Variants of SLC6A4 in depression risk following severe TBI

Abstract

Background: Post-traumatic depression (PTD) may be a result of several factors like secondary injury chemical cascades as well as psycho-social factors following traumatic brain injury (TBI). While the role of serotonin in the pathology and treatment of idiopathic major depression may be somewhat controversial, it is unclear what role serotonin may play in PTD following a TBI.

Objective: To assess serotonergic function and genetic risk for PTD development over 1 year following TBI.

Research design: Examination of variation in the serotonin transporter gene [SLC6A4 (5-HTTLPR, rs25331, and a variable number of tandem repeats variant in Intron 2)] in 109 subjects with moderate–severe injury. Depression was assessed using the Patient Health Questionnaire (PHQ-9) at 6 and/or 12 months post-injury.

Main outcomes and results: At 6 months post-injury, subjects with a history of pre-morbid mood disorders and 5-HTTLPR L-homozygotes were at greater risk for PTD. Contrary to major depression, subjects without pre-morbid mood disorders (n = 80) and S-carriers were 2.803-times less likely to be depressed compared to L-homozygotes. At 12 months post-injury, L_G-carriers were also less likely to experience PTD. Temporal analysis also showed 5-HTTLPR associations in PTD development across recovery.

Conclusions: This study suggests a unique injury- and temporally-specific interaction between TBI and genetic risk for depression.

Introduction

Post-traumatic depression (PTD) is the most common neurobehavioural complication following traumatic brain injury (TBI). Individuals with TBI are 10-times more likely than the general population to suffer from a depressive episode during their first year of recovery [1]. As individuals with TBI comprise a significantly understudied proportion of the yearly major depressive disorder (MDD) incidence, examination of this specific population is important and novel and begins to address an important gap in the literature. While studies have examined demographic variables (e.g. pre-morbid or family psychiatric illness, substance abuse or socioeconomic class) [2] in PTD development, they do not fully explain this increased risk for depression. For example, several studies suggest that pre-morbid mood disorder status is a risk factor for post-TBI depression, yet these studies also show an increased risk for subjects with no pre-injury history of mental health disorders [1, 3]. Other factors, like injury severity, as measured by the Glasgow Coma Scale [4], do not typically correlate with degree/incidence of depressive symptoms post-TBI [5].

With an already compromised central neurologic system, the addition of depression following injury can further hinder recovery [6]. Depression in TBI has also been associated with aggression, anxiety, suicidality, and increased healthcare costs [1, 2, 7]. Some studies demonstrate that remittance of depressive symptoms allows for greater recuperation of cognitive function post-TBI [8]. These observations suggest that early treatment of PTD may be a key component of effective recovery, yet intervention is complicated by a lack of efficacious treatment paradigms and a poor understanding of the underlying neuropathology in PTD.

Decades of research in MDD have explored the role of serotonin (5-HT) in depression aetiology and treatment [9]. The serotonin transporter (5-HTT), which modulates the duration and intensity of 5-HT action on its target receptors through re-uptake, has been an important and controversial target for genetic studies in major depression. The most frequently studied variation of this 5-HTT gene (SLC6A4, located on chromosome 17q11.1-17q12) is a 44 bp insertion/deletion in the promoter region of SLC6A4 and is known as the serotonin-transporter-linked polymorphic region (5-HTTLPR). The 5-HTTLPR has two common variants: the long (L) and short (S) allele, based on presence and absence of the insertion, respectively. The L-allele has a reported 3-fold higher transcriptional activity in vitro compared to the S-allele [10], possibly resulting in comparatively less synaptic 5-HT. One meta-analysis showed variation at 5-HTTLPR is implicated in major depression risk [11], but conflicting meta-analyses exist [12], potentially due to specific methodological differences. The meta-analysis by Karg et al. [11] included specific physical stressors (e.g. hip-dislocation) and showed the S-allele to be a significant factor in MDD risk. S-carriers also show increased emotionality, increased hypothalamic-pituitary-adrenal axis reactivity and greater reactivity of the amygdala in response to fearful faces [13]. However, as stated in a review by Homberg and Lesch [14], the S-allele is a conserved polymorphism across evolution (non-human primates also share this polymorphism) and likely has an efficacious role as well. Some argue that S-carriers, when compared to L-homozygotes, show better cognitive performance, with some caveats, in areas that may be related to increased attention. There is also a single nucleotide polymorphism (SNP, rs25531, A > G), within 5-HTTLPR that is found almost exclusively within the L-allele [15]. The G-variant results in an additional AP2 transcription factor binding site; in vitro studies have shown that the L_G-allele is functionally similar to the S-allele [16]. Also within SLC6A4, there is a variable number of tandem repeats (VNTR) within Intron 2 that results in three variants: 9, 10 and 12 repeats of a 17 bp repeat segment [17]. While it is not clear if the VNTR has a true functional effect on 5-HTT protein expression, the 9-allele has been associated with unipolar depression [17] and the 12-allele with bipolar disorder [18]. From this evidence, it is likely that genetic variation within SLC6A4 plays a complex role in depression and mood disorder risk.

