Major Depression Following Traumatic Brain Injury

Backgroud  Major depression is a frequent psychiatric complication among patients with traumatic brain injury (TBI). To our knowledge, however, the clinical correlates of major depression have not been extensively studied.

Objective  To determine the clinical, neuropsychological, and structural neuroimaging correlates of major depression occurring after TBI.

Design  Prospective, case-controlled, surveillance study conducted during the first year after the traumatic episode occurred.

Settings  University hospital level I trauma center and a specialized rehabilitation unit.

Methods  The study group consisted of 91 patients with TBI. In addition, 27 patients with multiple traumas but without evidence of central nervous system injury constituted the control group. The patients' conditions were evaluated at baseline and at 3, 6, and 12 months after the traumatic episode. Psychiatric diagnosis was made using a structured clinical interview and DSM-IV criteria. Neuropsychological testing and quantitative magnetic resonance imaging were performed at the 3-month follow-up visit.

Results  Major depressive disorder was observed in 30 (33%) of 91 patients during the first year after sustaining a TBI. Major depressive disorder was significantly more frequent among patients with TBI than among the controls. Patients with TBI who had major depression were more likely to have a personal history of mood and anxiety disorders than patients who did not have major depression. Patients with major depression exhibited comorbid anxiety (76.7%) and aggressive behavior (56.7%). Patients with major depression had significantly greater impairment in executive functions than their nondepressed counterparts. Major depression was also associated with poorer social functioning at the 6-and 12-month follow-up, as well as significantly reduced left prefrontal gray matter volumes, particularly in the ventrolateral and dorsolateral regions.

Conclusions  Major depression is a frequent complication of TBI that hinders a patient's recovery. It is associated with executive dysfunction, negative affect, and prominent anxiety symptoms. The neuropathological changes produced by TBI may lead to deactivation of lateral and dorsal prefrontal cortices and increased activation of ventral limbic and paralimbic structures including the amygdala.

Mood and anxiety disorders seem to be frequent psychiatric complicationsamong patients who have a traumatic brain injury (TBI).^1-6 Wehave published our findings for a group of 66 patients with acute TBI whowere followed up for more than 1 year.⁷ Duringthis period, 28 patients (42.4%) received a diagnosis of major depressionas diagnosed using a semistructured interview (Present State Examination)and DSM-III-R diagnostic criteria. Hibbard et al⁸ used a structured interview and DSM-IV criteria to identify Axis I psychopathological abnormalitiesin 100 adults with TBI who were evaluated, on average, 8 years after the traumaticepisode. The prevalence of major depression in this series was 61%.

More recently, Kreutzer et al⁹ studiedthe prevalence of major depressive disorder among a sample of 722 outpatientswith TBI, evaluated an average of 2½ years following TBI. Defined using DSM-IV criteria, major depression was diagnosed in 303patients (41.9%). Seel et al¹⁰ used the samemethod to assess the prevalence of depression among 666 outpatients enrolledat 17 US centers affiliated with the Traumatic Brain Injury Model SystemsProgram. Patients had moderate or severe head injuries and were evaluated10 to 126 months after the injury occurred. The prevalence of major depressionin this sample was 27%.¹⁰

A recent community study suggested an association between a historyof TBI and an increased lifetime prevalence of major depression.¹¹ Holsingeret al¹¹ found that the lifetime prevalenceof major depression among men who sustained a head injury during World WarII was 18.5% vs 13.4% for a comparable group who did not. In addition, currentrates of depression were higher in patients who sustained a brain injury 50years ago or longer, suggesting that patients with a head injury have recurrentdepressive disorder throughout their lifetime at a significantly higher frequencythan comparable patients without a head injury. Furthermore, Koponen et al¹² assessed the frequency of Axis I and Axis II disordersin a group of 60 patients followed up for 30 years after the TBI occurred.These patients were particularly vulnerable to develop depressive disorders,showing a lifetime prevalence of major depression of 26.7%.

Patients with depression and cerebrovascular disease show prominentexecutive dysfunction, particularly those patients with late-onset depressivedisorders and ischemic deep white matter and basal ganglia lesions.^13,14 Although executive function deficitshave been consistently described in patients with TBI,^15,16 toour knowledge, the relationship between mood disorders and executive dysfunctionhas not been previously described.

