Corticolimbic Blood Flow During Nontraumatic Emotional Processing in Posttraumatic Stress Disorder FREE

Posttraumatic stress disorder (PTSD) can occur after trauma exposure (eg, combat, assault, and disasters), and it is characterized by various altered emotional responses. Patients with PTSD not only experience intense negative emotional reactions when reminded of their trauma but also report anhedonia, social withdrawal, isolation, and decreased emotional expressivity, referred to as “emotional numbing.”¹ Characterizing the neural basis of these diverse, distorted emotional responses poses a major challenge to contemporary psychiatric research.

Functional neuroimaging techniques, such as positron emission tomography (PET) and functional magnetic resonance imaging, provide an opportunity to examine the neural correlates of these altered emotional responses in PTSD.^2- 6 To date, studies have focused primarily on brain activation in response to trauma-related stimuli. Exposure to combat sights and sounds in combat veterans with PTSD has been associated with relatively increased amygdala activation^7- 8 and relatively decreased ventral medial prefrontal cortex (MPFC) or subcallosal anterior cingulate cortex (ACC) activation.⁹ The recollection/imagery of combat and civilian traumatic events in PTSD has been associated with amygdala, orbitoprefrontal cortex (OPFC), and insular cortex¹⁰ activation and relatively decreased blood flow in the MPFC or ACC.^11- 14 These findings are central to current functional neuroanatomic models of PTSD^2- 6 and are supported by findings from animal experiments, lesion studies, and functional imaging in healthy individuals. These studies^15- 18 collectively implicate the same paralimbic (insula, MPFC, and OPFC) and limbic (amygdala and ACC) regions in perceiving and responding to emotionally salient stimuli and affective regulation. Although limbic and paralimbic frontal dysfunction has been consistently linked to trauma-related symptom provocation in PTSD,^2- 5 it remains unclear whether a similar neural pattern extends to nontraumatic probes. Do abnormal emotional responses “generalize” to nonspecific stimuli? The use of generic emotional stimuli that are not linked to an individual's trauma may thus be informative.

Initial findings of abnormal responses to emotional stimuli not related to trauma in PTSD point toward the same regions. Patients with PTSD exhibit exaggerated amygdala activity to masked¹⁹ and overtly presented²⁰ fearful faces, suggesting an abnormal amygdala response to general threat-related stimuli. In the study²⁰ using overtly presented fearful faces, a diminished MPFC response was also observed. Previously, Shin et al²¹ observed that PTSD subjects, unlike controls, did not activate the rostral ACC and MPFC in response to general negative words in an emotional Stroop paradigm. Similarly, Bremner and colleagues²² reported that during the retrieval of emotional words, the PTSD group exhibited deactivations/decreased regional cerebral blood flow (rCBF) throughout the frontal brain areas (ACC, OPFC, and MPFC). These findings, although implicating limbic and paralimbic involvement in affective dysfunction in PTSD, may reflect more perceptual or cognitive aspects of emotion than affective experience itself. Lanius and colleagues²³ included traumatic and nontraumatic emotional life events in a script-driven imagery study and reported lower ACC activation in PTSD across all imagery conditions.

Unlike emotional faces or words, complex visual pictures from the International Affective Picture System²⁴ directly elicit emotional experience and response.²⁵ Furthermore, these standardized pictures contain emotionally salient content unrelated to personal traumatic experience and, therefore, are suited for the investigation of general emotional experience. These emotional pictures serve as reliable probes of the paralimbic and limbic brain regions^26- 31 implicated in models of PTSD.^2- 6

In this PET activation study, we probed neural activity during general affective experience in PTSD by examining response to generally aversive (trauma-unrelated) visual stimuli. We studied individuals with PTSD in relation to 2 separate comparison groups: healthy controls never exposed to trauma and trauma-exposed individuals without PTSD. To our knowledge, this is the first large-sample functional imaging study of affective processing of emotional pictures comparing all 3 groups. We hypothesized that patients with PTSD would have differential responses to emotionally salient stimuli specifically in the amygdala, ACC, and paralimbic cortex (dorsal MPFC, ventral MPFC/OPFC, and insular cortex) relative to controls.

