Traumatic brain injury and age of onset of dementia with Lewy bodies

Abstract

Background

Traumatic brain injury (TBI) with loss of consciousness (LOC) has been associated with earlier onset of mild cognitive impairment, frontotemporal dementia, Parkinson’s disease, and Alzheimer’s disease (AD), but has not been examined as a risk factor for earlier onset of dementia with Lewy bodies (DLB).

Objective

The purpose of this study was to assess the association between a history of TBI and the age of onset of DLB.

Method

Data from 576 subjects with a clinical diagnosis of DLB were obtained from the National Alzheimer’s Coordinating Center (NACC). Analyses of Covariance examined whether self-reported history of remote TBI with LOC (i.e., > 1 year prior to the first Alzheimer’s Disease Center visit) was associated with earlier DLB symptom onset.

Results

Controlling for sex, those with a history of remote TBI had an approximately 1.5-year earlier clinician-estimated age of onset (F = 0.87, p = 0.35) and 0.75-years earlier age of diagnosis (F = 0.14, p = 0.71) of DLB compared to those without a history of TBI, though the differences did not reach statistical significance. Analysis of subjects with autopsy-confirmed diagnoses was underpowered due to the low number of TBI+ subjects.

Conclusions

Remote TBI with LOC was not significantly associated with DLB onset, despite being a significant risk factor for cognitive decline and earlier age of onset in other neurodegenerative conditions. Replication of these results using a larger cohort of DLB subjects with and without a TBI history who have undergone autopsy is indicated, as our TBI+ subjects did show a slightly earlier onset of about 1.5 years. Further investigations into other potential DLB risk factors are also warranted.

INTRODUCTION

A history of traumatic brain injury (TBI) with loss of consciousness (LOC) is increasingly recognized as a risk factor for all-cause dementia [1, 2], Alzheimer’s disease (AD) [3–9], mild cognitive impairment (MCI) [10], and frontotemporal dementia (FTD) [11–13]. However, there is limited research investigating the relationship between TBI and dementia with Lewy bodies (DLB). As a syndrome, DLB is characterized by the presence of at least two of the following core clinical features: fluctuating cognition, recurrent visual hallucinations, REM sleep behavior disorder and spontaneous features of parkinsonism. Pathologically, this neurodegenerative disorder is defined by the presence of cortical Lewy bodies consisting of α-synuclein [14]. Most studies examining TBI as a risk factor for synucleinopathies have focused on the related disorder of Parkinson’s disease (PD), with mixed results [15–19]. A 2013 meta-analysis of 22 studies investigating head trauma and PD found that TBI may increase the risk of PD by a pooled factor of 1.57 [16]. Additionally, a study of sibling pairs with PD found that sustaining a TBI of any severity was associated with an average 3.3-year earlier onset of PD [20]. A meta-analysis published in 2014 focused on mild TBI and PD analyzed five of the studies included in the 2013 meta-analysis [16] and reported a lack of association between mild TBI and PD [18]. The two largest population studies in these meta-analyses, from Denmark [19] and Sweden [21], both reported the risk for PD to be greatest soon after the head injury and dissipates over time, raising concerns for reverse causation, where head injuries preceding the PD diagnosis may be a consequence of prodromal PD rather than causing PD. To address this issue, a more recent investigation [15] assessed the risk of PD associated with TBI and non-TBI physical traumas, as well as different TBI factors, including age of injury, severity, frequency and the interval between injury and PD diagnosis, and reported a 44% increase in risk of being diagnosed with PD after TBI compared to non-TBI trauma, with significantly higher risk proportional to severity and frequency that supports a dose-response relationship. A subsequent study has also reported that an earlier age at first head injury is associated with PD, though the overall association between head injury and PD was not significant [17]. Several studies have reported on the accumulation of α-synuclein following TBI, both in cases of fatal head injury [22] as well as non-human TBI models [23, 24]. Along these lines, a recent study by Crane and colleagues reported that a history of TBI with LOC may be a risk factor for increased Lewy bodies at autopsy as well as the onset and progression of PD [25]. This evidence would suggest that TBI might also increase the risk for developing DLB or hasten its onset due to the accumulation of Lewy body pathology. However, despite this potential link, investigations into the relationship between TBI and DLB are limited.

