Neurobiological Basis of Failure to Recall Extinction Memory in Posttraumatic Stress Disorder Mohammed R

Home , Extinction (psychology)

ARCHIVAL REPORTS Neurobiological Basis of Failure to Recall Extinction Memory in Posttraumatic Stress Disorder Mohammed R. Milad, Roger K. Pitman, Cameron B. Ellis, Andrea L. Gold, Lisa M. Shin, Natasha B. Lasko, Mohamed A. Zeidan, Kathryn Handwerger, Scott P. Orr, and Scott L. Rauch Background: A clinical characteristic of posttraumatic stress disorder (PTSD) is persistently elevated fear responses to stimuli associated with the traumatic event. The objective herein is to determine whether extinction of fear responses is impaired in PTSD and whether such impairment is related to dysfunctional activation of brain regions known to be involved in fear extinction, viz., amygdala, hippocampus, ventromedial prefrontal cortex (vmPFC), and dorsal anterior cingulate cortex (dACC). Methods: Sixteen individuals diagnosed with PTSD and 15 trauma-exposed non-PTSD control subjects underwent a 2-day fear conditioning and extinction protocol in a 3-T functional magnetic resonance imaging scanner. Conditioning and extinction training were conducted on day 1. Extinction recall (or extinction memory) test was conducted on day 2 (extinguished conditioned stimuli presented in the absence of shock). Skin conductance response (SCR) was scored throughout the experiment as an index of the conditioned response. Results: The SCR data revealed no signiﬁcant differences between groups during acquisition and extinction of conditioned fear on day 1. On day 2, however, PTSD subjects showed impaired recall of extinction memory. Analysis of functional magnetic resonance imaging data showed greater amygdala activation in the PTSD group during day 1 extinction learning. During extinction recall, lesser activation in hippocampus and vmPFC and greater activation in dACC were observed in the PTSD group. The magnitude of extinction memory across all subjects was correlated with activation of hippocampus and vmPFC during extinction recall testing. Conclusions: These ﬁndings support the hypothesis that fear extinction is impaired in PTSD. They further suggest that dysfunctional activation in brain structures that mediate fear extinction learning, and especially its recall, underlie this impairment.

Key Words: Amygdala, classical, conditioning, hippocampus, post- whereas extinction recall refers to the retrieval and expression of traumatic, magnetic resonance imaging, prefrontal cortex, stress the learned extinction memory after a delay (25). Understanding disorders the basis of these processes is important, given that one of the main clinical characteristics of PTSD is exaggerated and persis- he pathophysiology of posttraumatic stress disorder tent fear responses to reminders of the traumatic event. It is also (PTSD) has been extensively studied over the past several important in that the current behavioral treatment of choice, T years with neuroimaging and probes such as script-driven exposure therapy, relies on extinction-based mechanisms imagery and visual emotional stimuli (reviewed in [1–4]). Studies (26,27). have identified a network of dysfunctional brain regions, inclu- Pavlovian fear conditioning is commonly employed to probe ding amygdala; hippocampus; and subregions of the medial the neurobiology of fear acquisition and its inhibition in rodents prefrontal cortex, including ventromedial prefrontal cortex (28–31), and it has also been used in psychophysiological (vmPFC) and dorsal anterior cingulate cortex (dACC). Individuals (32–34) and neuroimaging studies of humans (35–37). In this with PTSD typically show exaggerated amygdala and diminished procedure, conditioned responses (CRs) are formed when a hippocampal activation relative to control subjects (5–10). The conditioned stimulus (CS) is paired with an aversive uncondi- dACC has emerged as another brain region that seems to be tioned stimulus (US), such as a mild electric shock. These CRs hyperactive in PTSD (11–13). Most studies have shown that can then be diminished or extinguished by the repeated presen- vmPFC is hypoactive in this disorder (12,14–21), but a few have tation of the CS in the absence of the US. Pavlovian fear reported hyperactivity (10,13,22–24). Although findings from conditioning and extinction are relevant to the neurobiology of these studies provide insight into the pathophysiology of PTSD, PTSD, given that this disorder involves learned fear (27) that the function of these brain regions within the context of fear might persist for decades after the trauma exposure (38). Study- extinction learning and its recall (or retention) has not been ing them might elucidate mechanisms by which perseverant fear directly examined. Extinction learning refers to the gradual, responses occur. The hypothesis that extinction of conditioned within-session decrements of conditioned fear responses, fear is deficient in PTSD (3,39) is supported by de novo fear conditioning and extinction studies that have demonstrated deficient extinction learning (40). Moreover, we recently re- From the Department of Psychiatry (MRM, RKP, CBE, ALG, LMS, NBL, MAZ, ported psychophysiological data indicating that Vietnam veter- SPO, SLR), Massachusetts General Hospital and Harvard Medical School, ans diagnosed with PTSD have an acquired impairment in the Boston; Department of Psychology (LMS, KH), Tufts University, Medford; retention of extinction memory (41). McLean Hospital (SLR), Belmont, Massachusetts; and the Department of Neurobiological research has advanced our understanding of Veterans Affairs (SPO), Medical Center, Research Service, Manchester, New Hampshire. the mechanisms underlying extinction learning and recall. Nu- Address correspondence to Mohammed R. Milad, Ph.D., Department of merous studies conducted in rodents with various pharmacolog- Psychiatry, Harvard Medical School and Massachusetts General Hospital, ical and molecular manipulations and electrophysiological and 149 13th St., CNY 2614, Charlestown, MA 02129. E-mail: milad@nmr. microstimulation tools have indicated that extinction learning mgh.harvard.edu. and recall involve different cellular mechanisms and possibly Received Mar 16, 2009; revised Jun 25, 2009; accepted Jun 26, 2009. different brain regions (for review, see [42]). For example, studies