There is a lack of literature on the effects of TBI on the serotonergic system, and the relationship between serotonergic function to both mood and cognitive performance post-TBI has not been well studied. Several studies suggest a hypo-neurotransmission state following TBI, including monoaminergic neurotransmitter systems, that likely impacts chronic TBI pathology and cognitive dysfunction [19]. Some studies even suggest that anatomic location of injury, specifically left orbitomedial frontal lobe lesions that directly affect serotonergic innervation [20], is a major contributor to PTD development, but this pathology is not apparent in a large number of persons with PTD [5]. One study showed that serum serotonin profiles of subjects with TBI fall below control subject levels during recovery, with even lower levels observed in PTD [21]. While serotonergic systems have not been fully investigated in TBI, this neurotransmitter system is increasingly important to consider as anti-depressants like selective serotonin re-uptake inhibitors (SSRIs) are a routine part of care for those with MDD, yet seem to have limited effectiveness in PTD [7]. Interestingly, 5-HT_1A receptor agonist administration confers some neuroprotective and cognitive benefits when administered acutely in animal models of TBI [22, 23], but their role in PTD has not been studied.

Considering the increased risk for depressive episodes in the first year post-TBI and the lack of understanding regarding the PTD aetiology, this study aimed to investigate SLC6A4 genetic variation in PTD development. As recovery from TBI is a dynamic and complicated process, PTD was examined across the first year of recovery and in relation to pre-morbid mood disorders. It was hypothesized that genetic variation within SLC6A4 would influence depressive symptomology in subjects with TBI. It was expected that subjects with a history of mood disorders may have varied temporal onsets and genetic risk factors for PTD. It was also hypothesized that genetic associations with PTD could differ over time, wherein PTD occurring at early recovery time-points may be biologically-driven and reflect a low serotonin environment compared to later time-points, where PTD may be driven more heavily by self-awareness of deficits and other environmental stressors.

To the authors’ knowledge only one other study has investigated the role of SLC6A4 in PTD incidence at 1 year post-injury, with no significant associations identified [24]. However, this current study presents novel genetic associations for PTD across the first year post-TBI. The results show that subjects with pre-morbid mood disorders were more likely to develop depression post-TBI, and, consistent with major depression, tended to carry the S-allele. Subjects with pre-morbid mood disorders were also more likely to experience a persistent sub-type of PTD. Among those without a history of pre-morbid mood disorders, carriage of the S-allele was protective against PTD at 6 months post-TBI. At 12 months post-injury, the S-allele was not associated with PTD status, but the L_G-allele was protective against PTD. These data suggest a complex and dynamic gene-injury interaction with PTD over time and suggest a need for further investigation into the role of SLC6A4 in both PTD and general TBI recovery.

Methods

Participants

Included for analysis were 109 Caucasian participants aged 17–71 (mean age 35.33 ± 14 years) receiving care at inpatient and/or outpatient clinics within the University of Pittsburgh Medical Center (UPMC) after sustaining a moderate-to-severe, non-penetrating brain injury, with CT evidence of intracranial injury. Participants were a sub-set of a larger study investigating genetic factors and recovery in individuals with TBI. Participants were included in this analysis if they were cognitively able to report symptoms on the Patient Health Questionnaire–9 (PHQ-9) at 6 or 12 months post-injury.

Participants had a GCS score [4] of 3–15 (mean GCS, 7.76 ± 2.82, median = 7) when using the best GCS obtained within the first 24 hours post-injury. Demographic information including age, gender and education, was collected through chart review as well as subject and caregiver interviews. Anti-depressant usage at 6 and 12 months was extracted from both subject interview and chart review. Subjects were not stratified by anti-depressant use for analysis as many individuals with PTD have a reduced response to SSRIs [25] or are prescribed SSRIs for reasons other than depressive symptoms [26]. However, anti-depressant use was considered during multivariate analysis. A pre-morbid history of mood disorders, including depression, bipolar disorder and anxiety at each time point, was established by self-report and chart review.

Depressive symptom assessment

Depressive symptomology was evaluated at 6 and 12 months post-injury using the Patient Health Questionnaire-9, a brief self-report symptom inventory based on the nine DSM-IV diagnostic criteria for Major Depressive Disorder. The PHQ-9 requires participants to rate, on a scale between 0 (None) and 3 (Nearly Every Day), how often they experience symptoms of depression over a 2-week period. A higher total score reflects a greater number of and/or greater severity of depressive symptoms, with the maximum score being 27. The PHQ-9 can be used both as a continuous and categorical outcome measure. Participants with TBI were grouped as ‘depressed’ vs ‘non-depressed’ using the DSM diagnostic criteria as they map to specific PHQ-9 questions as previously described [27]. Individuals were categorized as depressed if they responded positively to at least five symptom-based questions on the PHQ-9, with at least one pertaining to a cardinal symptom (anhedonia or depression). This method has been validated in a population with TBI showing a sensitivity of 93% and a specificity of 89% compared to the Structured Clinical Interview for DSM Diagnosis (SCID), which is also modelled on DSM diagnostic criteria [27]. This method has also been examined for its ability to discriminate between chronic TBI and depression symptoms [28].

