Introduction
The role of traumatic brain injury (TBI) in long-term cognitive impairment and decline remains controversial. Although studies have reported that a history of TBI is a risk factor for late-life cognitive impairment and neurodegenerative disease, including Alzheimer’s dementia (Lye & Shores, Reference Lye and Shores2000; Plassman et al., Reference Plassman, Havlik, Steffens, Helms, Newman, Drosdick and Breitner2000; Wang et al., Reference Wang, Lin, Sung, Wu, Hung, Wang and Tsai2012), it is clear that other factors modify this relationship, including injury severity (Rohling et al., Reference Rohling, Binder, Demakis, Larrabee, Ploetz and Langhinrichsen-Rohling2011), frequency (Smith, Johnson, & Stewart, Reference Smith, Johnson and Stewart2013), age (Teasdale, Nicoll, Murray, & Fiddes, Reference Teasdale, Nicoll, Murray and Fiddes1997), as well as genetics (Friedman, Froom, & Sazbon, Reference Friedman, Froom and Sazbon1999) and sex (Stein, Reference Stein2001).
Because these factors influence mechanisms of recovery from TBI and susceptibility to later neuropathology (see Dardiotis, Grigoriadis, & Hadjigeorgiou, Reference Dardiotis, Grigoriadis and Hadjigeorgiou2012; Lye & Shores, Reference Lye and Shores2000), it is important to understand their role in long-term cognitive outcomes post-TBI. A major difficulty in comparing findings across studies is the variety of head-injury variables examined, study duration, sensitivity of outcome measures, and other key factors reported. Furthermore, very few studies have focused on multiple factors of interest such as genotype and sex within the one sample to understand their relative contribution to TBI outcome (Ost et al., Reference Ost, Nylen, Csabok, Blennow, Rosengren and Nellgard2008; Ponsford et al., Reference Ponsford, McLaren, Schonberger, Burke, Rudzki, Olver and Ponsford2011), or TBI in relation to neurodevelopment and change across the lifespan (Teasdale, Murray, & Nicoll, Reference Teasdale, Murray and Nicoll2005). Findings from these studies suggest the relationship between TBI outcome and factors such as genotype, may vary with sex and age. The present study uses a large, prospective dataset to examine three factors of current interest: APOE genotype, sex, and age, in relation to neuropsychological outcome after TBI. In addition, a range of TBI variables were examined in the one sample, including severity, frequency and age of first injury, with the objective of understanding how these TBI variables interact with the above factors to explain cognitive outcome across the lifespan.
Despite increasing knowledge of the molecular processes in brain injury, clinical and epidemiological evidence has been far from unanimous in supporting patterns of associations predicted by molecular studies. The biological response to TBI is highly variable across individuals and injury mechanisms. In general, neurotrauma initiates a complex cascade of molecular processes within the brain some of which exacerbate neuronal death and others which are neuroprotective. These processes include neuroinflammation, oxidative stress, excitoxicity, apoptotic cell death, neurodegeneration, and plasticity. Of the several genes identified as influencing this molecular response to injury, the gene for apolipoprotein E is the most studied and pertinent to the question of late-life neurodegeneration after TBI (Dardiotis et al., Reference Dardiotis, Grigoriadis and Hadjigeorgiou2012; Kokjohn, Maarouf, & Roher, Reference Kokjohn, Maarouf and Roher2012; Liu, Kanekiyo, Xu, & Bu, Reference Liu, Kanekiyo, Xu and Bu2013). Compared to TBI patients who carry only the ε2 and ε3 isoforms of the APOE gene, carriers of the ε4 isoform are more likely to die post-injury (Teasdale et al., Reference Teasdale, Nicoll, Murray and Fiddes1997), have a longer length of coma (Sorbi et al., Reference Sorbi, Nacmias, Piacentini, Repice, Latorraca, Forleo and Amaducci1995), greater neurological