OUP user menu

Six-month recovery from mild to moderate Traumatic Brain Injury: the role of APOE-ε4 allele

Laury Chamelian, Marciano Reis, Anthony Feinstein
DOI: http://dx.doi.org/10.1093/brain/awh296 2621-2628 First published online: 20 October 2004


The possession of at least one APOE-ε4 allele may be linked to poor outcome in patients with predominantly severe traumatic brain injury (TBI). In mild TBI, which accounts for approximately 85% of all cases, the role of the APOE-ε4 allele is less clear. Studies completed to date have relied on brief cognitive assessments or coarse measures of global functioning, thereby limiting their conclusions. Our study investigated the influence of the APOE-ε4 allele in a prospective sample of 90 adults with mild to moderate TBI in whom neuropsychiatric outcome 6 months after injury was assessed as follows: (i) a detailed neuropsychological battery; (ii) an index of emotional distress (General Health Questionnaire); (iii) a diagnosis of major depression (Structured Clinical Interview for DSM-IV); (iv) a measure of global functioning (Glasgow Outcome Scale); (v) an index of psychosocial outcome (Rivermead Head Injury Follow-up Questionnaire); and (vi) symptoms of persistent post-concussion disorder (Rivermead Post-Concussion Symptoms Questionnaire). No association was found between the presence of the APOE-ε4 allele and poor outcome across all measures. Given the homogeneous nature of our sample (mild to moderate injury severity), the uniform follow-up period (6 months) and the comprehensive markers of recovery used, our data suggest that the APOE-ε4 allele does not adversely impact outcome in this group of TBI patients.

  • APOE-ε4 allele
  • cognitive testing
  • mood
  • psychosocial outcome
  • head injury
  • GCS = Glasgow Coma Scale
  • GOS = Glasgow Outcome Scale
  • LOC = loss of consciousness
  • MANOVA = multivariate analysis of variance
  • PTA = post-traumatic amnesia
  • SRT = Simple Reaction Time
  • TBI = traumatic brain injury


It has been suggested (Sorbi et al., 1995; Jordan et al., 1997; Teasdale et al., 1997; Friedman et al., 1999; Kerr et al., 1999; Crawford et al., 2002; Chiang et al., 2003) that outcome following traumatic brain injury (TBI) is influenced by polymorphism of the apolipoprotein E (APOE) gene, located on chromosome 19. Of the three common alleles (ε2, ε3, ε4), APOE-ε4 allele has been the one associated with unfavourable cognitive (Friedman et al., 1999; Crawford et al., 2002) and functional recovery (Teasdale et al., 1997; Lichtman et al., 2000; Chiang et al., 2003), deposition of β-amyloid following head injury (Roberts et al., 1994; Nicoll et al., 1995), prolonged posttraumatic coma (Sorbi et al., 1995; Friedman et al., 1999), lower cerebral blood flow during the first 24 h after injury (Kerr et al., 1999) and greater neurological deficits in boxers with history of 12 or more professional bouts (Jordan et al., 1997). It has also been shown to act synergistically (Mayeux et al., 1995; Tang et al., 1996) and additively (Katzman et al., 1996) with a previous TBI as risk factors for Alzheimer's disease, although recent studies have failed to support these findings (O'Meara et al., 1997; Weiner et al., 1999; Guo et al., 2000; Plassman et al., 2000; Jellinger et al., 2001).

While informative, these studies have mostly focused on subjects with severe head injuries (Roberts et al., 1994; Nicoll et al., 1995; Sorbi et al., 1995; Friedman et al., 1999; Kerr et al., 1999; Lichtman et al., 2000; Crawford et al., 2002). With respect to mildly brain-injured individuals, who account for almost 85% of all TBI cases, the role of the APOE-ε4 allele is less clear. In addition, outcome measures have often been limited, relying on brief cognitive assessments (Jordan et al., 1997; Crawford et al., 2002) or on disability scales (Teasdale et al., 1997; Chiang et al., 2003) (such as the Glasgow Outcome Scale) (Jennett and Bond, 1975) that lack the necessary details in providing a thorough depiction of various aspects of recovery following head injury. Although a recent study (Liberman et al., 2002) with predominantly mild TBI patients recorded no significant relationship between APOE-ε4 allele status and a limited number of cognitive tasks 6 weeks following head injury, it did not assess psychosocial functioning. Our study investigated the influence of the APOE-ε4 allele on multiple measures of neuropsychiatric recovery in mild to moderate TBI at a follow-up period extended to 6 months after injury.