Depression was also evaluated across recovery in a sub-set of participants (n = 67) with both 6 and 12 month depression data. For this temporal analysis, each participant was categorized into a sub-type of PTD as follows: no depression (at either time point), transient depression (depressed at 6 months, but not at 12 months), persistent depression (depressed at both time points) and late-onset depression (depressed at 12 months only).

Genotyping

DNA was isolated from either blood using a simple salting out procedure or from cerebrospinal fluid using the Qiamp protocol from Qiagen (Valencia, CA). For genotyping serotonin transporter gene (SLC6A4) promoter polymorphism 5-HTTLPR, flanking primers 5′-TCCTCCGCTTTGGCGCCTCTTCC-3′ (forward) and 5′-TGGGGGTTGCAGGGGAGATCCTG-3′ (reverse) were used for DNA amplification (PCR) [15]. The rs25531 A > G single nucleotide polymorphism was concurrently detected by digesting the amplified fragments with the restriction enzyme MspI (New England Biolabs, Beverly, MA), where the A > G substitution creates an additional MspI site. Amplification products were simultaneously resolved by electrophoresis on 3.5% agarose gels and visualized by ethidium bromide staining and UV transillumination [29]. For genotyping the STin2 VNTR, flanking primers 5′-GTCAGTATCACAGGCTCGAC-3′ (forward) and 5′-TGTTCCTAGTCTTACGCCAGTG-3′ were used. Sequence confirmed controls of each genotype were run with each plate.

Genetic variant frequencies did not differ by demographics or clinical variables (data not shown). As this was a genetic association study, the potential for stratification effects was examined. All associations were analysed in a Caucasian population (n = 109). During this study, African-American subjects were also recruited and completed PHQ-9 assessments post-injury, yet the sample size for the African-American population with depression data (n = 10) was not sufficient for sub-analysis. There is also consistent evidence of race variation in SLC6A4 allele distribution [12] and, consistent with the literature, the allelic frequencies for the variants analysed differed by race [30–32] (data not shown). For these reasons, the analyses presented here are limited to the Caucasians sub-population described above.

Statistical analysis

The Statistical Package for Social Sciences (SPSS, version 20) was used for data analysis. Descriptive analysis included mean and standard deviation for continuous and ordinal variables such as age, GCS and education. Frequencies were calculated for categorical variables such as gender and anti-depressant use. Genetic information was analysed and categorized based on genotype and allele. Demographic and relevant clinical information was compared between depression groups and variant/genotype groups using Student’s t-tests or an ANOVA to compare means and Chi-Square or Fisher’s Exact to compare frequencies. In order to control for demographic/clinical or injury severity information and control for potential confounders, multivariate logistic regression was used when identifying associations with the absence/presence of PTD. Variables were entered in the logistic regression model based on their bivariate associations with depression. Groupings to assess multi-variable interaction effects were avoided to limit the possibility of having groups with small numbers in the multivariate regression. Variables with a p-value less than 0.3 in bivariate analysis were initially entered in each model. A backward selection method was then used to systematically remove non-significant variables from the model. The final regression model included variables that had a final p-value less than 0.2. However, additional regression analyses incorporated clinically relevant variables (gender, injury severity, education) that have been previously associated with TBI outcome risk, regardless of their p-value. These variables were forced into the model in order to (1) show stability of associations across multiple modelling methods and (2) add clinical relevance to these models.

Results

Population description

This study included 109 individuals with both genotype information and at least one PHQ-9 score (at 6 or 12 months). Of these 109 subjects, 35 were categorized as depressed at 6 months (38.46%, n = 91), while 29 subjects were depressed at 12 months (30.53%, n = 95); 46 subjects experienced depression at some point during the first year post-injury (42.20%, n = 109). Fifteen subjects (13.76%) reported some history of pre-morbid mood disorders, with 14 of these subjects having some element of depression as a part of their pre-morbid psychiatric presentation. The remaining subject had a history of anxiety. Subjects with pre-morbid mood disorders did not differ in age, education or injury severity compared to subjects without any history of mood disorders (data not shown). However, more subjects with pre-morbid mood disorders were taking an anti-depressant at 6 months (81.8%) compared to subjects with no history of mood disorders (33.8%, X² = 9.345, p = 0.003). This effect was reduced at 12 months (53.3% compared to 30%, X² = 3.079, p = 0.075). For those with no history of mood disorders, there were no statistically significant differences in demographic profiles between depressed and non-depressed groups (Table I).