To our knowledge, only one exploratory analysis, by Boot and colleagues, has evaluated TBI as a risk factor for developing DLB [26]. Those authors compared subjects with a clinical diagnosis of probable DLB (n = 147) to sex- and age-matched healthy controls (n = 294) in three longitudinal studies of aging and dementia at Mayo Clinic. In that investigation, TBI of any severity was not found increase the likelihood of DLB [26]. Recently, we have published a series of papers investigating the onset of neurodegenerative disorders in those with and without a history of TBI and LOC, showing an earlier onset of MCI, AD, and FTD in those with a history of TBI [10, 27–29]. To add to the sparse literature examining TBI history in DLB, the current study evaluated the relationship between a self-reported remote history of TBI with LOC and symptom onset in DLB.

MATERIALS AND METHODS

Subjects

Data were obtained from the National Alzheimer’s Coordinating Center (NACC) Uniform Data Set (UDS) and Neuropathology Data Set (NDS). NACC has collected demographic, genetic, clinical, and autopsy data on healthy controls and subjects diagnosed with various dementias via approximately annual visits from National Institute on Aging-funded U.S. Alzheimer’s Disease Centers (ADC) since September 2005 [30]. Subjects with data from UDS versions 1 and 2 who were aged 45 years or older and had a primary clinical diagnosis of DLB were selected for inclusion. A team of multi-disciplinary ADC clinicians or a single clinician at each ADC formulated the diagnosis of DLB using established diagnostic criteria [31]. In total, data were drawn from 31 ADCs in this sample, using data from September 2005 to March 2015. Available autopsy diagnoses using the 2005 3^rd consortium criteria [31] were reviewed to determine if the accuracy of clinical diagnoses was similar to recently published frequencies [32].

Measures

The NACC dataset consisted of three questions related to TBI history that were answered by the subject and/or informant during ADC visits. ADC clinicians asked subjects if they had ever sustained a TBI resulting in LOC < 5 minutes or ≥ 5 minutes, or if their TBI resulted in a chronic deficit. Responses were coded as absent, recent/active (TBI < 1 year ago or required ongoing treatment), remote/inactive (TBI ≥ 1 year ago with recovery or no current treatment), or unknown. Data comparing a history of single versus multiple TBIs were not available in this version of the NACC dataset. In addition, injury severity cut-offs in the available data were arbitrary, and TBIs resulting in LOC < 5 minutes or ≥ 5 minutes were combined into a single category of TBI with any LOC. To reduce the effect of recent injuries and to examine the effect of remote TBI, only those who reported a TBI occurring ≥ 1 year prior to the time of diagnosis were included. Subjects who sustained a TBI occurring < 1 year prior to the initial visit or anytime thereafter were excluded to reduce the likelihood of more acute effects. Finally, to reduce the possibility that residual cognitive deficits influenced the diagnosis of DLB, those who reported chronic deficits from their TBI were excluded.

The age of DLB onset was assessed using the NACC variable of the estimated age of symptom onset, defined as the ADC clinician’s estimated age of cognitive decline based on subject/informant report, medical records, and observation. Factors including demographics (e.g. sex, race, education), cardiovascular factors (e.g., smoking and history of stroke), first-degree family member with dementia, apolipoprotein E ε4 status (Apoe4; number of alleles), and depression (i.e., recent and lifetime history) were examined as potential covariates.

Statistical Analyses

Chi-square and independent t tests were used to assess whether subjects with a history of TBI with LOC (TBI+) and those without (TBI−) differed by demographics, number of years smoking, history of remote stroke, number of Apoe4 alleles, family history of dementia, and history of depression. Any significant differences that were identified in these variables were used as covariates in primary analyses. Analyses of Covariance (ANCOVA) were used to assess whether age of DLB onset significantly differed between TBI+ and TBI− groups. Missing data were assessed for each individual analysis and subjects missing relevant data were excluded. All analyses were conducted using IBM SPSS Statistics V24 (IBM Corp, SPSS Statistics V24, Armonk, New York, USA, 2015) with p < 0.05 as the level of significance.

RESULTS

Clinical, demographic, and sample characteristics

After excluding six subjects as outliers for time of dementia onset to diagnosis (i.e., ≥ 3 SDs above the mean), a total of 576 subjects with available TBI history and DLB diagnosis were included. Autopsy data concerning Lewy body disease were available in 191 of these subjects. Of these, 73% (n = 140) had an intermediate or high likelihood that the pathologic findings were associated with DLB clinical syndrome according to the 2005 Consortium Criteria [31]. The frequencies obtained in this sample are similar to recently published information [32].