deficits (Friedman et al., Reference Friedman, Froom and Sazbon1999), as well as poorer functional outcome several years following rehabilitation (Ponsford et al., Reference Ponsford, McLaren, Schonberger, Burke, Rudzki, Olver and Ponsford2011). The mechanisms underpinning the effect of APOE genotype are not well understood, but may influence the degree of Aβ plaque deposition within the brain (Johnson, Stewart, & Smith, Reference Johnson, Stewart and Smith2010). Acute effects of TBI such as widespread axonal injury, leads to the rapid accumulation of amyloid precursor proteins (APP) within the damaged axons (Gentleman, Nash, Sweeting, Graham, & Roberts, Reference Gentleman, Nash, Sweeting, Graham and Roberts1993) along with a co-accumulation of compounds that promote Aβ formation. This process can persist chronically (Chen, Johnson, Uryu, Trojanowski, & Smith, Reference Chen, Johnson, Uryu, Trojanowski and Smith2009). While some studies suggest that APP plays a protective role post-TBI (Thornton, Vink, Blumbergs, & VanDen Heuvel, Reference Thornton, Vink, Blumbergs and VanDen Heuvel2006), individuals with APOE ε4 + genotype experience increased post-injury Aβ deposition and reduced Aβ clearance (Hartman et al., Reference Hartman, Laurer, Longhi, Bales, Paul, McIntosh and Holtzman2002), which may in turn lead to an increased risk of Aβ plaque formation characteristic of dementia pathology. Compared to APOE ε3, the APOE ε4 isoform of the apolipoprotein E is also less able to repair synapses and protect neurons in the event of injury (Zhong, Scearce-Levie, Ramaswamy, & Weisgraber, Reference Zhong, Scearce-Levie, Ramaswamy and Weisgraber2008), thus hindering recovery in TBI patients with APOE ε4 + genotype.
Some epidemiological studies support this link between APOE genotype and TBI outcome and suggest the combination of TBI history and APOE ε4 + genotypes significantly elevates the risk of long-term cognitive decline or Alzheimer’s disease, beyond the effect of either TBI or APOE ε4 + genotype alone (Katzman et al., Reference Katzman, Glasko, Saitoh, Chen, Pay, Booth and Thomas1996; Rohling et al., Reference Rohling, Binder, Demakis, Larrabee, Ploetz and Langhinrichsen-Rohling2011; Sundstrom et al., Reference Sundstrom, Nilsson, Cruts, Adolfsson, Van Broekchoven and Nyberg2007; Tang et al., Reference Tang, Maestre, Tsai, Liu, Feng, Chung and Mayeux1996). The majority of these studies have included participants carrying the ε2 allele within the reference group despite evidence that ε2 may protect against degenerative neuropathology (Liu et al., Reference Liu, Kanekiyo, Xu and Bu2013). Evidence for the influence of APOE genotype on TBI outcome is not unanimous, however, as others report cognitive decline decades after severe TBI but with no relationship to APOE ε4 + genotype (Millar, Nicoll, Thornhill, Murray, & Teasdale, Reference Millar, Nicoll, Thornhill, Murray and Teasdale2003), and similarly no relationship with cognitive decline within the first two years following mild-moderate TBI (Chamelian, Reis, & Feinstein, Reference Chamelian, Reis and Feinstein2004; Rapoport et al., Reference Rapoport, Wolf, Herrmann, Kiss, Shammi, Reis and Feinstein2008). While there are numerous methodological differences across all studies, it is notable that the latter works used only mild/moderate TBI cases (Chamelian et al., Reference Chamelian, Reis and Feinstein2004; Rapoport et al., Reference Rapoport, Wolf, Herrmann, Kiss, Shammi, Reis and Feinstein2008) or a mixed-age sample with no information on the distribution of ages (Millar et al., Reference Millar, Nicoll, Thornhill, Murray and Teasdale2003). TBI severity is a known factor influencing cognitive outcome and decline (Plassman et al., Reference Plassman, Havlik, Steffens, Helms, Newman, Drosdick and Breitner2000; Polusny et al., Reference Polusny, Kehle, Nelson, Erbes, Arbisi and Thuras2011), and recently age has also been found to influence TBI outcome.