Table I.

 Demographics and allele distribution for subjects with no history of depression (excluding pre-morbid subjects), n = 94.

	6 months (n = 80)			12 months (n = 80)
Variable	Depressed (n = 27)	Not depressed (n = 53)	p-value	Depressed (n = 21)	Not depressed (n = 59)	p-value
Age (mean ± SD)	37.8 ± 14.6	33.3 ± 14.4	0.116	37.5 ± 14.2	34.0 ± 14.0	0.338
Education (mean ± SD)	13.0 ± 2.0	13.0 ± 1.9	1.000	13.2 ± 2.4	13.0 ± 1.7	0.615
GCS (mean ± SD)	7.9 ± 2.6	7.9 ± 2.8	0.979	8.0 ± 2.5	7.5 ± 2.9	0.549
Male, n (%)	21 (77.7)	46 (86.8)	0.345	17 (81.0)	47 (79.7)	1.000
Antidepressant use, n (%)	12 (44.4)	15 (28.3)	0.213	9 (42.9)	15 (25.8)	0.172
5-HTTLPR S-carriers, n (%)	11 (40.7)	35 (66.0)	0.035	13 (61.9)	37 (62.7)	1.000
5-HTTLPR LA-carriers, n (%)	23 (85.2)	43 (81.1)	0.763	19 (90.5)	47 (79.7)	0.334
5-HTTLPR LG-carriers, n (%)	6 (22.2)	7 (13.2)	0.345	0 (0.0)	12 (20.3)	0.019
VNTR 9- carriers, n (%)	3 (11.1)	1 (1.9)	0.109	1 (4.8)	3 (5.1)	1.000
VNTR 10- carriers, n (%)	16 (59.3)	27 (50.9)	0.636	12 (57.1)	30 (50.8)	0.800
VNTR 12- carriers, n (%)	18 (66.6)	45 (84.9)	0.083	15 (71.4)	47 (79.7)	0.544

Pre-morbidity and post-traumatic depression

Results comparing pre-morbid mood disorder status and PTD are presented in Table II. Those with pre-morbid mood disorders were more likely to be depressed at 6 months post-injury compared to subjects with no history of mood disorders (p = 0.006, n = 91, Figure 1(a)). This greater risk for depression was also observed at 12 months post-injury (p = 0.040, Figure 1(b)). Temporal analysis of pre-morbid status showed that subjects with pre-morbid mood disorders had a different distribution within temporal PTD sub-types compared to subjects with no history (p = 0.036) and were more likely to experience persistent PTD than other temporal sub-types of PTD (p = 0.015) (Figure 1(c)).

Table II.

 Demographics and analysis of pre-morbid history in PTD (6 and 12 months), n = 109.

	6 months (n = 91)			12 months (n = 95)
Variable	No history (n = 80)	Pre-morbid (n = 11)	p-value	No history (n = 80)	Pre-morbid (n = 15)	p-value
Age (mean ± SD)	34.8 ± 14.6	38.0 ± 13.6	0.495	34.9 ± 14.0	37.8 ± 12.0	0.462
Education (mean ± SD)	13.0 ± 2.0	12.1 ± 1.8	0.154	13.0 ± 1.9	12.1 ± 1.7	0.081
GCS (mean ± SD)	7.9 ± 2.7	6.5 ± 2.3	0.118	7.6 ± 2.8	7.2 ± 2.4	0.570
Male, n (%)	67 (83.8)	7 (63.3)	0.208	64 (80.0)	10 (66.7)	0.310
Antidepressant use, n (%)	27 (33.8)	9 (81.8)	0.006	24 (30.0)	8 (53.3)	0.075
Depressed, n (%)	27 (33.8)	8 (72.7)	0.016	21 (26.3)	8 (53.3)	0.040
5-HTTLPR S-carriers, n (%)	46 (57.5)	8 (72.7)	0.266	50 (62.5)	11 (73.3)	0.311
5-HTTLPR LA-carriers, n (%)	66 (82.5)	8 (72.7)	0.336	66 (82.5)	12 (80.0)	0.531
5-HTTLPR LG-carriers, n (%)	13 (16.3)	1 (9.1)	0.466	12 (15.0)	2 (13.3)	0.615
VNTR 9- carriers, n (%)	4 (5.0)	0 (0.0)	0.592	4 (5.0)	0 (0.0)	0.497
VNTR 10- carriers, n (%)	43 (53.8)	5 (45.5)	0.422	42 (52.5)	8 (53.3)	0.589
VNTR 12- carriers, n (%)	63 (78.8)	9 (81.8)	0.587	62 (77.5)	12 (80.0)	0.567

Figure 1.