Sixteen Vietnam veterans with combat-related PTSD (PTSD patients), 15 Vietnam combat-exposed veterans without PTSD (combat controls [CCs]), and 15 healthy age- and sex-matched individuals without previous trauma exposure (normal controls [NCs]) participated in this study. The data from 1 PTSD patient were corrupted and omitted from subsequent analysis. The PTSD diagnosis and evaluation of other Axis I psychiatric disorders were established using the Structured Clinical Interview for DSM-IV.³² Besides the PTSD diagnosis in the PTSD group, all the participants were free of current Axis I disorders. All the participants were right-handed and had not been taking psychotropic medications for at least 4 weeks before the study (6 weeks in the case of fluoxetine and antipsychotic medications). The participants had no history of major medical or neurologic illness, and those with histories of substance abuse or dependence were in remission for at least 6 months. To control for potentially confounding effects of past illicit drug use or abuse, we recruited controls matched for substance use history to PTSD patients. Results of urine toxicology screens were negative for all the participants. Magnetic resonance imaging confirmed the absence of any structural abnormalities. All the participants gave written informed consent after explanation of the experimental protocol, as approved by the University of Michigan institutional review board and the Ann Arbor Veterans Administration Medical Center subcommittee on human subjects.

As seen in Table 1, the groups had similar ethnic backgrounds and past psychiatric comorbidity. All the participants were aged 35 to 60 years. The ages of the PTSD patients were not significantly different from those of either control group; however, the CC group was older than the NC group (F_2,3 = 6.646; P = .003). The PET data analysis comparing controls incorporated age as a nuisance variable to account for this difference.

The emotional activation paradigm has been previously reported.³³ Participants viewed 2 sets of complex International Affective Picture System gray-scaled images: an aversive set containing negatively valenced/unpleasant images and a nonaversive set containing images of neutral valence. The sets were matched with respect to the presence of faces and human figures. Another set of blank images consisted of gray screens with a fixation cross, serving as a low-level control condition. All the image sets were matched on luminance using image-editing software (Photoshop 4.0; Adobe Systems, Mountain View, Calif).

We acquired PET, psychophysiologic, and behavioral ratings data while participants viewed blocks of aversive, neutral, or blank images. Fourteen images per block were displayed for 5 seconds each, with no interstimulus interval; no pictures were presented more than once. Data acquisition consisted of 2 runs of the 3 conditions (aversive, nonaversive, and blanks), counterbalanced across participants. During scanning, participants verbally rated each picture for aversive content on a 5-point scale from 1 (not at all unpleasant) to 5 (extremely unpleasant). During the blank condition, participants used a rating of 3 for each blank screen. We chose this appraisal/labeling task to support task compliance in the absence of empirical evidence that it could affect corticolimbic blood flow, as recently reported (see the “Comment” section).³⁴ The participants also participated in a script-driven imagery paradigm in a separate experiment, reported elsewhere.³⁵ The order of the experiments was counterbalanced across participants.

Skin conductance was recorded using a psychophysiologic monitoring system (model MP-100; BioPac Systems, Santa Barbara, Calif) and silver–silver chloride electrodes attached to the finger volar surface. Waveform peak and integral during image presentation were analyzed. Subjective ratings and skin conductance responses were analyzed using separate 3 × 3 (group [NCs, CCs, and PTSD patients] × condition [aversive, nonaversive, and blank]) analyses of variance.