Age may influence the brain’s response to injury (e.g., Adelson et al., Reference Adelson, Jenkins, Hamilton, Robicaud, Tran and Kochanek2001) and also the role of APOE ε4 + in the recovery process (Blackman, Worley, & Strittmatter, Reference Blackman, Worley and Strittmatter2007). Some have suggested that the effect of the ε4 allele on brain biology and function may be beneficial in the young but deleterious at older ages (Han & Bondi, Reference Han and Bondi2008). Although evidence for this type of association in healthy individuals’ cognitive performance is mixed (Bunce, Anstey, Burns, Christensen, & Easteal, Reference Bunce, Anstey, Burns, Christensen and Easteal2011; Ihle, Bunce, & Matthias, Reference Ihle, Bunce and Matthias2012), several studies that have used young adult samples (aged in their 20s and 30s) report improved global outcome (Willemse-van Son, Ribbers, Hop, van Duijn, & Stam, Reference Willemse-van Son, Ribbers, Hop, van Duijn and Stam2007) or faster recovery from TBI on neuropsychological measures of attention, executive function and episodic memory (Han et al., Reference Han, Drake, Cessante, Jak, Houston, Delis and Bondi2007). These studies however, had no older cohort against which the differential influence of APOE could be compared, and both used samples that were all or mostly male. Other works have reported the reverse pattern of findings (Teasdale et al., Reference Teasdale, Murray and Nicoll2005) where the APOE ε4 + genotype was associated with poorer global outcome 6 months post-injury in young patients. This difference decreased with increasing age, such that genotype was unrelated to outcome in the older patients.
Sex may also moderate the relationship between TBI and long-term cognitive outcome in adulthood. There is evidence that circulating estrogen and progesterone have neuroprotective roles in mouse models of brain damage (Ardelt et al., Reference Ardelt, Carpenter, Lobo, Zeng, Solanki, Zhang and Pike2012; Stein, Reference Stein2001). For example, female rats show less brain damage than male rats and ovariectomized female rats and senescent rats show the same degree of brain damage as male rats (see Roof and Hall, Reference Roof and Hall2000). Estrogen and progesterone are thought to minimize secondary injury post-TBI by reducing oxidative stress and neuronal death, as well as reducing cerebral edema and enhancing synaptogenesis (Stein, Reference Stein2001). Consistent with such findings, two meta-analyses of case-control studies have found a stronger association between past TBI and Alzheimer’s disease diagnosis in males than in females (Fleminger, Oliver, Lovestone, Rabe-Hesketh, & Giora, Reference Fleminger, Oliver, Lovestone, Rabe-Hesketh and Giora2003; Mortimer et al., Reference Mortimer, van Duijn, Chandra, Fratiglioni, Graves, Heyman and Rocca1991). However, a meta-analysis of functional outcomes post-injury found poorer return-to-work and somatic-symptoms outcomes in females (Farace & Alves, Reference Farace and Alves2000), and other studies have failed to find any effect of sex on TBI outcome (Renner et al., Reference Renner, Hummelsheim, Kopczak, Steube, Schneider, Schneider and Stalla2012). Some findings suggest that sex may also interact with genotype in predicting outcome of TBI. For example, Ponsford et al. (Reference Ponsford, McLaren, Schonberger, Burke, Rudzki, Olver and Ponsford2011) reported that negative outcomes in APOE ε4 + patients were more apparent in females than males, and particularly in patients aged over 55 years. The authors surmised that older females may experience less neuroprotection from estrogen and be more vulnerable to the effects of APOE ε4 + . Others have reported that the negative effect of APOE ε4 + on outcome following severe TBI is more apparent in males than females (albeit in a sample with mixed ages from 8 to 81 years) (Ost et al., Reference Ost, Nylen, Csabok, Blennow, Rosengren and Nellgard2008).
The complex interactions that appear to influence TBI outcome are difficult to examine in small samples with limited age cohorts and duration of follow-up. Here, we use a large prospectively studied sample to investigate several self-reported TBI variables (severity, age of injury, and frequency) and their interaction with sex and APOE genotype as predictors of cognitive performance. Additionally, we implement a lifespan approach by separately studying these interactions in young, middle-aged and older adults.