Pre-morbid mood disorder history in PTD risk across recovery. (a, b) Subjects with a history of mood disorders prior to a brain injury were more likely to experience PTD at 6 months (*a, p = 0.016, χ²= 6.207, n = 91) and 12 months (*b, p = 0.040, χ²= 4.369, n = 95). (c) There was also a significant association for pre-morbid mood disorder status in prediction of PTD sub-type (p = 0.036, χ²= 7.308, n = 81), such that subjects with pre-morbid mood disorders were more likely to have some form of PTD compared to subjects with no history (*p = 0.047). This difference was primarily related to more subjects with pre-morbid history experiencing persistent PTD (**p = 0.015).

When evaluating bivariate associations performed for the entire population, there were no significant genetic associations with PTD at 6 or 12 months (Table II). While not statistically significant, 16.6% of 5-HTTLPR S-carriers in our study reported a history of pre-morbid mood disorders (compared to 9.3% of non-carriers, p = 0.396, data not shown). As subjects with pre-morbid mood disorders were more likely to be depressed at 6 and 12 months post-TBI, it was hypothesized that an interaction between pre-morbid status and genotype might exist for PTD. Subjects were evaluated in bivariate analysis using a categorical variable comprising pre-morbid and S-carrier status; PTD rates did vary by pre-morbid history and by S-carrier status (Fisher’s Exact, χ²= 10.969, p = 0.007, Figure 2(a)). Post-hoc analysis showed that S-carriers with no pre-morbid mood disorders exhibited the lowest depression frequencies (23.9% depressed), while S-carriers with pre-morbid mood disorders had the highest depression frequencies (75.0% depressed, p = 0.009). Interestingly, at 12 months, the effects of pre-morbid history by S-carrier status on PTD status in bivariate analysis was reduced (p = 0.123) (Figure 2(b)). Interactions between other variants in SLC6A4 (rs25531, L_G and the VNTR in Intron 2) and pre-morbid mood disorder status were evaluated for PTD risk at both time-points, but no significant results were observed (data not shown).

Figure 2.

(a) Subjects grouped by 5-HTTLPR S-carrier status and pre-morbid mood disorder history show different rates of depression at 6 months (n = 91; Fisher’s exact, χ²= 10.696; p = 0.007) based on percentage depressed in each group (y-axis). Pair-wise contrasts show that subjects with pre-morbid mood disorders and who were also S-carriers, were more likely to be depressed, compared to S-carriers with no pre-morbid mood disorder (*p = 0.009); within subjects with no history of mood disorders, S-carriers were less likely to be depressed (**p = 0.035). (b) At 12 months, 5-HTTLPR S-carrier status and pre-morbid mood disorder history did not significantly differ in depression rates (p = 0.123). Pair-wise contrasts showed trending differences such that subjects with pre-morbid mood disorders and who were non-carriers were more likely to be depressed compared to subjects who had no history of mood disorders (with both S-carriers, #p = 0.089, and non-carriers, #p = 0.073, in this group).

Multivariate models including pre-morbid mood disorder status and also S-carrier status showed these variables were consistently the two most important factors in 6 month PTD risk. In a forced entry model with other covariates (age, gender, GCS, education level and anti-depressant use) entered, subjects with a history of pre-morbid mood disorders were still at the greatest risk for PTD (OR = 5.74: CI = 1.133–29.092, p = 0.035), followed by S-carrier status, where L-homozygotes had the next greatest risk (OR = 3.343, CI = 1.135–9.849, p = 0.029) for PTD. In a backwards stepwise model, pre-morbid status and S-carrier status were the only variables significantly affecting PTD status (Table III).

Table III.

 Multivariate analyses of genetic variants in PTD.

	Odds ratio	CI (95%)	p-value
(a) Pre-morbid status and 5-HTTLPR in all subjects at 6 months post-TBI, forced entry model
Variable (n = 90*)
Pre-morbid history	5.740	1.133–29.092	0.035
5-HTTLPR L-homozygotes	3.343	1.135–9.849	0.029
Antidepressant Use	1.450	0.478–4.400	0.512
Age	1.020	0.980–1.063	0.332
Gender	0.645	0.187–2.226	0.488
GCS	0.925	0.743–1.151	0.483
Education	1.069	0.816–1.400	0.629
(b) Pre-morbid status and 5-HTTLPR in all subjects at 6 months post-TBI, backwards conditional step-wise model
Variable (n = 90*)
Pre-morbid history	6.943	1.606–30.011	0.009
5-HTTLPR L-homozygotes	2.715	1.070–6.885	0.035
All other terms from Model (a) removed through 5 steps when p > 0.2.
(c) 5-HTTLPR in subjects with no history of mood disorders, forced entry model
Variable (n = 79*)
5-HTTLPR L-homozygotes	2.803	1.032–7.614	0.043
Antidepressant use	1.941	0.682–5.521	0.214
Age	1.014	0.977–1.051	0.466
Gender	0.537	0.144–2.006	0.355
GCS	0.983	0.808–1.196	0.865
Education	1.009	0.774–1.316	0.946
(d) 5-HTTLPR in subjects with no history of mood disorders, backwards conditional step-wise model.
Variable (n = 79*)
5-HTTLPR L-homozygotes	2.932	1.094–7.857	0.033
Antidepressant use	1.965	0.716–5.391	0.190
All other terms from Model (c) removed through 5 steps if p > 0.2.