Major depressive disorder is a syndrome of variable causes and probablydifferent underlying pathophysiological abnormalities. Neuroimaging studiesin patients with primary depression have found structural and metabolic abnormalitiesin regions of the prefrontal cortex, including dorsolateral prefrontal,^17-21 anteriorcingulate,^22-26 andorbitofrontal cortices.^27,28 Furthermore,recent neuropathological studies demonstrated that compared with control subjects,patients with familial unipolar and bipolar depression show a reduction inboth the density and number of glial cells in subgenual cortex²³ aswell as the density and size of neurons and glial cells in the orbital anddorsolateral aspects of the prefrontal cortex.^23,29-31

In the present study, we assessed the frequency of comorbid conditionssuch as anxiety disorders and aggressive behavior that would provide furtherinsight to the clinical phenomenology and the biological mechanisms leadingto the onset of major depression among patients with TBI. The relationshipbetween major depressive disorder and cognitive disturbance was analyzed usingan extensive neuropsychological battery of tests. Finally, the associationof major depression with the type, extent, and location of brain damage wasevaluated using more sensitive neuroimaging methods. We hypothesized thatfrontal lobe dysfunction, particularly in the left hemisphere, would be associatedwith major depression.

The study group consisted of 91 consecutive patients with closed headinjury admitted to the University of Iowa Hospitals and Clinics, Iowa City(n = 60) or the Iowa Methodist Medical Center, Des Moines (n = 31). Patientswith penetrating head injuries or those with clinical or radiological findingssuggesting spinal cord injury were excluded from the study. Patients withsevere comprehension deficits (ie, those who were unable to complete partII of the Token Test³²) that precluded a thoroughneuropsychiatric evaluation were also excluded from the study. In addition,27 patients with multiple traumas but without clinical or radiological evidenceof central nervous system involvement constituted our control group. Sixty-eight(74.7%) of the 91 patients with TBI were injured in a motor vehicle collision,16 patients (17.6%) by a fall, 3 patients (3.3%) by assault, and 4 patients(4.4%) by other mechanisms (eg, sport-related injuries). All 118 subjectsgave written informed consent for participation in this study.

Both patients with TBI and the controls were followed up at 3, 6, and12 months. The mean (SD) time of follow-up was not significantly differentbetween the TBI and the general trauma groups (9.38 [4.2] months and 9.26[3.9] months, respectively). We compared the demographic and clinical characteristicsof the patients who completed the study with those who dropped out of thestudy. Patients who dropped out were significantly younger (mean [SD], 27.2[7.0] years) than patients who remained in the study (38.2 [12.5] years) (F_[1,90] = 7.3, P<.009). Otherwise, therewere no significant differences between the groups for sex, race, socioeconomicor employment status, marital status, or educational level. In addition, therewere no significant differences between those who dropped out and those whocompleted the study in the severity of the TBI, degree of functional or cognitiveimpairment, premorbid social functioning, frequency of psychiatric disorders,or frequency of alcohol or other substance abuse.

Severity of the TBI was assessed using the 24-hour Glasgow Coma Scale(GCS)³³ score. According to this measurement,GCS scores between 13 and 15 defined mild head injury; between 9 and 12, moderatehead injury; and between 3 and 8, severe head injury. Patients with a GCSscore in the 12- to 15-point range but who underwent intracranial surgicalprocedures or were seen with focal lesions greater than 15 mL, however, wereconsidered to have had a moderate head injury.³⁴ Theoverall severity of the traumatic injury was assessed using the AbbreviatedInjury Scale.²¹

All patients were assessed by a psychiatrist (R.E.J. or R.G.R.) using2 semistructured interviews, a modified version of the Present State Examination,³⁵ designed to elicit symptoms of mood and anxiety disorder,and the Structured Clinical Interview for DSM-IV diagnoses.^36,37 Severity of depressive and anxietysymptoms were assessed using the Hamilton Depression Rating Scale³⁸ and the Hamilton Anxiety Scale,³⁹ respectively.Aggressive behavior was assessed using the Overt Aggression Scale.⁴⁰