Among the study sample, 526 reported no history of TBI (TBI−) and 50 reported a history of remote TBI with no chronic deficit (TBI+). Demographic information is provided in Table 1. The sample was largely Caucasian and generally well educated, with an average of 14.61 years of education (Table 1). Missing data were as follows: one subject was missing information regarding race, 13% (n = 76) were missing family history of dementia information, 31% (n = 181) were missing Apoe4 status, 7% (n = 38) were missing number of years smoking cigarettes, 4% (n = 25) were missing history of stroke, and 3% (n = 16) were missing recent and lifetime depression history. No subjects were missing education, sex, age of diagnosis, or age of estimated onset. There were significantly more males in the TBI+ group (86%) than the TBI− group (70%, p = 0.014), so sex was used as a covariate in subsequent analyses. The TBI+ and TBI− groups did not differ in race, years of education, number of years smoking cigarettes, family history of dementia, number of Apoe4 alleles, lifetime depression, depression within 2 years of diagnosis, or remote history of stroke (Table 1).

Table 1.

Demographics of a clinically-diagnosed DLB sample with and without a history of remote TBI

	TBI+ (n = 50)	TBI− (n = 526)	p-value
Males, n (%)	43 (86)	366 (67)	0.014*
Caucasian, n (%)	49 (98)	455 (87)	0.06
Education in years, M (SD)	15.3 (3.7)	14.5 (3.6)	0.17
Years smoking, M (SD)	13.7 (14.1)	11.2 (15.7)	0.25
Family history of dementia, n (%)	24 (50)	256 (57)	0.38
Number of APOE ε4 alleles			0.63
Zero APOE ε4 alleles, n (%)	22 (56)	181 (51)
One APOE ε4 allele, n (%)	13 (33)	146 (41)
Two APOE ε4 alleles, n (%)	4 (10)	29 (8)
Depression
Lifetime, n (%)	27 (56)	305 (60)	0.65
Within 2 years of diagnosis, n (%)	22 (46)	286 (56)	0.18
Remote history of stroke, n (%)	3 (6)	39 (8)	0.74

^*p < 0.05.

Note: APOE ε4 indicates apolipoprotein E ε4.

Of subjects with autopsy data with an intermediate or high likelihood that the pathologic findings were associated with DLB clinical syndrome, 131 reported no history of TBI (TBI−) and 9 reported a history of remote TBI with no chronic deficit (TBI+). Due to the limited sample size of TBI+ subjects, statistical analyses comparing groups were not carried out in this neuropathological sample.

TBI history and age of DLB symptom onset and diagnosis

The average estimated age of DLB onset was 1.49 years earlier for the TBI+ group (M = 68.08, SD = 8.96) compared to the TBI− group (M = 69.57, SD = 8.09). Similarly, the TBI+ group (M = 73.75, SD = 9.34) had a 0.75-year earlier diagnosis than the TBI− group (M = 74.5, SD = 7.71). Controlling for sex, no significant differences between TBI+ and TBI− groups (mean age of estimated DLB symptom onset, F (1, 573) = 0.87, p = 0.35; mean age of DLB diagnosis, F (1, 573) = 0.14, p = 0.71) were observed.

DISCUSSION

A history of TBI has been found to be related to an increased risk and earlier onset of several neurodegenerative conditions. In this study, we compared the average age of DLB onset in those with and without a history of TBI in 576 subjects with the clinical syndrome DLB in the NACC database. While the 50 subjects with DLB and a history of TBI had an approximately 1.5-year earlier onset of symptoms and 0.75-years earlier diagnosis of DLB compared to those without a history of TBI, the differences did not reach statistical significance.

This study marks only the second time in the literature that TBI has been investigated as a potential risk factor for DLB. The previous study was an exploratory investigation into 19 risk factors that may contribute to DLB conducted by Boot and colleagues [26], which showed that, in a cohort of 147 subjects with DLB, 18 of whom reported a history of head injury, a self-reported history of head injury was not associated with a significantly increased risk of developing DLB compared to control subjects (odds ratio of 1.4, 95% confidence interval 0.7 – 2.6, p value 0.33). Despite methodological differences, our findings are in keeping with the prior negative findings for a relationship between TBI and increased risk of DLB and support the conclusion that a history of TBI is not associated with significantly earlier onset or risk of developing DLB.