Methods
Study Population
The study population was from the Personality and Total Health (PATH) Project. The sample for the PATH project has previously been described in detail (Anstey et al., Reference Anstey, Christensen, Butterworth, Easteal, Mackinnon, Jacomb and Jorm2012) and has been shown to be representative of the 2001 Australian Census data for the area (Anstey et al., Reference Anstey, Christensen, Butterworth, Easteal, Mackinnon, Jacomb and Jorm2012). In brief, we surveyed three cohorts aged 20–24, 40–44, and 60–64 years at baseline with four yearly follow-ups. The sampling frames were derived from the electoral roll in Canberra and Queanbeyan, Australia. Electoral registration and voting are compulsory for Australian citizens. Figure 1 displays the full PATH sample (N=7485) and the samples excluded from analysis (N=220) (see exclusion criteria below). At Wave 1, 2077 of the selected participants in the 20s cohort had data for both TBI and APO ε4 + genotype, 2124 participants in the 40s cohort had both TBI and APOE genotype data, and 2132 participants in the 60s cohort had both TBI and APOE genotype data.
Fig. 1 Flow chart of sample selection and follow-up. *Study sample excludes participants reporting baseline stroke or epilepsy.
Exclusion/Inclusion Criteria
To examine the association between head injury and cognitive change, participants with any history of cerebrovascular events (including “mini-strokes” and TIAs) or epilepsy at baseline were excluded from the analysis. The 60s cohort included a small number of participants with dementia and mild cognitive disorders as classified by a previously described clinical assessment (Anstey et al., Reference Anstey, Christensen, Butterworth, Easteal, Mackinnon, Jacomb and Jorm2012), including comprehensive cognitive and neuropsychiatric evaluation. Thus, data from a total of 6333 participants were available for analysis at Wave 1. Sample attrition at each wave was due to death, loss to follow up, and refusal to respond to study-related questions, and represented on average 6.5% and 16% at wave 1 and 2 respectively.
Measures
Head-injury variables
Participant responses to TBI questions were combined across the three waves of data collection to form the following variables: (1) history of head injury, (2) number of head injuries sustained, (3) severity of head injury, and (4) age of first head injury (see Table 1). Data on the cause of TBI was also collected, but not analyzed in the present study due to small samples, but is nevertheless presented in a descriptive format in Table 2. The variable ‘Severity of TBI’ consisted of two levels – mild TBI and moderate/severe TBI. Mild TBI represented those who reported having a serious head injury but who reported a loss of consciousness (LOC) of less than 60 min or a period of memory loss [post-traumatic amnesia (PTA)] up to 1 day. Moderate to severe TBI was defined as cases who reported a loss of consciousness of 60 min or more and PTA of more than 1 day (Table 1). LOC has been used as an index of TBI severity in other epidemiological studies, and although the more commonly accepted cutoff for LOC in mild TBI is <30 min, this time interval was unavailable in the PATH dataset to enable this categorization. Thus, the more stringent cutoff was used here. As indicated in Table 1, the TBI frequency data were collected only for those injuries with a LOC > 15 min, and, therefore, does not include milder TBIs.
Table 1 TBI variables
*Questions for wave 3 pertained to the period since the last interview at wave 2.
†Those responding “uncertain” were excluded from analyses when using this variable.
††Assumed to represent post-traumatic amnesia (PTA).
Table 2 Baseline characteristics of the study population according to age cohort
Note. BMI = body mass index at baseline; TBI = traumatic brain injury.
Cognitive measures
Cognitive testing relevant to the present study included the following tests of episodic memory [delayed recall after the first trial of the California Verbal Learning Test (CVLT) (Baddeley, Emslie, & Nimmo-Smith, Reference Baddeley, Emslie and Nimmo-Smith1993)], verbal ability/vocabulary [Spot-the-Word test (STW) (Delis, Kramer, Kaplan, & Ober, Reference Delis, Kramer, Kaplan and Ober1987)], processing speed [Symbol-Digit Modalities test (Smith, Reference Smith1982)], working memory [Digit-Span backward (Wechsler, Reference Wechsler1955)], and simple reaction time.