Subjects were recruited prospectively from a traumatic brain injury clinic at a tertiary care hospital. All patients who have sustained a TBI and seen in the hospital's emergency room are routinely given a follow-up clinic appointment within weeks of injury and thereafter followed for a minimum of 6 months. A consecutive sample of 90 clinic attendees was enrolled in our study at the time of their first clinic assessment. Subjects were between 18 and 60 years of age and had sustained a non-penetrating mild (Esselman and Uomoto, 1995) [Glasgow Coma Scale (GCS) = 13–15; loss of consciousness (LOC) <20 min; post-traumatic amnesia (PTA) <24 h] or moderate TBI [GCS = 9–12; PTA >24 h but less than 1 week]. All participants underwent a thorough neuropsychiatric evaluation, including detailed cognitive testing 6 months after head injury, at which time a buccal smear was collected to determine APOE genotype. The study sample was split into those with (n = 19) and without (n = 71) the APOE-ε4 allele. These two groups were then compared on the neuropsychiatric measures outlined below, which were performed without prior knowledge of the patients' APOE status.

Background information

The demographic and TBI-related data collected included age, gender, race, marital and pre-injury employment status, level of education, type of occupation, history of alcohol and substance abuse, past psychiatric history, prior head injury, family history of psychiatric illnesses or dementia/Alzheimer's disease, mechanism of injury, head injury severity indices, such as GCS recorded at the emergency room (Levin et al., 1987), LOC, PTA (Russell and Smith, 1961), initial CT brain results. In addition, all subjects were assessed with the Abbreviated Injury Severity Score (AISS) (Civil and Schwab, 1988), which provides a measure of trauma severity to various body regions, including the head. The presence of physical pain and medication use were also recorded.

APOE genotyping

DNA was extracted from buccal epithelial cells using the Qiagen Blood Mini Kit, and amplified by PCR with primers specific for the APOE alleles ε2, ε3 and ε4: 5′-TCC AAG GAG CTG CAG GCG GCG CA-3′ and 5′-ACA GAA TTC GCC CCG GCC TGG TAC ACT GCC A-3′. Cycling conditions were as follows: 94°C for 4 min, 35 cycles of 94°C for 30 s, 66°C for 30 s, and 70°C for 1:30 min, with final extension at 70°C for 10 min. The amplimers were digested with the HhaI restriction endonuclease for 2 h and then electrophoresed on a 4% high-resolution agarose gel.

Neuropsychiatric evaluation

Glasgow Outcome Scale (GOS) (Jennett and Bond, 1975)

This widely used clinician-rated five-point scale assesses global adjustment to activities of daily living and general outcome. A score of 5 indicates a return to the premorbid level of functioning whereas lower scores denote a poor global outcome.

Rivermead Head Injury Follow-up Questionnaire (RHFUQ) (Crawford et al., 1996)

This is a five-point self-report scale with a total score ranging from 0 to 48. It addresses 10 aspects of a patient's functioning (relationships, domestic and vocational activities, ability to participate in a conversation) following TBI, and hence provides a more detailed description of psychosocial functioning than the GOS. High scores on the RHFUQ are indicative of poor recovery.

Rivermead Post-Concussion Symptoms Questionnaire (RPQ) (King et al., 1995)

This is a five-point self-report scale measuring the presence and the severity of 17 somatic symptoms commonly experienced following head injury. High scores on the RPQ indicate a greater level of physical distress.

Twenty-eight-item General Health Questionnaire (GHQ) (Goldberg and Hiller, 1979)

This questionnaire assesses self-reported symptoms of emotional distress. It contains four subscales of seven questions each, pertaining to somatic complaints, anxiety, social dysfunction and depression. For each question, the answer is chosen among four possible responses that are scored in a binomial fashion (0–0–1–1). High scores on the GHQ indicate a greater level of psychological distress.

Mood disorder section of the Structured Clinical Interview for the DSM-IV (SCID for DSM-IV) (First et al., 1994)

This was used to establish a diagnosis of major depression. The clinic's neuropsychiatrist who interviewed the study participants was blind to the subjects' cognitive data and APOE genotype.