The current findings with respect to DLB, however, are in contrast with our previous findings between TBI history and onset of several other neurodegenerative conditions. In our study of subjects with a well-characterized clinical history of AD, a history of TBI was associated with a significantly earlier average age of onset by approximately 2.5 years [27, 29]. Those findings are consistent with some [3–9], but not all studies investigating TBI in AD [33–35]. Similarly, TBI has been associated with an approximately 3-year earlier age of FTD symptom onset and diagnosis [10], supporting other studies that have found a significant relationship between TBI and increased risk of FTD [11–13]. In PD, TBI was concluded to be a risk factor (pooled OR = 1.57) for the development of PD in a meta-analysis of 22 studies [16]. The negative findings in this study are curious when compared to one study that found a 3-year earlier age of PD onset [20], which is neuropathologically similar to DLB. Taken together, these findings suggest that TBI may increase the risk of several neurodegenerative diseases to varying degrees, perhaps due to disease-specific neuropathological mechanisms.

Although TBI may be a risk factor for earlier onset of AD, FTD, and PD, why TBI did not appear to be significantly associated with the earlier onset of DLB remains unclear. PD has a similar neuropathological profile to DLB, with the accumulation of α-synuclein-containing Lewy bodies and Lewy neurites, and several studies have shown an increased risk of developing PD after TBI [15–17]. Interestingly, in their recent analysis of data from three prospective cohort studies with a total of 7,130 subjects, Crane and colleagues showed that a history of TBI with LOC was associated with incident PD, progression of parkinsonism, and findings of cortical Lewy bodies at autopsy. TBI history was not, however, related to cognitive impairment or dementia in PD that is typically associated with cortical Lewy bodies [25]. Gardner and colleagues have hypothesized that TBI may decrease motor reserve or accelerate a pre-existing neurodegenerative cascade, or even trigger a de novo cascade [15]. Jafari and colleagues posited neuroinflammation, mitochondrial dysfunction, or glutamate toxicity as mechanistic possibilities for the association between TBI and PD [16]. If these factors are indeed playing a role, then perhaps the pathological changes instigated by TBI promote either an earlier or more prevalent movement disorder manifestation of synucleinopathy to result in a PD phenotype as opposed to the cognitive manifestations of DLB. Alternatively, the impact of TBI on synucleinopathy may be restricted to brainstem-predominant α-synuclein pathology typical of PD, but not sufficient for more diffuse cortical spread of α-synuclein to the degree necessary to induce cognitive impairment and dementia seen in DLB. Clearly, further investigation is warranted to investigate the differential impact of TBI on neurodegenerative diseases.

There are several limitations and caveats to the present study. First, while the difference in age at onset and diagnosis of DLB in those with a history of TBI was not statistically significant, the TBI+ sample did have an onset 1.49 years earlier than the TBI- group, and showed a symptom onset that was 0.75 years earlier. When compared to our previous studies investigating earlier onset in other neurodegenerative diseases in relation to a history of TBI, substantially smaller group sizes due to the lower frequency of DLB in the NACC dataset could make it difficult to detect a statistically significant difference in the present sample. Although this is the largest study to date investigating TBI and risk of DLB, future studies could aim to investigate this relationship in larger samples, perhaps using combined datasets. In addition, due to the confines of the available data, age of injury was not obtained in the previous versions (i.e., UDS versions 1 and 2) of NACC’s data collection forms, making it impossible to explore the interactions between more recent versus remote TBIs. Similarly, other injury-related information, such as the mechanism of injury, subsequent injury-related medical complications, TBI severity, or history of multiple TBIs were not available in the previous versions of NACCs data collection forms and thus were not incorporated into our analysis. These are important factors for characterizing TBI, and should be assessed in future studies to better understand the role of TBI and risk of neurodegenerative diseases including DLB. Also, given the limited autopsy data available at this time, the diagnosis of DLB was based on clinical criteria. Lastly, as with many other NACC-based investigations, subjects in the current study were largely well-educated, male Caucasians who were recruited via clinician referral or self-referred into a voluntary longitudinal observational study, which may not accurately represent the general population.

In conclusion, our findings presented here do not suggest a statistically significant relationship between a self-reported history of remote TBI with LOC and an earlier onset of DLB symptoms or diagnosis. This is consistent with previous findings suggesting that TBI history is not associated with developing DLB [26]. Our present findings, however, contrast with our previous studies indicating that a history of remote TBI with LOC was associated with a significantly earlier onset of MCI, AD and FTD by approximately 2.5 to 3 years [10, 27–29]. The mechanisms by which TBI may increase the risk of developing different neurodegenerative diseases, or even the phenotypic expression within a neuropathological process, remains poorly understood and warrants further investigation. Such studies should focus on comparing the effects of TBI on AD and FTD as well as the motor and cognitive presentations of synucleinopathies.