APOE genotyping
DNA was extracted from buccal swabs using QIAGEN DNA Blood kits (#51162; QIAGEN, Hilden, Germany). Two single-nucleotide polymorphisms (SNPs; rs429358 and rs7412) were genotyped to identify APOE genotypes comprised of the APOE-ε2, -ε3, and -ε4 alleles using TaqMan assays (Life Technologies, Carlsbad, CA) as previously described (Jorm et al., Reference Jorm, Mather, Butterworth, Anstey, Christensen and Easteal2007). All genotypes containing the ε4 allele were combined as the composite APOE ε4 + (ε4/ε4; ε4/ε3; ε4/ε2) genotype. Because of the potential protective effects of APOE ε2 allele, we excluded individuals with ε3/ε2 and ε2/ε2 from the reference group. Thus, APOE ε4 + was compared against APOE ε3/ε3.
Procedure
Participants were assessed individually either at their home or at The Australian National University. A trained interviewer administered all neuropsychological tests. Participants used a palmtop computer to answer the survey questions, with assistance from the interviewer if required. The study was approved by the Australian National University Research Ethics Committee, the protocol was completed in accordance with the Helsinki Declaration, and all participants provided written consent.
Statistical Analyses
Four head injury variables were investigated as predictors of cognitive performance. Separate linear mixed-effects models were created for each head injury variable, in which sex, APOE genotype, age cohort, time in study, and their interactions were modeled as predictors (fixed effects) of cognitive performance. Raw scores on the cognitive tests were standardized relative to each age cohort and converted to t-scores (Mean=50; SD=10). For all cognitive tests, including reaction time measures, higher t-scores represent better performance. Each model was adjusted for participants’ level of education and baseline BMI. Intercept and time in study were modeled as random effects. Specifically, we were interested in significant interactions between head injury and time, indicating that groups defined according to their head injury status showed different rates of cognitive change. The results of these analyses are reported under longitudinal analyses (Table 3). We also examined interactions between head injury and APOE ε4 genotype, and head injury and sex which are reported under Baseline Effects in Table 3.
Table 3 TBI as a predictor of baseline and longitudinal cognitive function (B coefficients (standard errors))
§Comparison group for all TBI terms is the No TBI group; comparison group for genotype is APOE ε3/ε3, and for sex comparison is to Female.
*p <.05, ** p <.01.
Results
Cohort Sample Characteristics
The three age cohorts varied significantly in baseline demographic characteristics including sex, body mass index, and years of education (see Table 2). APOE genotype frequencies did not vary with age group (p =.10) and were consistent with Hardy-Weinberg equilibrium predictions within each sample (p >.05 for all) and with other previously published estimates. TBI characteristics differed across the three age cohorts. The youngest age cohort had a higher frequency of self-reported serious head injuries, more childhood injuries, and more reports of multiple head injuries than did the older age cohorts. The middle-aged cohort reported a higher frequency of moderate/severe head injury relative to the other cohorts. The cohorts also differed in the cause of TBI, with more traffic-related TBIs in middle age and more sports related TBIs in the young cohort.
Results of Linear Mixed-Effects Models
Baseline effects involving APOE ε4
Interactions between APOE genotype and TBI in predicting baseline cognitive performance was evident only in the 20s and 40s cohort. There were no effects relating to APOE genotype in the 60s cohort (Table 3). In the 20s cohort, history of head injury predicted lower baseline episodic memory performance in APOE ε4 carriers compared to non-carriers (i.e., APOE ε3 homozygotes) with APOE ε4 carriers scoring on average 5.40 (SEM=1.79) T score points lower than non-carriers (p<.05; Figure 2A). In the 20s cohort, moderate/severe head injury in APOE ε4 carriers was also associated with poorer memory (see Figure 2B; Table 3), and similarly APOE ε4 carriers reporting childhood TBI also had poorer memory (Figure 2C). However, APOE ε4 carriers reporting first TBI as an adult showed better verbal ability than did non-carriers (Figure 2D). APOE ε4 genotype in the context of multiple TBI predicted poor episodic memory in the 20s cohort (Figure 2E). In the 40s cohort, APOE ε4 carriers who sustained their first TBI as a child had slower response times than did non-carriers reporting childhood TBI [B=5.92, (SEM=2.15), p<.01] (Figure 2E).