Family history of psychiatric disorders was assessed for first-degreerelatives using the family history method using Research Diagnostic Criteria.⁴¹ The Mini-Mental State Examination⁴² wasused as a global measure of cognitive functioning. Impairment in activitiesof daily living was assessed using the Functional Independence Measure.⁴³ Psychosocial adjustment was quantitatively assessedusing the Social Functioning Examination and Social Ties Checklist.⁴⁴ The Social Functioning Examination is a semistructuredinterview assessing different aspects of psychosocial adjustment such as qualityand satisfaction of interpersonal relationships, performance of home and familyresponsibilities, work experience, social activities, economic practices,stability of family income, living environment, spiritual beliefs, and useof community resources. Scores range from 0.0 to 1.0 with the higher scoresindicating greater impairment. The Social Ties Checklist is a 10-item questionnairethat determines the number of social connections (eg, frequency of seeingfriends or membership in clubs, churches, or other organizations) availableto the patient. Scores range from 0 to 10 with the higher scores indicatingfewer social connections. Initial Social Ties Checklist scores assessed socialsupport networks prior to the traumatic episode. The reliability and validityof each of these instruments has previously been demonstrated in populationswith brain injuries.⁴⁵

Computed tomographic scans and occasionally magnetic resonance (MR)imaging were obtained as part of the standard clinical evaluation in the emergencyand neurosurgery departments of the participating institutions. The nature,extent, and location of traumatic lesions were classified in accord with theTraumatic Coma Data Bank criteria and registered using the appropriate forms.⁴⁶ A neurologist trained in the assessment of structuralneuroimaging scans, who was blind to the results of the psychiatric examination,read all of the scans.

In addition a research MR image was obtained in patients with TBI atthe time of the 3-month evaluation using a 1.5-T scanner (GE Sigma, Milwaukee,Wis) at the radiology department of the University of Iowa. The tools of alocally developed software package, BRAINS (Department of Psychiatry, Universityof Iowa, Iowa City), were used to generate volumetric data. This softwarepermits cross-modality image registration, automated tissue classification,automated regional identification, cortical surface generation, volume andsurface measurement, 3-dimensional visualization of surfaces, and multiplanartelegraphing. The validity and reproducibility of morphometric analysis usingthe aforementioned software has been reported in previous studies.^47-54

To quantify gray matter volume of the frontal lobe, an MR image−basedparcellation method was used.⁵⁵ This methodsubdivides the frontal lobe into 11 functionally relevant subregions on thebasis of individual gyral and sulcal topography. Morphometric tracings wereperformed by a research assistant who was extensively trained in this techniqueand who was blind to the psychiatric diagnosis and group assignment of theparticipants.

Participants underwent neuropsychological assessment evaluated by anexperienced neuropsychologist (D.M.) at the 3-month follow-up visit. Analysesincluded in this article focused on memory and frontal-executive functioning,as assessed by the following 8 tests: Rey Auditory Verbal Learning Test⁵⁶ (delayed recall trial); Rey Complex Figure Test⁵⁶ (delayed recall trial); Trail Making Test⁵⁷ (A and B/A ratio); Multilingual Aphasia Examination⁵⁸ (controlled oral word association); Stroop Color-WordInterference Test⁵⁹; Wisconsin Card-SortingTest (the number of categories achieved and the number of perseverative errors).⁶⁰

Comparison of the groups used simple χ² analyses whenthe expected frequencies were sufficiently large and the Fisher exact testwhen the χ² test was inappropriate. Because some of our continuousmeasures were clearly nonnormally distributed, we chose the Mann-Whitney testfor comparing the groups. To remain consistent, the Mann-Whitney test wasused for all such comparisons. Data are given as mean (SD).

According to their initial GCS scores and initial computed tomographicdata, of the 91 patients who sustained a TBI, 40 patients (44.3%) had a mildTBI, 30 patients (32.5%) had a moderate TBI, and 21 patients (23.2%) had asevere TBI. Although their GCS scores fell within the mild range, 10 patientswere classified as having a moderate TBI because they required a surgicalevacuation procedure and/or they were initially seen with focal lesions largerthan 15 mL. In accord with the Traumatic Coma Data Bank Classification, 66(72.1%) of the 91 patients had diffuse lesions and 25 patients (27.9%) hada mass or focal pattern of injury.

The demographic and background characteristics of the TBI and controlgroups are given in Table 1. Therewere no significant differences between the TBI and control groups for age,sex, race, or socioeconomic status. There were no significant between-groupdifferences in Abbreviated Injury Scale and Functional Independence Measurescores. Thus, compared with patients with TBI, the controls had experiencedcomparable traumatic injuries and had a similar degree of functional impairment.In addition, there were no significant differences between the control andTBI groups in the frequency of personal history of psychiatric disorders orpersonal history of substance abuse.