Resumption of work or studies

Patients were asked if they had resumed work or studies. Those who had not returned to work or studies because of injuries other than their TBI (n = 36) were excluded from this part of the analysis.

Cognitive battery

Wechsler Adult Intelligence Scale—III: working memory (Wechsler, 1997a). This measure of attention and working memory is a composite of the scores computed from the following subsets: Digit span, Arithmetic and Letter sequencing.

Wechsler Memory Scale—III: logical memory I and II (Wechsler, 1997b). This assesses verbal memory by examining the patient's ability to recall two orally presented stories immediately (I) and after a 30-min delay (II).

California Verbal Learning Test-II: total, long delay free recall and recognition hits (Delis et al., 2000). This provides an assessment of learning and memory of verbal material. The subject is presented with a 16-item shopping list over five free recall trials. We recorded the sum of recalled words from List A across the five trials and from delayed free-recall as well as from recognition trials.

Brief Visuospatial Memory Test—Revised: immediate and delay total recall (Benedict, 1997). With multiple trials, it is possible with this test to study the patient's visuospatial learning and memory. Testing involves the reproduction and subsequent delayed recall of a series of geometric designs.

Paced Auditory Serial Addition Task (Gronwall, 1977). This task assesses information processing speed and sustained and divided attention. The subject is required to add consecutive pairs of tape-recorded digits so that each number is added to the one immediately preceding it. Four series of digits are presented, each at an increasingly quick rate of number presentation: 2.4, 2.0, 1.6 and 1.2 s. The number of correct responses for each of the four series was recorded.

Controlled Oral Word Association Test (Spreen and Benton, 1969). This measures verbal association fluency, tapping into higher-level executive functioning abilities. Subjects are asked to generate as many words as possible beginning with a given letter (F, A or S). Three trials are administered, each trial lasting 60 s. Proper nouns, numbers and the same word with a different suffix are excluded. The sum of admissible words generated during the three trials was recorded.

Wisconsin Card Sorting Test (WCST): total and perseverative responses (Heaton et al., 1993). This provides measures of mental flexibility and problem solving. Subjects are required to sort cards according to specific categories (colour, shape, number) based on feedback from the examiner. The total number of categories achieved and the number of perseverative responses were recorded.

Simple reaction time (SRT) (Feinstein et al., 1992). This gives an index of basic psychomotor speed. The test comprised 60 trials for each hand. The imperative stimulus to which the subject had to react was the filling of a square either to the left (for the left hand) or the right (for the right hand) of a central blank square on the computer monitor. The subject reacted by pushing either the left or right button on a button box. The right-hand responses were completed before proceeding to the left-hand ones. Prior to the imperative stimulus, an arrow appeared in the central square pointing in the direction of the square to be filled. The arrow appeared 1.6, 0.8 or 0.2 s before the imperative stimulus, each for 25% of the time. For the remaining 25% the arrow appeared simultaneously with the imperative stimulus. The order of the interval was assigned randomly, to prevent the subject anticipating the exact occurrence of the stimulus. The interval between the end of one trial and the appearance of the arrow for the next trial was also assigned randomly between 1 and 4 s.

Choice reaction time (CRT) (Feinstein et al., 1992). The test comprised 80 trials. As in the SRT, the imperative stimulus to which the subjects had to react was a filling of a square either to the left or right of the central blank square. A mixture of warned and cued CRT trials was used. In the warned trials, a cross appeared in the central square prior to the imperative stimulus. This indicated that the stimulus was about to appear, but not which side. In the cued trials, the arrow appeared in the central square pointing in the direction of the square to be filled. The 80 trials were equally and randomly divided between warned and cued responses. Within each 40, half the responses were right and half were left. The timing for the cross or arrow to appear prior to the imperative stimulus was the same as for the SRT and was also assigned randomly to prevent anticipation.

Vocabulary subscale of the Wechsler Abbreviated Intelligence Scale (Wechsler, 1999). This was used to provide an estimate of premorbid intelligence quotient.

Statistical analysis

Patient groups with and without the APOE-ε4 allele were compared using t tests for continuous demographic/injury, psychosocial and cognitive variables and χ2 analyses for categorical variables. Fisher's exact test was reported when appropriate. A 1% level of significance was chosen to adjust for multiple comparisons. In addition, two separate multivariate analyses of variance (MANOVAs) were performed on the 6-month neuropsychiatric and cognitive outcome measures. For each of the MANOVAs, the maximum P value was set at 0.05 (two-tailed test), as is recommended when conducting multiple comparisons (Keppel, 1982).


