reported with equivalent frequency in the pRBD and non-pRBD subgroups of the total population and yet the clinicopathological relationships we detected were preserved. Regardless, future studies that PSG-validate the MSQ in populations with antecedent exposure to head trauma will be important to further refine our understanding of RBD symptoms in the setting of RHI and provide generalizable context for the present findings.

Conclusions

In summary, compared to the general population, pRBD was >30x more frequent in a convenience sample of CTE participants in association with increased years of contact sports participation. Participants with pRBD exhibited a higher frequency of LBD; the majority of pRBD participants did not have LBD, however, but rather an increased frequency of p-tau pathology within the dorsal and median raphe nuclei. Overall, RHI may predispose to pRBD in CTE through brainstem pathology involving tau and α-synuclein dependent and independent mechanisms.