The PET procedures have been described previously.³³ In brief, PET was performed using the Siemens ECAT EXACT and HR+ scanners (CTI Molecular Imaging Inc, Knoxville, Tenn), which yield 47 or more sections simultaneously for whole-brain coverage at an axial field of view of approximately 15 cm and a resolution of 4.5 mm full-width half-maximum. Although both scanners have similar technical specifications, any differences between them were minimized as follows: (1) an algorithm derived from point-source calculations (as supplied by CTI) was performed on data acquired from the HR+ scanner to equate sensitivities; (2) data were reconstructed using the same 0.5 Hanning filter to yield an effective full-width half-maximum of 10 to 11 mm; (3) data were collected on the same number of participants in each group using each scanner (EXACT: 10 PTSD patients, 10 CCs, and 11 NCs; HR+: 6 PTSD patients, 5 CCs, and 4 NCs); (4) statistical analysis was based on intrasubject subtraction (comparisons within subjects with fixed-effects first-level analyses); and (5) scanner type was entered as a categorical nuisance variable into the analysis and confirmed that scanner type did not affect any of the activation foci reported in the “Results” section.

Participants were positioned in the PET scanner gantry, using the orbitomeatal line as a reference and a forehead restraint to reduce intrascan movement. For each PET scan, 10 to 15 mCi of [¹⁵O]H₂O was given as an intravenous bolus. Data were collected in 3-dimensional acquisition mode as a single 60-second frame beginning 2 seconds after the radioactivity arrived in the brain. Lights were dimmed, and ambient noise (from cooling fans) was minimal. Scans were separated by 12 minutes. Stimuli presentation began 5 seconds before image acquisition.

All PET images of integrated tissue counts were decay corrected and reconstructed using filtered back-projection with a Hanning filter to yield an effective full-width half-maximum of 10 to 11 mm, with standard attenuation and scatter correction.³³ The PET data were analyzed using statistical parametric mapping (SPM99; Wellcome Department of Cognitive Neurology, London, England). The PET images were realigned to the first image in the scanning session, spatially normalized to the Montreal Neurological Institute brain template, adjusted for global values, and smoothed using an isotropic Gaussian filter of 12-mm width (full-width half-maximum). The data were fit to a general linear statistical model according to standard processing methods.^36- 39 Hypotheses were tested as contrasts in which linear combinations of the model parameters were evaluated using t statistics. Relevant contrasts between conditions (nonaversive vs blank and aversive vs nonaversive) were first performed for each participant³⁷ and then entered into a second-level random-effects within-group analysis using a 1-sample t test on contrast images obtained in each participant for each comparison of interest, treating participants as a random variable.⁴⁰ We used a conservative random-effects analysis designed to permit general population inferences and between-group comparisons. Resulting statistical parametric maps of the t statistic at each voxel were transformed into z scores. We had a priori hypotheses about differential activation between groups in specific brain regions (amygdala, insula, ACC, MPFC, and OPFC) and, therefore, using the small-volume correction toolbox in SPM99, we examined significant activations in a search area covering 672 mL of brain tissue comprising 2 volumes: (1) one with dimensions of 140 × 60 × 50 mm centered at [0, −10, −5] to cover the amygdala and insula and (2) another with dimensions of 40 × 70 × 90 mm centered at [0, 35, 15] to cover the entire medial wall of the frontal cortex (ACC, MPFC, and OPFC). For the within-group analysis, activation foci in this search volume were considered significant if they survived a P<.005 corrected for multiple comparisons using the false discovery rate procedure in SPM99 such that only 5 of every 1000 voxels are falsely accepted as significant,⁴¹ with a cluster extent threshold (k) of greater than 5 contiguous voxels. For the between-group analysis, we first performed an omnibus F test for a group effect to test for the presence of activations in a priori regions before any post hoc pairwise t tests (eg, NC>PTSD patients). Only foci that exhibited a significant effect of group in the omnibus test were considered to be significant in subsequent between-group comparisons. Significance thresholds for a priori regions exhibiting between-group differences (2-sample t tests) were set at the same level (P<.005, false discovery rate corrected for multiple comparisons). Outside a priori regions, we report activations surviving correction for multiple comparisons across the entire brain (false discovery rate–corrected P<.005).

