Missense Mutation of Brain Derived Neurotrophic Factor (BDNF) Alters Neurocognitive Performance in Patients with Mild Traumatic Brain Injury: A Longitudinal Study

The predictability of neurocognitive outcomes in patients with traumatic brain injury is not straightforward. The extent and nature of recovery in patients with mild traumatic brain injury (mTBI) are usually heterogeneous and not substantially explained by the commonly known demographic and injury-related prognostic factors despite having sustained similar injuries or injury severity. Hence, this study evaluated the effects and association of the Brain Derived Neurotrophic Factor (BDNF) missense mutations in relation to neurocognitive performance among patients with mTBI. 48 patients with mTBI were prospectively recruited and MRI scans of the brain were performed within an average 10.1 (SD 4.2) hours post trauma with assessment of their neuropsychological performance post full Glasgow Coma Scale (GCS) recovery. Neurocognitive assessments were repeated again at 6 months follow-up. The paired t-test, Cohen’s d effect size and repeated measure ANOVA were performed to delineate statistically significant differences between the groups [wildtype G allele (Val homozygotes) vs. minor A allele (Met carriers)] and their neuropsychological performance across the time point (T1 = baseline/ admission vs. T2 = 6th month follow-up). Minor A allele carriers in this study generally performed more poorly on neuropsychological testing in comparison wildtype G allele group at both time points. Significant mean differences were observed among the wildtype group in the domains of memory (M = -11.44, SD = 10.0, p = .01, d = 1.22), executive function (M = -11.56, SD = 11.7, p = .02, d = 1.05) and overall performance (M = -6.89 SD = 5.3, p = .00, d = 1.39), while the minor A allele carriers showed significant mean differences in the domains of attention (M = -11.0, SD = 13.1, p = .00, d = .86) and overall cognitive performance (M = -5.25, SD = 8.1, p = .01, d = .66).The minor A allele carriers in comparison to the wildtype G allele group, showed considerably lower scores at admission and remained impaired in most domains across the timepoints, although delayed signs of recovery were noted to be significant in the domains attention and overall cognition. In conclusion, the current study has demonstrated the role of the BDNF rs6265 Val66Met polymorphism in influencing specific neurocognitive outcomes in patients with mTBI. Findings were more detrimentally profound among Met allele carriers.


Introduction
Mild traumatic brain injury (mTBI) due to road traffic accident (RTA) is one of the most common forms of head injury, afflicting millions of people worldwide [1][2][3]. The complex pathophysiology of mTBI and the biochemical responses that occurs thereafter frequently results in cognitive, affective or behavioral deficits [4][5][6]. A wide variety of complaints and symptoms have been reported [7][8][9][10][11]. The predictability of these deficits are however not straightforward [12]. Crawford et al (2002) and Pruthi et al (2010) in their respective studies noted that the extent and nature of recovery in patients with mTBI are usually heterogeneous and not substantially explained by the commonly known demographic and injury-related prognostic factors [5,13], despite having sustained similar injuries or injury severity [12,[14][15].
Six missense mutations in BDNF [82][83][84], namely the rs6265, rs1048218, rs1048220, rs1048221, rs8192466 and rs139352447 were selected. The rs6265 variant has been well studied for its involvement in modulating recovery from brain injury but has yet to be investigated in the Malaysian population. The rs1048220 and rs1048221 are within the crucial protease cleavage site for proBDNF and are reported to impair proBDNF cleavage; and rs1048220 and rs104218 have been associated with Alzheimer's disease [85][86][87][88][89]. However, none of these missense mutations have been explored in brain injury with the exception of rs6265 (BDNF Val66-Met) [18,20,32,40,46,58,66,75]. Hence, the objective of our study was to assess the effects and association of variations within BDNF in relation to neurocognitive performance among patients with mTBI.

Materials and Methods
A total of 61 patients with mTBI who presented to the Emergency Department of University Malaya Medical Center, Kuala Lumpur between April 1 st , 2013 and August 31 st , 2014 were recruited prospectively. We defined mTBI as an acute head injury, consisting of non-penetrating head trauma resulting in one or more of the following: confusion/disorientation; loss of consciousness (LOC) less than 30 minutes; posttraumatic amnesia (PTA-less than 24 hours in duration); transient focal neurological signs or seizures; and Glasgow Coma Scale of 13 to 15 upon acute clinical evaluation. These patients were assessed with baseline computed tomography (CT) scans of the brain in the emergency department using a Siemens Somatom Sensation 16 CT scanner (Siemens AG, Berlin, Germany). A neuroradiologist (NR) and a neurosurgeon (VN) who were blinded to the clinical diagnosis independently evaluated the images for each patient. Patients who met the study criteria were admitted to the observation ward for 24 hours. Informed consent was obtained upon explaining the objectives of the study and as well as the research protocols/procedures as per the approved guidelines of our local Ethics Committee for the study (UM/EC/949.15). Thirteen patients were later dropped from this study as some refused screening of their genetic profiles, while others were later lost to follow-up, leaving the final sample of 48 patients with their DNA analyzed for genotyping.
Genotyping DNA was obtained from leukocytes using the phenol-chloroform extraction method [90]. Details of the six SNPs that were examined in this study are in Fig 1. The SNPs were genotyped using Taqman1 allelic discrimination assays and genotyping was carried out on a 7500 Fast Real-Time PCR machine (Applied Biosystems) using standard protocols as recommended by the manufacturer. Genotypes were confirmed by polymerase chain reaction (PCR) and Sanger sequencing in a random subset of individuals to determine the error rate for each of the Taqman SNP assays (see Fig 2 for the primer sequences).