Fig. 2 TBI as a predictor of baseline cognitive: interactions with APOE genotype. (A) TBI history and genotype as predictor of episodic memory in 20s cohort. (B) TBI severity and APOE genotype as predictors of episodic memory in 20s cohort. (C) Age of first TBI and APOE genotype as predictors of episodic memory in 20s. (D) Age of first TBI and genotype as predictors of verbal ability in 20s cohort. (E) TBI frequency and genotype as predictors of memory in 20s cohort. (F) Age of first TBI and genotype as predictors of reaction time in 40s cohort. (Error bars represent ±1 SEM.)
Baseline effects involving sex
Interactions between gender and TBI in predicting baseline cognitive performance was apparent in all three age groups, although the pattern of association differed across cohorts. In the 20s cohort, males with TBI in adulthood demonstrated poorer verbal ability relative to females (Figure 3A). Males with a single TBI also had poorer verbal ability relative to females (Figure 3B). In the 40s cohort, childhood TBI was associated with poorer memory in females relative to males, and females reporting a single TBI had reduced processing speed compared to males (Figure 3D) (B=4.86, (SEM=1.92), p<.05). A similar pattern was apparent in the 60s cohort. Here, female participants reporting moderate/severe TBI had reduced memory performance relative to their male counterparts [B=8.75, (SEM=4.58), p<.05]. Females reporting mild TBI also had slower reaction times compared to males [B=4.88, (SEM=2.50), p<.05].
Fig. 3 TBI as a predictor of baseline cognitive performance: interactions with sex. (A) Age of first TBI cause and sex as predictors of verbal ability in 20s cohort. (B) TBI frequency and sex as predictors of verbal ability in 20s cohort. (C) Age of first TBI and APOE genotype as predictors of memory in 40s cohort. (D) TBI frequency and sex as predictors of processing speed in 40s cohort. (E) TBI severity and sex as predictors of memory in 60s cohort. (F) TBI severity and sex as predictors of reaction time in 60s cohort. (Error bars represent +/-1 SEM.)
Longitudinal effects
TBI history predicted longitudinal decline in episodic memory [by -0.39 (SEM=0.13)] t-score points per year (p<.01; see Table 3) in the 20s cohort, and also in the 40s cohort [by -0.27 (SEM=0.13), p<.01]. Individuals aged in their 60s who reported a history of TBI showed longitudinal decline in speed of processing, where performance slowed by 0.26 (SEM=0.10) t-scores per year (p<.05) (Table 3). In this cohort, a similar rate of decline in processing speed was associated with reporting a single TBI relative to no TBI. Multiple TBI was associated with decline in memory for the 20s cohort, while single TBI predicted decline in memory in the 40s cohort. In the 20s cohort, both the childhood TBI and adulthood TBI were associated with decline in memory scores over time (Table 3). Adulthood TBI was also associated with longitudinal reduction in memory scores for the 40s cohort, whereas this variable was associated with decline in processing speed for the 60s cohort.
Discussion
Four head injury variables were investigated as predictors of longitudinal cognitive change and baseline cognitive performance within three age cohorts. In addition to finding TBI-related cognitive decline in young as well as older adults, the results indicated that sex and APOE genotype interacted with TBI to predict baseline differences in cognitive performance in several cognitive domains. However, these interactions varied across age groups.