We also performed a correlational analysis to determine whether clinical symptoms of PTSD patients were associated with rCBF patterns. Beck Depression Inventory, Clinician-Administered Posttraumatic Stress Disorder Scale, and Toronto Alexithymia Scale scores were entered as covariates of interest in the whole-brain voxelwise analysis. In addition, interregional correlation analyses in each group examined the relationship between rCBF responses in the left amygdala and the rest of the brain given recent evidence of a functional relationship between the amygdala and prefrontal regions (ACC and MPFC) in PTSD.^42- 44 Data extracted from a functional region of interest in the left amygdala (MNI [Montreal Neurologic Institute] coordinates: [−16, −6, −28]), which was observed to be differentially activated in the NC and PTSD groups, was correlated with whole-brain, voxelwise changes in rCBF for the aversive vs nonaversive contrast.

Aversive pictures were rated more negatively than neutral pictures (F_1,43 = 231.4; P<.001); between-group differences (P = .64) and group × condition interactions (P = .78) were not significant. Aversive pictures also elicited a greater skin conductance response (F_2,42 = 15.8; P<.001); between-group differences and group × condition interactions were not significant (P = .91 and P = .92, respectively) (Table 2).

Within-group analyses focused on the amygdala, insula, and medial frontal cortex as a priori regions of interest. The omnibus F test for a group main effect did not reveal any significant foci in the nonaversive vs blank contrast. For the aversive vs nonaversive contrast, a group effect was detected in the amygdala ([−18, −8, −28]; z = 2.65; P<.005) and in the ventral MPFC ([−8, 58, −20]; z = 2.64; P<.005) (Table 3 and Table 4).

In the nonaversive vs blank comparison, no activations or deactivations in the amygdala (eg, decreased rCBF) were noted in any group; there were no between-group differences. In the aversive vs nonaversive comparison, greater amygdalar rCBF response to aversive pictures was observed in NCs and CCs but not in PTSD patients (Table 3). In between-group comparisons, NCs and CCs had greater activation of the left amygdala to aversive pictures than the PTSD group (Table 4 and Figure). In the aversive vs blank comparison, NCs also had greater activation of the left amygdala ([−28, 0, −26]; z = 3.00; k = 30) than the PTSD group. Similarly, CCs had greater amygdala activation ([−22, 0, −16]; z = 2.92; k = 100) than the PTSD patients. Both group differences (NC>PTSD and CC>PTSD) were contributed to by increased (eg, higher) activity in the control groups and the absence of response (rather than deactivation) in the PTSD group.

In the nonaversive vs blank comparison, neither activations nor deactivations were noted in the ventral MPFC/OPFC in any group. In the aversive vs nonaversive comparison, NCs activated the ventral MPFC, but trauma-exposed groups showed no activations in this region (Table 3). In the PTSD group, significant deactivations were observed in the ventral MPFC ([−10, 64, −16]; z = 3.47; k = 188). Consequently, NCs had greater activation of the ventral MPFC than the PTSD group due to activation in the NCs and deactivation in the PTSD patients (Table 4 and Figure). No differences between trauma-exposed groups were detected.

In the nonaversive vs blank comparison, neither activations nor deactivations of the ACC were noted in any group. Between-group differences were not significant. In the aversive vs nonaversive comparison, all 3 groups activated the right dorsal ACC (Table 3). No between-group differences were noted.

In the nonaversive vs blank comparison, all 3 groups activated dorsal MPFC (Table 3). No differences between groups were detected. In the aversive vs nonaversive comparison, only NCs activated the dorsal MPFC, but no between-group differences were detected.

In the nonaversive vs blank comparison, only CCs activated the insula, but no between-group differences were detected (Table 3). In the aversive vs nonaversive comparison, only NCs activated the bilateral insula, but no between-group differences were detected (Table 3).