*One subject was missing Education information.

Genetic associations with PTD among those without pre-morbid mood disorders

Among those with no history of mood disorders, Table I (n = 94), S-carriers had lower PTD rates at 6 months post-injury (Fisher’s exact, p = 0.035), with no significant differences by S-carrier status at 12 months (p = 1.00) (see Figures 3(a) and (b)). L_A- and L_G-carrier status (rs25531) (L_A/L_A vs L_A/L_G, Fisher’s exact, χ²= 0.952, p = 0.457) was not associated with PTD at 6 months (Figure 4(a)). However, at 12 months post-injury, no L_G-carriers were depressed (p = 0.019, Fisher’s exact, χ²= 5.025, n = 80) (Figure 4(b)). Genotype and carrier status of the VNTR in Intron 2 of SLC6A4 was also examined in relation to PTD risk, with no significant associations (Table I). All genetic variants were examined as modulators of depression severity (PHQ-9 scores) with no significant findings.

Figure 3.

5-HTTLPR S-carriers in PTD risk across recovery. (a, b) Evaluation of the S-allele carrier frequencies in only subjects with no history of mood disorders shows S-carriers are less likely to experience PTD at 6 months compared to non-carriers (*p = 0.035, a), with no significant effect of the S-allele at 12 months (p = 1.00, b). (c) There was also a significant association for S-carrier status in prediction of PTD sub-type (p = 0.013), such that S-carriers, compared to non-carriers, were less likely to experience PTD in any form (*p = 0.020). This difference was primarily related to less S-carriers experiencing transient PTD (**p = 0.004).

Figure 4.

5-HTTLPR LG-carriers in PTD risk across recovery. (a, b) Evaluation of the LG-allele carrier frequencies in only subjects with no history of mood disorders shows LG-carriers status was not associated with PTD risk at 6 months (a, p = 0.345, χ²= 6.207, n = 91). However, LG-carriers were less likely to experience PTD at 12 months compared to non-carriers (*p = 0.030, b). (c) There was also a significant association for LG-carrier status in prediction of PTD sub-type (p = 0.011), such that LG-carriers, compared to non-carriers, were more likely to experience transient PTD (*p = 0.005) compared to non-carriers.

Temporal analysis of PTD showed unique genetic associations in subjects with no history of pre-morbid mood disorders. S-carrier status predicted PTD sub-type (Fisher’s exact, χ²= 10.496, p = 0.013, Figure 3(c)), such that S-carriers were less likely to experience PTD in any form (compared to L-homozygotes, p = 0.020). More specifically, S-carriers were less likely to experience a transient depression (compared to L-homozygotes, p = 0.004). There was also a significant association for L_G-carrier status in prediction of PTD sub-type (Fisher’s exact, χ²= 12.856, p = 0.011) at 12 months, such that L_G-carriers were more likely to experience transient PTD compared to non-carriers (p = 0.005) (Figure 4(c)). Temporal analysis of the VNTR in Intron 2 did not show any significant predictors of PTD sub-type.

Multivariate analysis at 6 months, controlling for anti-depressant use, gender, injury severity and education, showed a robust association with 5-HTTLPR such that L-homozygotes were 2.803 (95% CI = 1.032–7.614, p = 0.043) times more likely to be depressed compared to S-carriers. Backwards elimination of variables in this model showed that only S-carrier status and anti-depressant use predicted PTD status at 6 months (Table IIIc/d).

Discussion

This study presents evidence for a temporally-influenced, injury-specific genetic risk profile for depression following TBI. While pre-morbid mood disorder history was the most influential predictor of PTD, after controlling for this variable, L-homozygotes were at the next greatest risk for PTD 6 months post-injury. This finding suggests a protective role for the S-allele, which was further confirmed in multivariate analysis that included only those without pre-morbid mood disorders. Interestingly, the presence of the L_G variant (compared to L_G absence) was associated with lower PTD risk at 12 months post-TBI. These PTD genetic risk profiles differ from findings reported with MDD and are dependent on time after injury, suggesting a TBI-specific genetic risk relationship. These unique associations with SLC6A4 in PTD call for further investigation into the role of the serotonergic system post-TBI, specifically in earlier recovery periods where intervention may be highly beneficial.

While this study showed a novel, protective association for S-carriers in PTD risk at 6 months post-TBI, particularly for subjects with no history of pre-injury mood disorders, temporal PTD analysis also supported these results by showing that S-carriers were specifically less likely to experience a transient depression, with no difference in other PTD-sub-types. These findings also show a protective association for the L_G-allele at 12 months post-injury in these subjects, with L_G-carriers primarily experiencing either no or transient PTD.