Written consent was provided from all subjects. The study was approved by the Sunnybrook and WCH Research Ethics Board.


The mean age for the total study sample (N = 90) was 33 years (SD 12.6). Subjects were predominantly male (60%), Caucasian (76.7%) and had sustained a mild head injury (56.7%). The frequencies for the APOE-ε2, ε3 and ε4 alleles were 14, 71 and 15%, respectively, with the following genotypes: APOE 2/3 = 14 (15.5%); APOE 2/4 = 3 (3.3%); APOE 3/3 = 57 (63.3%); APOE 3/4 = 16 (17.8%). There were no homozygotes for APOE-ε2 and APOE-ε4 alleles.

Comparisons between those with and without the APOE-ε4 allele revealed no significant differences on demographic and injury-related variables (Table 1). In terms of global, physical and psychosocial functioning, both groups had similar outcomes, including returning to work or school 6 months after injury (Table 2). Cognitive function did not differ between the groups on all measures tested (Tables 3). In the two separate MANOVAs, no significant differences were apparent for neuropsychiatric [F(7,61) = 0.4; P = 0.9] and cognitive [F(20,56) = 0.6; P = 0.9] outcomes.

View this table:
Table 1

Demographic and injury-related characteristics compared between those with and without the APOE-ε4 allele

DemographicsAPOE-ε4 positive (n = 19)APOE-ε4 negative (n = 71)t test/χ2P values
Mean (SD*)Mean (SD)
Age (years)31.2 (13.3)34.1 (12.3)t(df,88) = 0.90.4
Gender (male)52.6%62.0%χ2(df,1) = 0.50.5
RaceFisher's exact test0.8
Marital status (single or divorced)57.9%63.4%χ2(df,1) = 0.20.7
Education (beyond high school)52.6%40.8%χ2(df,1) = 0.80.4
Employment (employed)52.6%78.6%χ2(df,1) = 5.10.1
Occupation (professional/semiprofessional)18.8%14.5%Fisher's exact test0.7
Past alcohol abuse47.4%38.0%χ2(df,1) = 0.50.5
Past substance abuse31.6%11.3%Fisher's exact test0.1
Prior TBI31.6%22.9%Fisher's exact test0.5
Past psychiatric history22.2%17.1%Fisher's exact test0.7
Family psychiatric historyχ2(df,2) = 3.90.1
    Injury-related characteristics
Mechanism of injury (MVA-related)73.7%63.4%χ2(df,1) = 0.70.4
Loss of consciousnessFisher's exact test0.2
    Dazed or LOC <20 min83.3%93.8%
    LOC >20 min16.7%6.2%
Post-traumatic amnesiaχ2(df,1) = 0.80.4
    <24 h47.4%59.2%
    >24 h and <1 week or sedated52.6%40.8%
Glasgow Coma Score at the emergency roomFisher's exact test0.7
CT scan abnormalities61.1%a36.2%bχ2(df,1) = 3.50.1
AISS12.6 (9.5)13.5 (10.3)t(df,72) = 0.30.8
Pain symptoms56.3%65.5%χ2(df,1) = 0.50.5
Medication intake56.3%45.8%χ2(df,1) = 0.50.5
  • a n = 18

  • b n = 58.

  • AD = Alzheimer's disease; MVA = motor vehicle accident; LOC = loss of consciousness; AISS = Abbreviated Injury Severity Score.

View this table:
Table 2

Neuropsychiatric 6-month outcomes compared between those with and without the APOE-ε4 allele

APOE-ε4-positive (n = 19)APOE-ε4-negative (n = 71)t test/χ2P values
Mean (SD*)Mean (SD)
GOS4.3 (0.5)4.3 (0.6)t(df,71) = −0.30.7
RHFUQ17.9 (14.0)18.7 (13.7)t(df,75) = 0.20.8
RPQ19.6 (18.2)24.6 (18.1)t(df,72) = 1.00.3
    Somatic1.9 (2.8)2.5 (2.6)t(df,76) = 0.80.4
    Anxiety2.2 (2.6)3.0 (2.7)t(df,76) = 1.10.2
    Social dysfunction2.1 (2.6)3.2 (2.8)t(df,74) = 1.40.2
    Depression0.7 (1.5)1.0 (1.9)t(df,74) = 0.50.6
    Total7.0 (9.0)9.6 (8.6)t(df,74) = 1.10.3
SCID depressed18.2%a11.8%bFisher's exact test0.6
Return to work/studiesχ2(df,1) = 0.90.3
  • a n = 11

  • b n = 51

  • c n = 13

  • d n = 41.