No significant correlations were detected between brain activation to aversive pictures in the PTSD group and Clinician-Administered Posttraumatic Stress Disorder Scale, Toronto Alexithymia Scale, or Beck Depression Inventory scores. However, the absence of a significant finding may be due to a type II error related to the current sample size.

In relation to the left amygdala, negative correlations were observed only in the dorsal ACC for all 3 groups (NCs: [2, 8, 36]; z = 3.57; CCs: [0, 26, 20]; z = 3.33; and PTSD patients: [−8, 20, 28]; z = 3.64). Positive correlations were observed in the MPFC in the control groups (ventral MPFC—NCs: [−4, 66, 18]; z = 4.19; CCs: [−2, 62, −18], z = 3.44; and dorsal MPFC—NCs: [4, 26, 66]; z = 3.65; CCs: [−2, 42, 56]; z = 3.86). However, given the post hoc nature of these interregional correlation analyses and the temporal resolution of the PET technique, we caution against conclusive interpretations of causal or reciprocal relationships.

This PET activation study examined the neural correlates of processing emotionally salient stimuli unrelated to personal traumatic experience in PTSD and observed differential corticolimbic responses in the 3 groups studied. Both NCs and CCs activated the amygdala in response to aversive stimuli, whereas the PTSD group did not show a response in this region; this difference was statistically significant in group comparisons (NC>PTSD and CC>PTSD). During negative emotional experiences, NCs also activated ventral frontal regions (ventral MPFC/OPFC), whereas both combat-exposed groups exhibited either no response (CCs) or deactivation (PTSD patients); the difference between NCs and PTSD patients was statistically significant. Unlike NCs, the combat-exposed groups showed no insular response to aversive pictures. All 3 groups activated regions within the dorsal medial frontal cortex (dorsal ACC or dorsal MPFC) to aversive and nonaversive pictures; however, only NCs activated the dorsal MPFC to negatively valenced pictures. Collectively, these findings highlight differential neural responses in the limbic (eg, amygdala) and paralimbic prefrontal cortex (ventral MPFC) to general affective experience in PTSD.

Our results suggest altered activity in the amygdala in response to trauma-unrelated emotional stimuli in PTSD. Animal, human lesion, and functional imaging studies suggest that the amygdala has a critical role in emotion,^18,45- 47 in detecting fear-related cues and fear conditioning,^{15,18,48- 49} and in perceiving salience or general vigilance.^27,31,50 Accordingly, both non-PTSD groups (CCs and NCs) activated the amygdala during exposure to emotionally aversive stimuli. Although PTSD patients reported a negative subjective experience while viewing these pictures, they did not mount a similar increase in amygdalar activity. Our findings contrast with those of Rauch et al¹⁹ and Shin et al,²⁰ who demonstrated an exaggerated response in PTSD patients to masked and overtly presented fearful faces, respectively. One potential explanation for this discrepancy is that the aversive pictures may not specifically evoke fear but rather a more general negative emotional state.^18,51 Furthermore, the aversive stimuli did not contain material that was personally relevant to the PTSD group, and some studies,^7- 8,10,52 but not others,² have suggested that amygdala activation in PTSD is reserved for trauma-specific content Alternatively, this finding could reflect the emotional flattening observed in PTSD patients, who often recognize that they do not respond affectively to emotional situations. Behavioral evidence suggests that the subjective experience of PTSD patients to aversive pictures is similar to that of controls,^53- 54 and neuropsychologic studies suggest that the amygdala is critical not for recognition or identification of the emotional value but rather for mounting appropriate psychophysiologic responses.⁵⁵ Thus, the preserved subjective response and the lack of amygdalar response in our patients are consistent with the emotional numbing interpretation and may reflect the chronic emotional numbing observed in PTSD patients. However, because our study did not use positive-valence stimuli, the emotional numbing interpretation is necessarily preliminary.