These findings differ from 5-HTTLPR associations in MDD risk, where the S- and L_G-allele are most consistently identified as risk genotypes [11], implying a unique injury-induced neuropathology in PTD. Temporal PTD-sub-type analysis of both S-carrier and L_G-carrier PTD associations show these findings are also temporally-specific during TBI recovery. Interestingly, this study found no PTD associations with the VNTR in Intron 2 of SLC6A4. However, it has been reported previously that the 5-HTTLPR has a dominant role over 5-HTT expression, with the VNTR acting only as a modulator [33]. One other study has examined S-allele 5-HTTLPR associations with PTD 12 months post-injury with no significant results [24], findings which are consistent with the results in this study. While our report found the L_G-allele was associated with no PTD at 12 months, this previous study did not examine the L_G-allele separate from the S-allele. Also, subjects in that study were not stratified by pre-morbid mood disorders; Our study only showed genetic associations with PTD when pre-morbid subjects were excluded or whose effects were adjusted for in multivariate analysis (Table III). Similarly, studies in post-stroke depression suggest S-homozygotes show increased rates of depression, but many of these studies did not exclude or separately analyse subjects with a history of pre-injury depression [34].

The results of this study, with 5-HTTLPR genotype PTD risk associations that differ from MDD, suggest that depressive symptomology in TBI may arise from a unique, TBI-specific pathology. In MDD, several studies show associations between S-carriers and increased risk for MDD (particularly in relation to stressful life events [11]), yet these findings are inconsistent, and effect sizes are small [12]. Other studies show that the L-allele is associated with completed suicide, nicotine dependence, and attention deficit hyperactivity disorder [35]. While these studies all continue to suggest a role for 5-HTTLPR in psychopathology, they remain controversial, in part, due to a lack of a clear underlying mechanism for these associations. Some work suggests that cortical 5-HT levels vary by 5-HTTLPR genotype when considering both in vitro [36, 37] and in vivo [38, 39] binding studies. However, other conflicting studies exist [40–42]. More consistently, S-carriers show increased emotionality with higher amygdala reactivity, as well as a general hypervigilance towards environmental stimuli and some advantages in cognitive functioning (see Homberg and Lesch [14] review). While this behavioural adaptation could be advantageous in some aspects, it may also pre-dispose S-carriers to mood disorders [12–14]. Recently, Ho et al. [43] showed that SERT availability was reduced in subjects with major depression, but that this was not associated with SLC6A4 genotype. Similarly, Kobiella et al. [44] showed recently that 5-HTT binding potential did not mediate the relationship between 5-HTTLPR and amygdala reactivity, suggesting that 5-HTTLPR may not have a direct effect on 5-HTT availability. Kobiella et al. [44] suggest that 5-HTTLPR effects on neurodevelopment, and not 5-HT levels, may confer genotype differences in adult cognition and MDD risk.

The effects of 5-HTTLPR, either through its impact on 5-HT levels, 5-HTT functionality or relationships with cognition and emotionality, are likely to interplay with TBI pathology to influence PTD development. In a highly stressed system like the injured brain, it is possible 5-HTTLPR may more greatly impact cortical 5-HT levels or 5-HTT availability. TBI induces a global hypo-neurotransmission state (this is especially evident in the dopaminergic system, for a review see Bales et al. [19]) and in experimental TBI, there is evidence of a chronic hypo-serotonergic state [45]. Thus, a protective S-allele, with possibly lower 5-HT re-uptake, may indicate a less severe hypo-serotonergic state post-TBI in S-carriers compared to L-homozygotes. However, it is also important to note that 5-HTTLPR can be influenced by epigenetic regulation, particularly since TBI induces a global hypo-methylation state [46]. Interestingly, the human 5-HTTLPR S-allele is associated with increased methylation compared to the L-allele [47]; methylation at 5-HTTLPR has also been associated with increased depressive effects of a stressor [48].

The effect of 5-HTTLPR on morphology of the fronto-limbic system may aid in understanding unique PTD associations, and these circuits show dysfunction in both TBI and depression [5]. In the neurodegenerative state of TBI [49], hippocampal atrophy frequently occurs [50]. Hippocampal atrophy is also a common finding in MDD that is reportedly more severe in depressed L-homozygotes [51, 52], suggesting a potential ‘advantage’ for S-carriers, especially in reference to PTD. Morphological differences in the fronto-limbic system may also explain increased emotionality (increased amygdala reactivity) and cognition (better decision-making, risk aversion) of S-carriers. As depression influences cognitive recovery post-TBI [8], genetic influences on cognitive recovery may impact PTD development. As attentional deficits are a hallmark of TBI, an increased hypervigilant state pre-injury (attention to emotional/environmental stimuli) in S-carriers could be beneficial post-injury; similarly, studies show persons with ADHD tend to be L-carriers [53]. Better cognitive function at early time-points (6 months) post-injury may be protective against depressive symptomology. Future studies examining the effect of 5-HTTLPR on post-TBI cognitive recovery in relation to fronto-limbic function may aid in understanding how this polymorphism influences PTD risk.