0006-3223/09/$36.00 BIOL PSYCHIATRY 2009;66:1075–1082 doi:10.1016/j.biopsych.2009.06.026 © 2009 Society of Biological Psychiatry. Published by Elsevier Inc. All rights reserved. 1076 BIOL PSYCHIATRY 2009;66:1075–1082 M.R. Milad et al. suggest that, in addition to its role in fear acquisition, the excluded from the data analysis because of excessive motion in amygdala seems to be implicated in extinction learning, whereas the scanner. There remained 16 PTSD (6 women, 10 men) and 15 the vmPFC (corresponding to the infralimbic cortex in rodents) TENC subjects (8 women, 7 men, Fisher’s exact test p ϭ .48). and hippocampus seem to be involved in extinction recall (29,31,42–44). In contrast, a region dorsal to the vmPFC in rats, Psychodiagnostics, Demographics, and Psychometrics viz., the prelimbic cortex, has been found to promote condi- The Clinician-Administered PTSD Scale (CAPS) conferred tioned fear expression (45,46,47). PTSD diagnostic status. The Structured Clinical Interview for Neuroimaging studies have recently examined extinction DSM-IV Axis I Disorders (SCID) determined the presence of circuitry in healthy humans. In a study with functional magnetic other mental disorders. The TENC subjects with any current resonance imaging (fMRI) (48), amygdala was activated during mental disorder were excluded. The PTSD subjects with current extinction learning, whereas vmPFC was activated during extinc- substance dependence were excluded, as were subjects who had tion recall. More recently, we reported that vmPFC and hip- used any psychotropic medication within 4 weeks before partic- pocampus are co-activated during extinction recall and that the ipation (1 year for neuroleptics). Type of trauma and current degree of such activation is positively correlated with psycho- comorbid disorders appear in Table 1, as do group mean age, physiologically measured extinction retention (49), as is vmPFC education, total CAPS scores, and age at first trauma exposure. thickness (50). In contrast, thickness and functional activation of the dACC, homologous to rat prelimbic cortex, are correlated Fear Conditioning, Extinction, and Testing Procedures with expression of conditioned fear in humans (37). Thus, there The previously described (37,49) 2-day experimental protocol is converging evidence in rodents and humans implicating the is summarized in Figure 1. All subjects selected a level of shock vmPFC and hippocampus in extinction recall and the dACC in they regarded as highly annoying but not painful, to be used in fear expression. Finally, the amygdala seems to be involved both the experiment. On day 1, during the Habituation phase, the in fear expression and extinction learning, which might lead to to-be extinguished CSϩ (CSϩE), unextinguished CSϩ (CSϩU), ambiguous predictions. and the CS that is never to be paired with the shock (CSϪ) (four The objective of the present study was to examine the of each) were presented in a counterbalanced manner within neurobiological basis of deficient extinction recall in PTSD with either the to-be conditioning or the to-be extinction context. a focus on the aforementioned brain regions. While in a 3-T fMRI During the Conditioning phase, two CSϩs (e.g., red and blue scanner, PTSD and trauma-exposed non-PTSD control (TENC) lights) were depicted within a photograph of a distinct room subjects underwent a 2-day Pavlovian fear conditioning and (conditioning context), and each was paired with the US at a extinction procedure that we have previously used in healthy partial reinforcement rate of 60%. A third CS (e.g., a yellow light) (39,51,52) and PTSD subjects (41). Skin conductance response was also depicted within the conditioning context but never (SCR), a commonly used measure in human fear studies (47,53) paired with the US (CSϪ). There were 8 CSϩEs,8CSϩUs, and 16 served as the dependent measure of conditioned responding. On CSϪ trials. When the shock US occurred, it followed the CSϩ day 1, subjects underwent fear conditioning to two pictures of offset without delay. The shock electrodes remained attached to differently colored lamps, followed by extinction for one of the subject’s fingers during all subsequent phases, and subjects them. Day 2 tested recall of the extinction that had been learned were instructed throughout the experiment (except during the the previous day by contrasting responses to the previously Habituation phase) that they “may or may not receive the electric extinguished and unextinguished stimuli. Several hypotheses were tested. First, we predicted impaired Table 1. Description of Demographics, Comorbidities, and Types of extinction recall as measured by SCR in PTSD. Not only would Trauma Exposure in Cohort Studied this represent a replication of our previous report (41); it would also extend this finding to PTSD caused by civilian trauma. PTSD TENC p Second, we predicted lesser vmPFC activation during extinction Demographics learning in the PTSD group. However, no directional predictions Age 33.6 (Ϯ 3.1) 30.4 (Ϯ 3.4) .8 were made for amygdala activation during extinction learning, Education 15 (Ϯ .53) 15.9 (Ϯ .74) .43 because of its ambiguous role described in the preceding text. Mean age at trauma exposure 17 (Ϯ 3.5) 22.4 (Ϯ 3.86) .30 Third, we predicted lesser vmPFC and hippocampal activations CAPS Score 66 (Ϯ 6.04) 10.5 (Ϯ 2.66) Ͻ.0001 and greater amygdala and dACC activations during (impaired) Current Comorbidities (n) extinction recall in the PTSD group. Fourth, we predicted that the Major depression 5 0 magnitude of extinction recall, indexed by percent extinction Panic disorder 2 0 retention, would be positively correlated with vmPFC and hip- Alcohol abuse 1 0 pocampal activations and inversely correlated with amygdala Other substance abuse 2 0 and dACC activations across all subjects. Eating disorders 2 0 Type of Trauma Exposure (n) Motor vehicle accidents 2 2 Methods and Materials Sexual assaults 8 2 Physical assaults 4 6 Subjects Childhood abuse 6 1 A total of 19 PTSD patients and 20 trauma-exposed non-PTSD Combat 3 0 control subjects were recruited from the community. After a full Witness to traumatic events 3 4 explanation of the study’s procedures, written informed consent The number of comorbid disorders and types of trauma shown might was obtained in accordance with the requirements of the Part- exceed the number of subjects, because a subject might have had more ners Healthcare System Human Research Committee. All subjects than one comorbid disorder or type of traumatic event. The Ϯ symbol completed participation in the 2-day fear conditioning and designates SEM. PTSD, posttraumatic stress disorder; TENC, trauma-ex- extinction paradigm. Three PTSD and five TENC subjects were posed non-PTSD control subject; CAPS, Clinician-Administered PTSD Scale. www.sobp.org/journal M.R. Milad et al. BIOL PSYCHIATRY 2009;66:1075–1082 1077 A. Day 1aD y 2 CdiiiConditioning EiiExtinction LLearning i RRecall ll Shock