Task instructions could affect activations in the limbic (eg, amygdala) and prefrontal cortex regions. Recent imaging studies^34,56- 58 have shown that when healthy volunteers label their emotions while viewing emotionally salient stimuli, amygdala activation is attenuated, whereas medial and lateral frontal cortex activity is enhanced. Although this may have affected the activation pattern, all 3 groups performed a similar task, and both control groups activated all relevant regions, including the amygdala and medial prefrontal regions, to emotional stimuli. If the absence of amygdala activations in PTSD patients is task dependent, future studies should specifically investigate the effect of cognitive-emotion interactions on corticolimbic brain activity during emotion processing in PTSD.^20,42

Neither combat-exposed group exhibited ventral MPFC activation to aversive pictures. The NCs activated ventral prefrontal regions, whereas PTSD patients had deactivation in this area, suggesting ventral MPFC hypoactivity. This deactivation to affective salience may be PTSD specific, which is consistent with reported absent or decreased MPFC activity in PTSD patients engaged in recall and imagery of personal trauma.^9,12- 13 Current observation further extends the evidence of diminished MPFC activation in PTSD beyond traumatic reexperiencing or symptom generation and suggests the potential presence of a more generalized hypofunction. This interpretation is consistent with findings of greater rCBF decreases in the ventral MPFC in PTSD patients during emotional words retrieval.²² Ventral MPFC deactivations have been associated with enhanced self-referential processing.⁵⁹ The MPFC is hypothesized to be involved in reexperiencing the “feelings” of one's emotional past,⁶⁰ in emotional awareness,⁶¹ and in reflecting on one's own emotions/feelings.⁶² Furthermore, an appropriate stress response seems to depend on the functional integrity of the MPFC^63- 64 and of the medial prefrontal-amygdalar pathway.^65- 66 The MPFC mediates the inhibition of conditioned fear after extinction,^67- 68 and lesions to this area interfere with fear extinction⁶⁹ and impair emotional regulation and social interactions.⁷⁰ An attenuated MPFC response to negative-valenced stimuli has been described in alexithymia,⁷¹ and “affect instability” has also been associated with reduced ventral MPFC function.^72- 74 Based on these converging observations, it is plausible that MPFC hypofunction may be specifically related to the difficulties that PTSD patients have in exhibiting appropriate emotional reactivity, extinguishing fear, or mounting an appropriate stress response.

Combat controls did not activate or deactivate ventral MPFC to emotional stimuli. Although group differences were not noted between control groups or between CCs and PTSD patients, this response might be related, at least partially, to a compensatory mechanism or an adaptation to trauma exposure not related to PTSD pathogenesis. A positive interregional rCBF correlation between the ventral MPFC and the left amygdala was observed only in CCs and NCs. Prefrontal activity may be functionally linked to amygdaloid activity,^{33- 34,65- 66} and effective coupling of PFC-amygdala function could serve to regulate negative emotions and as a buffer against psychiatric disorders.^58,75- 80 At present, the evidence regarding alterations in amygdala-MPFC coupling in PTSD is mixed. Shin and colleagues^20,42 reported that amygdalar responses are inversely related to those of the MPFC in PTSD patients using a similar analytic approach in 2 separate studies, whereas Gilboa and colleagues⁴³ found a positive relationship between the amygdala and the ventral MPFC (using a partial least squares “functional connectivity” approach). The lack of detectable group differences in the ventral MPFC warrants caution against the specificity that the ventral MPFC serves an adaptive compensatory function in CCs; however, the finding prompts further investigation of the complex relationship between the PFC and the amygdala as neural markers of resilience to psychiatric disease after stress.