Increased PTD incidence among subjects with pre-morbid mood disorders, regardless of 5-HTTLPR status, is consistent with other studies that have reported pre-morbid status as a significant predictor in PTD development [54]. This study also found that pre-morbid mood disorder history was significantly associated with a persistent PTD sub-type. Other studies have shown that subjects with pre-morbid mood disorders were more likely to develop PTD at earlier time points compared to subjects with no history of mood disorders [3]. As this study did not investigate depression status specifically at the time of injury, it is possible PTD at early time-points could be a continuation of depression at injury. However, Bombardier et al. [1] reported similar rates and onset of depression in subjects with depression at injury compared to subjects with a pre-injury history for depression that were not depressed at the time of the injury. Neuropathology associated with pre-morbid mood disorders (decreased neurogenesis in the hippocampus [55], altered neuroplasticity and functionality in fronto-limbic circuits, etc.) may overlap and/or interfere with normal recovery post-TBI, resulting in increased risk for PTD. Furthermore, S-carriers with pre-morbid mood disorders were at greater risk for PTD compared to S-carriers with no history of mood disorders. It is possible that some of the suggested advantages of the S-allele in protection against PTD may be ameliorated by pre-injury neuropathology associated with pre-morbid mood disorders. Although there were no significant genetic predictors of pre-morbid status, subjects with pre-morbid mood disorders did tend to be S-carriers (Table II), consistent with MDD literature.

In this study, the L_G- and S- alleles, believed to result in similar 5HTT expression levels based on other studies, have different temporal associations with PTD risk, particularly for those without pre-morbid mood disorders. It is possible that the L_G-allele’s protective association has similar possible mechanisms as previously mentioned for the S-allele. Yet, it is not clear why these associations are temporally specific. While the L_G association needs to be replicated in a larger sample, a temporal difference might be explained in several ways. One possibility may be due to the regulation of the L_G allele by the transcription factor AP-2, as the L_G allele has an additional AP-2 binding site compared to the L_A allele (also, the S-allele does not have this site) [56]. Importantly, there is evidence to suggest that AP-2 function might highly impact injury × gene interactions [57]. It is also not clear how allele-specific methylation might affect these associations.

While this study contains several novel and neurobiologically relevant findings, there are some caveats to consider. This study was limited by population size, and future analyses will need to include larger samples. Also, this study did not exclude subjects that were on anti-depressant treatment. Patients with TBI are given SSRIs and other anti-depressants frequently for a range of issues common to TBI, including problems associated with sleep, agitation and behaviour, in addition to treatment for a depressive episode [26]. Subjects taking anti-depressants were more likely to be depressed in this study, suggesting that, although subjects were taking these medications, they were still experiencing significant depressive symptoms, a finding similar to previous reports of reduced effectiveness of SSRIs in populations with TBI [25]. Also, serotonergic genes may modulate antidepressant efficacy in individuals with TBI as well as rates of adverse events, as one study of a population with PTD showed S-homozygotes (compared to L-carriers) experienced less adverse effects of anti-depressants [25]. As the effect of 5-HTTLPR on pharmacotherapy has also been suggested in non-injured populations [35], future studies are needed to examine pharmacogenetic effects of anti-depressant treatment in PTD. Another caveat to consider is how gender effects may moderate 5-HTTLPR risk profiles in PTD. This study does not report significant gender differences in PTD development nor in genetic risk profiles, but is likely under-powered for this analysis. Women are more at risk for MDD [58], yet TBI is more prevalent in men [59]. Thus, sex will be an important factor to consider in future and larger studies of genetic risk associations with PTD.

As serotonin signalling influences many mechanisms that impact TBI recovery (e.g. neurogenesis, regeneration and synaptic plasticity) [60], this study also highlights the need to understand serotonergic functioning post-TBI. While genetic associations and interactions in a specialized population like PTD are highly intriguing, understanding how these mechanisms function across the spectrum of depression and cognitive recovery post-TBI may also aid in understanding brain repair mechanisms in general.

Conclusion

With the increased risk for depression following TBI and the paucity of effective treatments, there is an immense need for a better understanding of underlying pathology of PTD. The results of this study, with 5-HTTLPR genotype PTD risk associations that differ from MDD and interact with pre-morbid mood disorder status, suggest that depressive symptomology in TBI likely arises from a TBI-specific pathology. This study also serves as evidence of important gene × injury interactions that may differ from previously reported associations in non-injured populations and may significantly impact recovery and treatment course for those with TBI.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Sources of Support: CDC: R49 CCR323155; DOD: W81XWH-071-0701; NIH R01NR008424, and R01 HD048162.