Role of APOE Genotype
The cognitive performance of adults in their 20s and 40s was significantly predicted by the interaction between TBI and APOE genotype. We found that APOE ε4 carriers in their 20s with a history of TBI had poorer baseline episodic memory performance compared to non-carriers. TBI severity also interacted with genotype in this age group, such that participants with moderate to severe injury and the APOE ε4 + genotypes had poorer episodic memory relative to non-carriers with moderate/severe injuries. One possibility is that the detrimental influence of genotype on memory in the 20s cohort may have been driven by severity of injury, and would not be apparent in the absence of those with moderate/severe TBI. In the middle-aged cohort, APOE genotype predicted slower reaction time in the context of childhood TBI. These findings contrast with reports that young APOE ε4 carriers have better global or cognitive outcomes after TBI than non-carriers (Han et al., Reference Han, Drake, Cessante, Jak, Houston, Delis and Bondi2007; Willemse-van Son et al., Reference Willemse-van Son, Ribbers, Hop, van Duijn and Stam2007), and also broader studies indicating that APOE ε4 + genotypes confer cognitive benefits in cognitively normal young adults (Mondadori et al., Reference Mondadori, de Quervain, Buchmann, Mustovic, Wollmer and Schmidt2007). The only finding that may support the pattern of results in these previous works is that APOE ε4 carriers reporting adulthood TBI in the 20s cohort demonstrated better verbal ability than did non-carriers.
Another possible explanation for the discrepancy between ours and previous findings is that our results reflect longer-term cognitive outcomes associated with retrospective TBI history, whereas prior studies focused on relatively shorter-term outcomes (1 month to 3 years post TBI). Young participants in our study who reported adulthood TBI would have been only a few years post-TBI. Our findings are instead consistent with that of Teasdale et al (Reference Teasdale, Murray and Nicoll2005) who reported a negative impact of APOE ε4 genotype on TBI outcome in young patients—an effect which became negligible with advancing age. We recently examined APOE genotype as a predictor of cognitive performance in healthy participants of the PATH dataset and failed to find evidence of a beneficial effect of APOE ε4 + genotype in the young cohort (Bunce et al., Reference Bunce, Bielak, Anstey, Cherbuin, Batterham and Easteal2013). The present work extends this to show that such a relationship is not evident in this dataset even in the context of TBI.
The fact that we found no association between baseline cognition, TBI, and APOE genotype for adults in their 60s contrasts with several studies (Plassman et al., Reference Plassman, Havlik, Steffens, Helms, Newman, Drosdick and Breitner2000; Rohling et al., Reference Rohling, Binder, Demakis, Larrabee, Ploetz and Langhinrichsen-Rohling2011; Sundstrom et al., Reference Sundstrom, Nilsson, Cruts, Adolfsson, Van Broekchoven and Nyberg2007). However, it is consistent with others that have failed to find a relationship between genotype and long-term cognitive outcome in older adults (Rapoport et al., Reference Rapoport, Wolf, Herrmann, Kiss, Shammi, Reis and Feinstein2008; Teasdale et al., Reference Teasdale, Murray and Nicoll2005). It is also possible that any associations between APOE genotype and cognitive outcome in older adults may have been masked by an interaction with sex as was reported by Ponsford et al (Reference Ponsford, McLaren, Schonberger, Burke, Rudzki, Olver and Ponsford2011) in their study. We did not examine interactions between genotype and sex in our study.
Role of Sex
Our findings suggest that the interaction between sex and TBI in predicting cognitive outcome varies across the lifespan. Specifically, we found that compared to females, males in their 20s with a single head injury or TBI as an adult had poorer verbal. In contrast, the effect of sex was largely reversed in both the older cohorts, with females showing lower baseline cognitive performance following TBI. Middle-aged females with multiple TBIs had poorer episodic memory than did males, and females a single TBIs had slower processing speed relative to males. Similarly, in the 60s females reporting either a mild or moderate TBIs had slower reaction times and memory respectively than did males in these categories. Although we did not directly examine participant levels of estrogen and progesterone in this study or control for hormone replacement therapy, the pattern of associations between sex, age, and cognitive correlates of TBI is consistent with age-related changes in the level of female hormones. A similar pattern was observed by Ponsford et al (Reference Ponsford, McLaren, Schonberger, Burke, Rudzki, Olver and Ponsford2011) where global outcome 2 years post-TBI was poorer in females than in males over the age of 55 years. Estrogen is thought to act as a potent antioxidant in the context of injury, while progesterone may act to reduce neuronal hyperexcitability post-injury (Roof & Hall, Reference Roof and Hall2000). Furthermore, estrogen therapy has been found to improve outcomes following cerebral ischemia and TBI in experimental models (Roof & Hall, Reference Roof and Hall2000; Stein, Reference Stein2001). Previous reports also indicate greater history of TBI among males with Alzheimer’s disease, but not in females (Fleminger et al., Reference Fleminger, Oliver, Lovestone, Rabe-Hesketh and Giora2003; Mortimer et al., Reference Mortimer, van Duijn, Chandra, Fratiglioni, Graves, Heyman and Rocca1991). While our study suggests that cognitive outcomes are poorer for older females with TBI history, it is not clear whether this will trigger or exacerbate pathological decline and dementia. Further follow-up of this sample will provide insight into the relationship between sex, TBI, and dementia.