All the groups activated the dorsal ACC in response to emotionally salient pictures; however, only NCs mounted a dorsal MPFC response to aversive pictures. The left amygdala rCBF was negatively correlated with the dorsal ACC rCBF in all 3 groups. We also observed a positive correlation between left amygdala activity and that of the dorsal MPFC in both control groups but not in PTSD patients. Although group differences in the dorsal frontal cortex were not detected, this may be a result of individual variability of response in this region to the negatively valenced pictures, and the present study may not have had adequate power to detect a small difference. These findings are not entirely consistent with previous evidence suggesting that dorsal portions of the ACC and MPFC may be hypofunctioning in PTSD, leading to diminished cognitive function or affective regulation during emotional processing. For example, Shin and colleagues²⁰ reported less ACC recruitment during the emotional counting Stroop task in combat-related PTSD²¹ and diminished ACC activity in response to overtly presented fearful faces. Furthermore, Lanius and colleagues^12- 13 demonstrated failure to activate the ACC and the dorsal MPFC during the recall and imagery of personal traumatic memories. Reduced pregenual and dorsal ACC volumes^81- 82 and decreased neuronal integrity (as indexed by N-acetylaspartate) in the ACC have been observed in PTSD.⁸³ These observations raise the notion that alterations in these regions may explain some deficits in cognitive functioning during emotional processing observed in PTSD patients and may be specific to PTSD. Such an interpretation is supported by animal, human lesion, and functional imaging studies, which collectively implicate dorsal frontal regions in affective regulation,^{17- 18,34,76,84- 86} including the cognitive control of negative emotional experience^{76- 77,86- 88} and cognitive-emotion interactions.^{29,34,89- 90} The potential functional consequences of differences in dorsal frontal cortical activity during emotion regulation in PTSD await further investigation.

Several limitations of this study are worth mentioning. First, the inclusion of PTSD patients with a history of comorbid conditions (eg, depression or substance abuse) and the exclusion of current active comorbid Axis I disorders are strengths and limitations. On the one hand, their inclusion matches better the representative, realistic combat veteran population; however, comorbid mood disorder and substance abuse, whether remote or current, opens the possibility that these psychiatric disorders contributed to the obtained results. However, previous psychiatric comorbidity was comparable among the groups. Similarly, because the traumatic event occurred several decades before this study, chronicity may have been a contributing factor. Also, only male combat veterans with PTSD were studied; therefore, the results do not necessarily generalize to civilian PTSD (eg, abuse, assault, and disasters) or to women. In addition, only aversive/negatively valenced stimuli were used, and future studies are needed to characterize brain activation to positive stimuli. In a cross-sectional study, it also cannot be determined whether the altered brain activation patterns preceded or followed traumatic exposure. However, the inclusion of healthy controls without previous trauma and combat-exposed individuals without PTSD allows examination of the impact of traumatic experience on brain activity. Last, aversive visual stimuli depicting dead bodies/physical wounds, although few in number, may have reminded combat-exposed individuals of their traumatic experience. All the individuals, however, were debriefed after PET, and none reported that the stimuli reminded them of their personal combat experience.

In conclusion, the present study investigated the neural basis of general affective experience in PTSD. By studying controls with and without previous traumatic exposure compared with PTSD patients, we can start differentiating trauma- and PTSD-specific brain activation changes. Ventral MPFC and amygdala hypoactivity to emotional pictures seems to be specifically related to PTSD. These findings prompt future neuroimaging studies to study general emotional processing and the relationship between prefrontal cortical regions and the amygdala to deepen our understanding of the neural circuitry in PTSD.

Correspondence: Israel Liberzon, MD, Department of Psychiatry, University of Michigan Medical School, 9D University Hospital, Box 0118, 1500 E Medical Center Dr, Ann Arbor, MI 48109-0118 (liberzon@umich.edu).

Submitted for Publication: July 9, 2004; final revision received June 14, 2005; accepted July 7, 2005.

Funding/Support: This study was supported by a Traumatic Stress Initiative Merit Award from the Department of Veterans Affairs and the Department of Defense (Drs Liberzon and Fig) and by the Ann Arbor Veterans Affairs Health System.

Previous Presentation: This work was presented in part at the Society for Biological Psychiatry annual meeting; May 16, 2002; Philadelphia, Pa.