Mood disorders were significantly more frequent among patients withTBI. Of 91 patients with TBI, 47 patients (51.6%) developed a mood disorderat some time during the first year after injury compared with 6 (22.2%) of27 patients with multiple traumatic injuries but without central nervous systeminvolvement (χ²₁ = 7.3, P =.006). Of the 91 patients, 47 patients met DSM-IV criteriafor mood disorder due to TBI, 30 patients (33%) presented with major depressivefeatures, 9 patients (9.9%) had depression without major depressive features,while the remaining 8 patients (8.8%) had manic or mixed features. Of the27 controls, 2 patients (7.4%) had major depressive disorder and 4 patients(14.8%) had depression without major depressive features. The frequency ofmajor depressive disorder was significantly greater in the patients with TBI(P = .008, Fisher exact test). Of the 91 patientswith TBI we have excluded 17 patients who developed mania or subsyndromaldepression; thus, the final number of participants included in the study were74 (ie, 30 patients with major depressive features and 44 nondepressed patients).

Patients with a mood disorder due to TBI with major depressive featureswill be the subject of the present article and will be compared with thosepatients who did not develop mood disorders during the first year after theTBI occurred. We have previously shown that patients with posttraumatic manicand hypomanic syndromes have different clinical correlates than those withmajor depressive disorder.⁶¹ It is conceivablethat post-TBI manic syndromes have different pathophysiological mechanisms.These syndromes will be the focus of an independent study. On the other hand,in contrast with what has been demonstrated in other depressed populationssuch as geriatric patients, there is no empirical evidence validating minoror subsyndromal depressions occurring after TBI as a helpful construct withdistinct clinical implications. For example, we did not find an effect ofminor depression on the long-term outcome of patients with TBI.⁶² Inaddition, there is a high degree of overlap of minor depression with otherprevalent conditions such as adjustment disorders and postconcussive syndromes.In fact, the frequency of minor depression was not significantly differentamong our patients with either TBI or general trauma. On the other hand, majordepressive disorder has been more extensively validated in different TBI samples,and the specificity and sensitivity of DSM-basedcriteria adequately demonstrated.

Major depressive disorder following TBI was significantly associatedwith the presence of anxiety disorders. Of 30 patients with major depressivedisorder, 23 (76.7%) met diagnostic criteria for a comorbid anxiety disordercompared with 9 (20.4%) of 44 patients who did not develop a mood disorderbut met criteria for an anxiety disorder during the first year following TBI(χ²₁ = 24.3, P<.001).Of these 23 patients who had major depression and a coexisting anxiety disorder,14 patients presented with generalized anxiety features, 2 patients had generalizedanxiety and panic attacks, and 7 patients met diagnostic criteria for posttraumaticstress disorder.

Major depression was also associated with the occurrence of aggressivebehavior that was categorized using the Overt Aggression Scale. Of the 30patients with major depression, 17 patients (56.7%) demonstrated significantaggressive behavior (Overt Aggression Scale scores >3 or a score of 3 withphysically aggressive behavior against self or others) compared with 10 of44 patients who showed the same level of aggression without mood disorderduring the first year after the TBI (χ²₁ = 8.9, P = .003).

Of the 30 patients who developed major depression, 15 patients (50%)received the diagnosis at the initial evaluation, 9 patients (30%) receivedthe diagnosis at the 3-month follow-up, and 6 patients (20%) received thediagnosis at the 6-month follow-up. Thus, half of the patients developed majordepression during the subacute period following TBI. We obtained follow-updata in 24 (80%) of the 30 patients with major depression during the firstyear after the TBI occurred. The average time of follow-up was 10.5 months(range, 3-12 months) with 20 patients (67%) completing the 12-month follow-upevaluation. Primary physicians prescribed antidepressants in 8 (33%) of 24patients. Among the 16 patients who did not receive antidepressant therapy,major depressive disorder had a mean duration of 5.8 (2.7) months. The meanfor duration of depression among patients treated with antidepressants was4.7 (2.7) months, respectively.

There were no significant differences between the patients with majordepression and those who did not have major depression for age, sex, race,socioeconomic status, marital status, educational level, or annual income.However, the number of patients who were unemployed at the time of TBI wassignificantly greater for the patients with major depression (P = .004, Fisher exact test) (Table2).

The frequency of previous psychiatric disorders is summarized in Table 3. When compared with nondepressedpatients, patients with TBI who developed major depression had a significantlyhigher frequency of personal history of mood disorders (P = .01, Fisher exact test) and personal history of anxiety disorders(P = .05, Fisher exact test).