TBI and Cognitive Decline
Our data show that 8-year cognitive decline was significantly predicted by self-reported TBI. In the 20s and 40s cohorts, TBI predicted significant decline in episodic memory but this was not apparent in the older cohort. It is possible that effects on episodic memory reduce over time, but this result may also reflect a greater difficulty in recalling events in the older cohort. Furthermore, the oldest cohort had the lowest baseline scores in episodic memory, and this may have made detection of subsequent decline more difficult. In older adults, however, we found that TBI predicted long-term cognitive decline primarily in speed of processing (Millar et al., Reference Millar, Nicoll, Thornhill, Murray and Teasdale2003; Sundstrom et al., Reference Sundstrom, Nilsson, Cruts, Adolfsson, Van Broekchoven and Nyberg2007). We did not find TBI variables to predict progressive decline in episodic memory in this age group, which contrasts with some previous studies (Rohling et al., Reference Rohling, Binder, Demakis, Larrabee, Ploetz and Langhinrichsen-Rohling2011). Together with the findings for APOE genotype, our data for the older adults suggest that a history of TBI is associated with late-life cognitive decline and impairment in this relatively young older cohort, but there was no evidence for a role for APOE genotype in TBI related impairment in this age group (Millar et al., Reference Millar, Nicoll, Thornhill, Murray and Teasdale2003). Decline in other cognitive domains, and in particular memory, may emerge as the cohort ages. The rate of decline observed in this study is relatively small, amounting to a fraction of a standard score per year. While this is of little clinical significance, the pattern of interaction with genotype and sex is of interest for understanding outcomes in specific patient groups.
In conclusion, our study supports previous findings that TBI predicts long-term cognitive decline in older adults. In addition we found TBI predicted cognitive decline in younger adults. Furthermore, APOE genotype and sex interacted with TBI history in predicting baseline cognitive impairment, and these interactions varied across the three age cohorts. Strengths of the present study include its prospective design, follow-up of over 6000 community based individuals over a period of 8 years, and use of standard neuropsychological measures. Our study also included three age cohorts and multiple TBI variables to provide a thorough examination of TBI relationships with long-term cognitive performance. Significant limitations of the present study include the retrospective self-report of injuries, the non-standard categorization of mild and moderate/severe TBI, and the significantly lower number of reported TBIs in the older cohorts. Furthermore, there was also limited information regarding the timing of TBI events relative to the cognitive assessments.
Acknowledgments
The authors are grateful to the PATH Chief Investigators, Chantal Meslin, Patricia Jacomb, Karen Maxwell, and the PATH interviewers. Susan Tan and Karen Nunweek provided technical assistance with the APOE genotyping. The study was supported by the National Health and Medical Research Council (NHMRC) of Australia Unit Grant No. 973302, Program Grant No. 179805, NHMRC project grant No. 157125, grants from the Australian Rotary Health Research Fund, and the Australian Brewers Foundation. Prof Anstey was funded by NHMRC Fellowships No. 1002560, and Dr. Cherbuin by ARC Fellowship 120100227. Dr. Bunce was funded by the Leverhulme Trust, UK. The authors declare there are no financial or other relationships that could be interpreted as a conflict of interest affecting this manuscript. The authors have no conflicts of interest to declare.