  • GOS = Glasgow Outcome Scale; RHFUQ = Rivermead Head Injury Follow-up Questionnaire; RPQ = Rivermead Post-Concussion Symptoms Questionnaire; GHQ = General Health Questionnaire; SCID = Structured Clinical Interview for DSM-IV.

View this table:
Table 3

Cognitive performances compared between those with and without the APOE-ε4 allele

APOE-ε4 positive (n = 19)APOE-ε4 negative (n = 71)t testP values
Mean (SD*)Mean (SD)
Time between injury and cognitive testing (days)208.8 (68.6)200.7 (53.3)t(df,88) = −0.50.6
WASI vocabulary (premorbid IQ)55.4 (10.2)53.5 (12.0)t(df,84) = −0.60.5
WAIS-III working memory29.3 (7.0)28.31 (8.1)t(df,84) = −0.50.6
WMS logical memory story I45.8 (15.8)43.7 (11.6)t(df,86) = −0.60.5
WMS logical memory story II30.5 (9.0)27.3 (8.8)t(df,85) = −1.40.2
CVLT-II total56.6 (10.3)53.9 (11.4)t(df,88) = −0.90.3
CVLT-II long delay free recall12.2 (3.5)11.3 (3.5)t(df,87) = −1.00.3
CVLT-II recognition hits14.4 (2.7)14.8 (1.7)t(df,22) = 0.70.5
BVMT-R total25.8 (6.5)22.9 (8.5)t(df,85) = −1.40.2
BVMT-R delay9.8 (2.4)9.2 (2.6)t(df,84) = −0.90.4
PASAT 2.4 s42.5 (10.0)38.3 (11.5)t(df,83) = −1.40.1
PASAT 2.0 s37.8 (8.9)34.6 (10.6)t(df,81) = −1.20.2
PASAT 1.6 s30.8 (7.5)27.2 (9.2)t(df,81) = −1.50.1
PASAT 1.2 s23.2 (6.6)21.1 (6.6)t(df,80) = −1.20.2
COWAT36.7 (13.2)35.2 (10.7)t(df,86) = −0.50.6
WCST total5.4 (1.4)4.9 (1.7)t(df,85) = −1.10.3
WCST perseverative responses13.7 (12.4)21.2 (20.9)t(df,85) = 1.50.1
SRT, mean, both hands (s)359.3 (145.1)1037.2 (5316.0)t(df,81) = 0.50.6
CRT, mean warned trials, both hands (s)444.4 (116.0)464.0 (181.5)t(df,80) = 0.40.8
CRT mean cued trials, both hands (s)382.2 (104.5)407.8 (176.1)t(df,80) = 0.60.6
CRT grand mean of cued and warned trials (s)413.3 (107.5)437.5 (180.4)t(df,80) = 0.50.6
  • WASI = Wechsler Abbreviated Intelligence Scale; WAIS = Wechsler Adult Intelligence Scale; IQ = intelligence quotient; WMS = Wechsler Memory Scale; CVLT = California Verbal Learning Test; BVMT = Brief Visuospatial Memory Test; PASAT = Paced Auditory Serial Addition Task; COWAT = Controlled Oral Word Association Test; WCST = Wisconsin Card Sorting Test; SRT = Simple Reaction Time; CRT = Choice Reaction Time.


We did not find an association between the presence of the APOE-ε4 allele and poor outcome across multiple behavioural domains 6 months following mild to moderate TBI. This finding is supported by the close group matching of patients with and without the APOE-ε4 allele with respect to demographic and injury-related variables that may influence recovery from mild to moderate TBI (Williams et al., 1990; van der Naalt, 2001). To date, our study is the first to address outcome in a homogeneous sample of TBI patients (mild to moderate injury severity) at a uniform follow-up period (6 months) with an extensive array of tests that incorporated validated and reliable indices of mood, behaviour and cognitive disturbance. With regard to the latter, a detailed neuropsychological battery covering different cognitive domains was employed.