There were no significant differences between those who were depressedand those who were not in the frequency of previous or concurrent alcoholor other drug abuse. Furthermore, a history of alcohol and/or other drug abusewas not significantly more frequent in the group with major depression (χ²₁ = 0.64, P = .42). Interestingly,we did not observe a significant association between having a personal historyof mood or anxiety disorders and posttraumatic depressive disorders amongthe controls (ie, patients without central nervous system injury). Finally,the frequency of psychiatric disorders in first-degree relatives of patientswith TBI was not significantly different between the those with and thosewithout major depression.

These findings are summarized in Table 4. There were no significant differences between those withand those without major depression in activities of daily living impairmentas measured by Functional Independence Measure scores or in global measuresof cognitive function such as Mini-Mental State Examination scores. InitialSocial Ties Checklist scores were not significantly different between thesegroups. On the other hand, compared with patients who did not develop depression,patients with major depression had significantly higher Social FunctioningExamination scores at the 6-month follow-up (χ²₁ =11.4, P<.001, Mann-Whitney test) and at the 12-monthfollow-up (χ²₁ = 4.6, P =0.03, Mann-Whitney test). These findings are consistent with poorer psychosocialoutcome in the group with major depression.

We analyzed memory and executive functioning among patients who hadmajor depression at the time of the neuropsychological evaluation (ie, 3-monthfollow-up visit). The results are given in Table 5. Compared with the nondepressed group, those who were depressedhad lower scores on all 8 tests. Because the 2 groups differed significantlyin age, we transformed these variables into ranks and tested the between-groupsdifferences using analysis of covariance, covarying for age. There were significantdifferences in the Wisconsin Card-Sorting Test number of perseverative errors(F_1,4 = 4.76, P = .03) and Trail MakingTest B/A ratio (F_1,51 = 5.82, P = .02).

Patients with TBI who had a personal history of depressive disorderswho were not depressed at the time of the neuropsychological evaluation didnot differ from patients with TBI without a psychiatric history in memoryand executive functioning tests, suggesting that the neuropsychological deficitsobserved among patients with major depression were not the result of a chronicmood disorder that preexisted brain trauma and the current depressive episode.

The frequency of mild, moderate, and severe TBIs was not significantlydifferent between those with and those without major depression. There wereno significant intergroup differences in GCS scores or in Abbreviated InjuryScale scores (Table 4). Thus,severity of trauma was similar in both groups.

A research MR image taken at the 3-month follow-up visit provided thedata necessary to conduct a volumetric analysis of frontal lobe regions. Agroup of 17 patients with major depressive disorder was compared with 17 patientswho did not develop a mood disorder during the first year of follow-up. Thegroups were matched for age (±2 years), sex, and severity of TBI. Thosewith and those without major depression were not significantly different withregard to the frequency of diffuse or focal patterns of brain injury, thelocation of the lesion, or the volume of focal (mass) lesions. In addition,there were no significant intergroup differences in the frequency of previousor current alcohol or other drug abuse, which has been associated with volumetricchanges. We also compared these 34 patients with the remaining 57 patientswith TBI who were excluded from the volumetric analysis. There were no significantdifferences between the groups in demographic variables, severity of TBI,as well as in the cognitive, activities of daily living, or psychosocial impairmentvariables. Thus, these groups were essentially comparable.

There were no significant differences between those with and those withoutdepression in total brain volume, total gray matter volume, or total whitematter volume. Temporal, parietal, and occipital lobe gray matter volumes(standardized to percentages of total gray matter volume) were also similarbetween the 2 groups. However, patients with major depression had significantlydecreased frontal gray matter volumes compared with those in the nondepressedgroup (χ²₁ = 10.5, P =.001, Mann-Whitney test). Furthermore, patients with major depression hadsignificantly smaller left frontal gray matter volumes than the those in thenondepressed group (χ²₁ = 7.1, P = .008, Mann-Whitney test).

Using a previously validated parcellation method,⁵⁵ weidentified 3 frontal areas that have previously been associated with moodregulation and the pathophysiology of mood disorders. These were the orbitofrontalcortex (ie, orbitofrontal cortex and straight gyrus), the medial-frontal cortex(ie, anterior cingular gyrus and caudal medial frontal cortex), and the lateralprefrontal cortex (ie, superior middle and inferior frontal gyri). The superior,middle, and inferior frontal gyri include the dorsolateral and ventrolateralprefrontal cortex that are otherwise difficult to measure in a reliable waybecause of the lack of consistent anatomical landmarks. The results of thisanalysis are summarized in Table 6.Regional gray matter volumes have been normalized as percentages of the totalgray matter volume.