Neurocognitive assessment
The screening module of Neuropsychological Assessment Battery (S-NAB Form 1) was used to assess the neurocognitive performance of the patients by a clinical neuropsychologist (VV). The assessments were done once the patient had recovered to a GCS score of 15 and was not under any trauma related physical or emotional distress. The S-NAB comprises a comprehensive set of neuropsychological tests (refer to Fig 3), with demographically corrected norms for adults between the ages of 18 to 97 years. Five cognitive domains i.e attention, memory, language, visuospatial and executive functions are evaluated through this battery. This battery consists of 12 individual tests across the five domains aforementioned. A total of 16 T scores are then derived, 14 of which contribute toward five separate Screening Index (domainspecific) scores and one Total Screening Index score [91][92]. The S-NAB Form 2 was used to  1 Reference minor allele frequency as reported in http://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs=6265. 2 Reference minor allele frequency as reported in http://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs=1048218. 3 Reference minor allele frequency as reported in http://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs=1048220. 4 Reference minor allele frequency as reported in http://www.ncbi.nlm.nih. gov/projects/SNP/snp_ref.cgi?rs=1048221. 5 Reference minor allele frequency as reported in http://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs= 8192466. 6 Reference minor allele frequency as reported in http://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs=139352447.  repeat the same subtests in the screening module at 6 months by the same neuropsychologist to assess the neurocognitive performance longitudinally.

Statistical analysis
All data management and analyses were performed using the SPSS statistical software (Version 22.0). Independent t-test was used to establish the differences in demographic features of the sample, if any, based on their BDNF SNPs and allele status. The mean differences of the standard score (SS) across the time points against the allele carrier status [wildtype G allele (Val homozygotes) vs. minor A allele (Met carriers)] were then analyzed using the paired t-test for both categories. The Cohen's d effect size (ES) was also used to measure the magnitude of the differences and a comparison of the ES was done. Repeated measure ANOVA was then performed to delineate statistically significant differences between the groups and their neuropsychological performance across the time point (T 0 = baseline/ admission vs. T 1 = 6 th month follow-up), The Bonferroni post hoc correction for both multiple comparison and confidence interval adjustment were administered. Any statistically significant (p< 0.05) major effects and interaction were then noted. To assess the association between the allele carrier status and neurocognitive performance, the Spearman correlation coefficient test was also used.

Demographic Characteristics
The demographic characteristics of the study patients are presented in Fig 4. The study patients were predominantly young males (87.5%), within the age range of 18 to 53 (75.0%) with a mean age of 27.4 (SD 8.9). These patients had an average of 11.4 (SD 2.0) years of formal education. The baseline (T 1 ) neuropsychological assessment was conducted after the full GCS recovery of the patients with an average turnaround time of 4.8 hours (SD 7.9) post trauma, while the repeat neuropsychological assessment was done at an average of 6.1 (SD 0.1) months. In order to look at clinically meaningful markers influenced by specific genotypes, we stratified the group according to the SNPs (involving only rs6265 and rs1048218 as the rest of the SNPs were monomorphic as discussed below) and their allele status. No statistically significant differences were observed within the groups except in the presence of LOC (t = -2.026, df = 46, p = 0.049), with a higher incidence among the A minor allele carriers.

Genetic results and correlation with neurocognitive performance
Genotype distribution and minor allele frequency. Six BDNF mutations were examined, of which four (rs1048220, rs1048221, rs8192466, rs139352447) were found to be monomorphic. Only rs6265 and rs1048218 were found to be polymorphic in our population (refer to Fig  1). Both the controls and patients conformed to the Hardy-Weinberg equilibrium for rs6265 and rs1048218. The minor allele frequency for rs6265 was 46.7% in controls compared to the patients (43.6%), but this was not significantly different (p = .380). The high MAF values is comparable to what has been reported previously [93][94][95][96][97] and as annotated for East Asians in the HapMap (41.8%) and 1000genomes (48.8%).
The rs1048218 variant had a low MAF in our population (2% in controls) that is also similar to what has been reported previously and in HapMap and 1000genomes. As the variant was present at a similar frequency in both the patients and controls, no further correlation analysis was performed with this variant.
BDNF rs6265 vs. neurocognitive performance. Individuals with the A minor allele (corresponding to Met carriers-Met homozygotes/ Met heterozygotes) generally performed more poorly on neuropsychological testing in comparison to those with the wildtype G allele (corresponding to the Val homozygotes) at both time points.  No statistically significant associations were observed across the neurocognitive domains and specific allele status with the exception of the memory SS score at 6 months (r = -0.412, p = 0.05). The memory scores of patients with the A allele were observed to be significantly lower at six months follow-up.