Our data support and extend the results of a recent study (Liberman et al., 2002) that examined cognitive dysfunction 6 weeks following mild to moderate TBI. No relationship between cognitive performance and the APOE-ε4 allele was found. In this report, TBI severity was based on a single variable, namely the Glasgow Coma Scale, the cognitive battery was limited in scope, and no measures of mood and behaviour were included. Despite these limitations, the data provided some evidence that recovery from mild to moderate TBI in the subacute phase was not dependent on genetic markers, a conclusion that our data now extend to the 6-month watermark, with the added caveat that neither mood, psychological distress nor additional aspects of cognition appear to be linked to the presence of the APOE-ε4 allele. Our delineation of TBI severity based on a confluence of three variables, namely GCS, PTA and duration of LOC, adds weight to our findings. In addition, our sample provided a fair representation of patients with head injury, since, unlike the previous report, we included subjects with premorbid psychiatric history or with alcohol/drug use problems, making our results generalizable. In this regard, our APOE-ε4-positive group had a lower employment rate despite a higher education level and greater incidence of prior TBIs, past psychiatric illness, alcohol and substance abuse, major depression and brain CT scan abnormalities. Although none of these findings approached statistical significance, there may be theoretically a relationship between the possession of the APOE-ε4 allele and neurocognitive dysfunction, which in turn leaves patients at increased risk of injury. However, our method did not allow us to answer this question. To do so, it would have required us to include a third subject group composed of patients who were APOE-ε4-positive but who had never sustained a traumatic brain injury.

Indirect support for our findings comes from another source. In the MIRAGE study (Bachman et al., 2003), in which various possible aetiological factors for Alzheimer's disease were examined in 443 African Americans and 2336 Caucasian Americans, no significant interaction was found between the APOE-ε4 allele and a number of risk factors of poor outcome, of which TBI was one. The racial breakdown of the sample was necessary, given the higher prevalence rate of the APOE-ε4 allele in patients of African descent (Zekraoui et al., 1997; Corbo and Scacchi, 1999). In this regard, it is germane to note that the 15% occurrence rate for the APOE-ε4 allele in our sample was consistent with the predominantly Caucasian population that attends our hospital and from which our study sample was derived. A unique study (Nathoo et al., 2003) that focused exclusively on the Zulu tribe in South Africa once again found that, despite the high prevalence of the APOE-ε4 allele in 110 subjects, outcome following mostly mild to moderate TBI was not linked to APOE polymorphism. Language barriers may have precluded a detailed cognitive assessment in determining recovery, which was based exclusively on the GOS.

Our data refute the findings from three other studies (Jordan et al., 1997; Teasdale et al., 1997; Chiang et al., 2003) that reported an association between the presence of the APOE-ε4 allele and poor outcome following predominantly mild to moderate TBI. In two of these (Teasdale et al., 1997; Chiang et al., 2003), outcome was based on a single measure, the GOS, a five-point Likert scale that provides a coarse measure of functioning after a head trauma. Neither of these studies looked at mood or cognition, two indices that provide the most sensitive measure of recovery following brain injuries that are deemed either mild or moderate. In the third study (Jordan et al., 1997), the ill effects of repetitive mild TBI were investigated in 30 retired and active boxers. To assess outcome, the authors devised a 10-point Chronic Brain Injury Scale that incorporated three dimensions, namely motor, cognitive and behavioural. Boxers who were considered to have received ‘high exposure’, (i.e. those with more that 11 professional bouts) and found to be APOE-ε4-positive were the most impaired. However, the assessment of cognition was based on the Mini-Mental State Examination (MMSE) (Folstein et al., 1975), which lacks sensitivity in patients at the less severe end of the TBI spectrum. In addition, deficits on the Chronic Brain Injury Scale appear to be quantified based on clinical observations, for which standardized testing instruments were not used (except for the MMSE). Furthermore, the small sample size adds to the difficulty when it comes to data interpretation.