Compared with those in the nondepressed group, patients with major depressionshowed significantly decreased left lateral frontal cortex volumes (χ²₁ = 9.7, P = .001, Mann-Whitneytest). This difference was owing to significantly smaller left inferior frontalgyrus volumes (χ²₁ = 6.8, P =.009, Mann-Whitney test) as well as smaller left superior frontal gyrus volumes(χ²₁ = 3.8, P = .05, Mann-Whitneytest) and left middle frontal gyrus volumes (χ²₁ =3.2, P = .08, Mann-Whitney test). Although rightlateral frontal volumes were smaller among those with major depression, thisdifference did not reach statistical significance. There were no significantdifferences between the 2 groups in orbitofrontal or medial prefrontal volumes.

To our knowledge, this is the first prospective study analyzing theclinical, neuropsychological, and neuroimaging correlates of major depressivedisorder following TBI using a complete neuropsychological battery and a refinedstructural MR imaging analysis. The principal findings of this study may besummarized as follows: major depressive disorder was observed in 30 (33%)of 91 patients during the first year after TBI. Major depressive disorderwas significantly more frequent among patients with TBI than among patientswith traumatic injuries of comparable severity but without involvement ofthe central nervous system. Compared with the group of patients who did notdevelop a mood disorder, those with major depression were more likely to havea personal history of mood and anxiety disorders. Most of the patients withmajor depression exhibited comorbid anxiety disorders and aggressive behavior.Patients with major depression showed significantly greater impairment inproblem-solving ability and cognitive flexibility than their nondepressedcounterparts. Major depression was also associated with poorer social functioningat the 6- and 12-month follow-up visits, as well as significantly reducedleft prefrontal gray matter volumes, particularly the ventrolateral and dorsolateralregions.

Before discussing the implications of this study, we should acknowledgeits methodological limitations. First, most of our patients were young malesof caucasian origin, reflecting the demographic characteristics of Iowa. Thus,our findings may not pertain to other groups of patients with TBI. Althoughwe made a special effort to obtain complete longitudinal data, 16 (21.6%)of 74 patients included in the present analysis were unavailable for follow-up.Similar or greater attrition rates are common in the vast majority of longitudinalstudies of patients with TBI. As aforementioned, the clinical and demographiccharacteristics of patients who dropped out were essentially similar to thoseof patients who completed the study. Thus, we believe that attrition had onlya limited effect on the findings that were addressed earlier. Given theselimitations, what are the most important implications of the present study?

Major depressive disorder as well as total mood disorders were significantlymore frequent in patients who sustained TBIs than in patients with similarbackground characteristics who underwent similar levels of stress (eg, motorvehicle collisions) but who did not sustain brain injury. Thissuggests that neuropathological processes associated with TBI constitute animportant contributing factor to the development of these mood disorders.

The reported frequency of major depression (33%) is consistent withthe findings of 2 recent large cross-sectional studies of patients with TBI.^9,10 These patients, however, were assessedlater during the course of recovery.

Patients with major depression were more likely to have a history ofmood and anxiety disorders and it is reasonable to assume that they were moreprone to develop psychiatric symptoms and major depression when exposed tosignificant stress. However, the fact that we did not observe the same effectof psychiatric history among patients without brain injury suggests that thisfactor might not play a decisive causative role. Interestingly, the frequencyof alcohol or other drug abuse was not significantly different between the2 groups, a fact that suggests that the pathological changes in reward andmood regulation circuits observed in patients with addictive disorders didnot convey a significantly higher risk of developing major depressive disorderin this group of patients with TBI. Finally, although it does not excludethe role of genetic factors in the cause of post-TBI depression, a familyhistory of mood or anxiety disorders was not significantly more frequent amongpatients with major depression.

Although patients with major depression and TBI were not different fromthose patients with TBI only with regard to global measures of cognitive functionsuch as the Mini-Mental State Examination score, they were significantly impairedin neuropsychological tests assessing executive functioning.