Discussion
We explored the prevalence and possible association of six BDNF mutations with specific neurocognitive functions in patients with mTBI over time. We observed a possible protective effect of the G allele in rs6265, with better performance in the domains of attention, executive function, memory, and overall cognition among patients with mTBI. The "finer" performance by those patients with the wildtype G allele in both neurocognitive and neurobehavioral measures, have been consistently reported by other studies involving other CNS pathologies as well [32,88,[98][99].
For example, McAllister et al (2012) demonstrated that patients with mTBI reported better performance on measures of processing speed longitudinally among the G allele homozygotes as opposed to those who were homozygote for the A allele [32]. Similarly, Chao, Kao and Porton (2008) reported that the age of schizophrenia onset in a group of non-related African American patients (n = 42) was significantly later in G allele homozygotes [88]. Moreover, Perkovic et al (2014) showed that patients with schizophrenia who were G allele homozygotes were better responders to pharmacotherapeutic treatment and demonstrated significant improvement of clinical symptoms, including lesser conceptual disorganization and delusions [99].
While some studies have reported increased neurocognitive vulnerability in G allele homozygotes [69,[100][101][102], we believe that the superior neurocognitive performance we observed in our study may be due to the possibility that the G homozygous allele is associated with increased activity dependent secretion of BDNF, increased synaptic plasticity, and better hippocampus dependent memory and cognitive performance [99]. These mechanistic processes have been well explicated in the works of Egan et al (2003) and Kauppi et al (2013) [29,72]. The divergent influence of haplotype specific variants cannot be overlooked as well.
On the other hand, patients with the A allele in our study revealed mostly non-significant trends of impaired neurocognitive performance with some interactions seen across the time points in the domains of executive function and overall cognition. Additionally, the memory domain saw statistically significant interaction over time and was also negatively associated with allele status, with evidence of regressing memory function at 6 months among the patients with the G allele. Kauppi et al (2013) noted that this preferential effect on memory is in line with the role of BDNF in the molecular processes underlying memory acquisition. Mechanistically, hippocampal or para-hippocampal BDNF is secreted during neural activation [72]. The increased postsynaptic level of BDNF is known to influence the formation of new synapses in the late phase long term potentiation (LTP), a process that is crucial for the acquisition and storage of long term memories. A successful completion of this crucial molecular process is however impeded when the activity dependent secretion and intracellular trafficking of BDNF is reduced in the protein [29,72].
Taken together, findings of the current study are the following: (1) only two non-synonymous alterations of the amino acid were present in our study population (rs6265/Val66Met and rs 1048218) and the rest of the remaining variants were monomorphic in nature; (2) mTBI patients with BDNF rs6265 Val homozygous allele showed significant differences in their neurocognitive performance and were more likely to perform better than the Met carriers in the domains of attention, memory, executive function and overall performance, both acutely and over time; (3) the Met allele carriers of BDNF rs6265 had considerably low standard scores in most neurocognitive domains observed longitudinally; (4) there was a significant main effect of the time points, and the influence of specific allele status on neurocognitive performance observed; and (5) longitudinal change in memory performance with evidence of deteriorating performance among the A minor allele group (Met carriers) was observed. Strengths of the study include a relatively well characterized homogeneous group of patients in terms of injury type or severity, a short time frame from the time of injury to neurocognitive testing, detection of early neuropsychological deficits in the acute stage, and a consistent reassessment interval at 6 months post trauma. However, there are certain limitations in our study that are worth noting. First, the method of dichotomizing the patients' allele carrier status category (wildtype G allele vs. A minor allele, both the homozygous and heterozygotes) may have unequally diminished the dual allele effect of the A minor allele homozygous vs. heterozygotes (Met/Met vs. Val/Met) on the neurocognitive performance. Additionally, the sample size representing each arms of the rs6265 polymorphism was rather small and should be increased in future longitudinal studies.

Conclusion
In conclusion, the current study has demonstrated the role of the BDNF rs6265 Val66Met polymorphism in influencing specific neurocognitive outcomes in patients with mTBI. Findings were more detrimentally profound among Met allele carriers. Our results strongly suggest that the role of the Val66Met polymorphism in influencing neurostructural alterations and cognitive and behavioral changes post-mTBI should be further explored. Such investigation in future studies may have significant influence over the ways in which mTBI patients are currently managed and their outcomes predicted.