The best evidence linking the APOE-ε4 allele with poor outcome comes from patients who have sustained predominantly severe TBI, but once again some of the methodological limitations inherent in the mild to moderate group apply, particularly the absence of valid measures of mentation (Sorbi et al., 1995; Friedman et al., 1999; Kerr et al., 1999; Lichtman et al., 2000; Crawford et al., 2002). In two studies, both with small sample sizes, the emphasis was on psychological and neurosurgical indices of outcome, and here the data showed an association between the APOE-ε4 allele and prolonged duration of coma (Sorbi et al., 1995) and a reduction in cerebral blood flow during the first 24 h following trauma (Kerr et al., 1999). In the latter, the combination of an APOE-ε4 allele and reduced blood flow was linked to worse 3-month outcome, as assessed by the GOS and the Disability Rating scale. Of note, however, was that poor outcome was defined as ‘dead’ or ‘vegetative’ as per the GOS ratings, yet the category ‘severe disability’, which applied to most of the non-APOE-ε4 allele bearers, was not included in the negative outcome group. Consequently, this methodology may have led the authors to overestimate the impact of the APOE-ε4 allele on poor recovery.

In other studies with larger sample sizes and more comprehensive markers of outcome, it was unclear whether validated assessment procedures were used. An investigation (Friedman et al., 1999) of 69 patients with predominantly severe TBI found that patients with the APOE-ε4 allele were almost six times more likely to remain comatose for more than 7 days and were 14 times less likely to have a good overall functional outcome 6–8 months after TBI. This global outcome index was derived from a composite set of examinations: mobility, seizures, speech, mood and cognition. However, no mention was made in the protocol of how the latter two indices were assessed. In addition, the overall outcome was designated as excellent versus suboptimal based on an arbitrary cut-off point. The limitations of this study were voided by Lichtman et al. (2000), who used the Functional Independence Measures (FIM) to study the effect of the APOE-ε4 allele on recovery in a group of patients who had completed a course of acute in-patient rehabilitation. The FIM assesses a patient's functioning across six areas: self-care, sphincter control, mobility, locomotion, communication and social cognition. Although the APOE-ε4 allele was linked to lower scores on the motor subscale, no association was found with cognition. When more sensitive psychometric tests are used, the yield is better, with 6-month correlations reported between the APOE-ε4 allele and memory deficits, as per the California Verbal Learning Test (Crawford et al., 2002). However, this relationship did not extend to measures of executive functioning, and once again mood was not part of the assessment. This investigation, which also included subjects with mild to moderate head injury, neglected to control for depression when evaluating cognitive performances despite accumulating evidence in the literature suggesting a strong association between major depression and poor performance on cognitive testing following mild to moderate TBI (Barth et al., 1983; Bornstein et al., 1989; Levin et al., 2001; Fann et al., 2001).

In summary, our study, which made use of a well-matched control group and widely used, validated measures of mood, behaviour and cognition, failed to elucidate any genetic predisposition to adverse outcome 6 months after trauma. This finding is of clinical relevance. The emotional (Mooney and Speed, 2001), physical and economic costs (Feinstein and Rapoport, 2000; Yasuda et al., 2001) of mild head injury are considerable. Attempts at providing routine treatment to all patients have been disappointing in terms of reducing the morbidity (King et al., 1997; Wade et al., 1997, 1998; Paniak et al., 1998, 2000). Finding a predictable marker of poor outcome would offer many advantages, allowing resources to be focused in the immediate aftermath of injury on those patients deemed vulnerable. The APOE-ε4 allele offered one possible marker in this regard, but the data thus far suggest, at least in those with mild to moderate TBI, that outcome may be more closely linked to other factors. Future research involving catecholaminergic (McAllister et al., 2004) and 5-HT receptor subtypes (Lopez-Figueroa et al., 2004; Roth et al., 2004) might offer additional clues with respect to the genetic influence on outcome following mild to moderate TBI. However, before this question can be answered with greater certainty, the results of a 25-year follow-up study of patients with severe head injury (Millar et al., 2003) that failed to find an association between poor outcome and the APOE genotype need replicating, but this time in subjects whose brain injuries are milder. Since 15% of mild TBI patients remain persistently symptomatic 1 year after injury (Alexander, 1995), a further extension to the follow-up period may unmask deficits that are in part genetically modulated.


A. F. is supported by funding from the Canadian Institutes of Health Research, Grant 36535. We would also like to thank Marilyn Slater, MLT, for molecular genetics testing.


View Abstract