Traumatic brain injury has been consistently associated with damageto the prefrontal cortex, basal ganglia, and the white matter tracts thatconnect these structures. Executive dysfunction and depression may be relatedto the same pathophysiological mechanism (ie, the disruption of these fronto-striatal-thalamiccircuits). Patients with major depression did not differ from the nondepressedcontrols with TBI in regard to the severity of TBI or the frequency and overallextent of frontal lobe lesions identified in their initial neuroimaging studies.Certainly, the selective volumetric changes in left prefrontal cortex observedamong patients with major depression may contribute to their cognitive deficits.It is conceivable, however, that a mechanism specific for depression (eg,abnormal aminergic modulation of prefrontal structures) may also produce impairmentin executive functioning as observed in the present group of patients.

Major depression was associated with reduced gray matter volume in thelateral aspects of the left prefrontal cortex. We have previously reportedon the selective involvement of left prefrontal and left basal ganglia lesionsin patients with acute major depressive disorder following TBI.⁶³ Otherstudies of secondary depressive disorders have also found decreased metabolicrates in inferior frontal regions in patients with Parkinson disease,⁶⁴ Huntington disease,⁶⁵ andcaudate stroke.¹⁸

It is unclear if the reduced prefrontal volumes observed in patientswith major depressive disorder are the result of the pathophysiological mechanismsinitiated by TBI or they constitute a preexistent trait associated with anincreased risk to develop mood disorders. Brain atrophic changes can be observedamong patients with chronic mood disorders.^30,66,67 Weanalyzed the effect of a history of anxiety or depressive disorders on frontallobe volumetric measures. There were no significant differences between patientswith a history of depressive or anxiety disorders and patients without a historyof psychiatric illness in total frontal lobe volume, total frontal gray andwhite matter volumes, or gray matter volumes of frontal subregions includingleft inferior frontal gray matter volume. We can also hypothesize that socialdeprivation and unemployment can be associated with prefrontal cortex changes.However, unemployed patients did not show significant reductions in prefrontalvolumes. In fact, patients with and without major depression who were unemployedhad higher prefrontal volumes than their employed counterparts. Thus, thereis no evidence to support the idea that asymmetric differences in frontallobe volume preexisted the brain injury and we believe that the decreasedleft frontal lobe volume is the result of resolving traumatic lesions approximately3 months after the TBI occurred.

Recent experimental studies of TBI suggest that diffuse neuronal damageand cell loss may progress over weeks to months after the initial insult inselectively vulnerable regions of the neocortex, hippocampus, thalamus, andstriatum.^68-70 Onthe other hand, neuroimaging studies of patients with TBI have demonstrateddelayed cerebral atrophy on computed tomographic scan and MR imaging,^71-73 as well as alteredmetabolic patterns consistent with neuronal loss and inflammation as evidencedby proton magnetic resonance spectroscopy.^74,75 Furthermore,behavioral outcome seems to be more strongly correlated with delayed ratherthan early imaging findings.^76,77

Whatever the case may be, major depression could result from deactivationof more lateral and dorsal frontal cortex and increased activation in ventrallimbic and paralimbic structures including the amygdala.^78-80 Thecognitive abnormalities observed in patients with TBI and major depressionare consistent with left lateral prefrontal dysfunction. Interestingly, highlevels of amygdala activation may be associated with an increased prevalenceof anxiety symptoms and negative affect,⁸¹ apattern of symptoms that closely resembles what we observed in our group ofpatients with TBI. Moreover, faulty prefrontal modulation of medial limbicstructures could explain the impulsive and aggressive behavior observed inthese patients.^82,83

Major depressive disorder is a frequent complication of TBI that exertsa deleterious effect on the recovery process and psychosocial outcome of patientswith brain injuries.^62,84 Biologicalfactors such as the involvement of the prefrontal cortex and probably otherlimbic and paralimbic structures may play a significant role in the complexpathophysiology of major depression. Future studies need to further characterizethese factors to identify patients with TBI who are at high risk of developingmajor depression and to design appropriate therapeutic interventions.

Corresponding author and reprints: Ricardo E. Jorge, MD, MEB/PsychiatryResearch, 500 Newton Rd, Iowa City, IA 52242 (e-mail: ricardo-jorge@uiowa.edu).

Submitted for publication February 20, 2003; final revision receivedJune 17, 2003; accepted June 25, 2003.

This study was supported in part by grants MH-40355, MH-52879, and MH-53592from the National Institute of Mental Health, National Institutes of Health,Bethesda, Md.

We thank Russell Hansen for imaging analysis and Teresa Kopel and StephanieRosazza for their support during this study.