Context Context with CS BB. 8 CS+E 16 CS+E 8 CS+E 8 CS+U 8 CS+U 16 CS- 16 CS- 16 CS-

Figure 1. Schematic of experimental protocol. (A) Pictures showing the visual contexts used in the experiment, within which conditioned stimuli (CS) were presented. In this example, pictures of an ofﬁce and a conference room represent conditioning and extinction (E) contexts, respectively, whereas the blue light represents the CSϩ that was paired with the shock and later extinguished (CSϩE). Extinction recall was conducted on day 2. (B) Schematic representation of the different phases of the experiment. A second CSϩ (red light, not shown) was presented during the conditioning phase but was not presented during the extinction learning phase (unextinguished, CSϩU). The same CSϩ was then presented during the recall phase. A third light (yellow, not shown) was presented throughout the different phases of the experiment and was never followed with the shock (CSϪ). The numbers of each stimulus type presented during the conditioning, extinction learning, and extinction recall are indicated. Gray shading represents the extinction context. Habituation phase is not shown. shock.” However, shocks were only delivered during the Con- Image Acquisition ditioning phase. After an approximate 1-min break, the Extinc- The image acquisition parameters were identical to those tion Learning phase began. During this phase, the CSϩE was previously used in our laboratory (49). Briefly, a Trio 3.0-Tesla depicted within a photograph of another distinct room (extinc- whole body high-speed imaging device equipped for echo tion context) and presented in the absence of the US, whereas planer imaging (Siemens Medical Systems, Iselin, New Jersey) the CSϩU was not presented. There were 16 CSϩE and 16 CSϪ with an 8-channel gradient head coil was used. Head movement trials. On day 2, during the Extinction Recall phase, 8 CSϩE, 8 was restricted with foam cushions. After an automated scout CSϩU, and 16 CSϪ trials were again presented depicted within image was obtained and shimming procedures were the extinction context. For each trial during the experiment, the performed, high-resolution three-dimensional magnetization context picture was presented for 9 sec: 3 sec alone followed by prepared rapid gradient echo sequences (repetition time/echo 6 sec in combination with the CSϩE, CSϩU, or CSϪ. The mean time [TR/TE]/flip angle ϭ 7.25 msec/3 msec/7°; 1 mm ϫ 1mmin intertrial interval was 15 sec (range: 12 sec–18 sec). All experi- plane ϫ 1.3 mm) were collected for spatial normalization and mental phases were conducted while blood-oxygen-level depen- positioning the subsequent scans. Scans with T1 (TR/TE/flip dent (BOLD) signal data were being acquired via fMRI. angle ϭ 8 sec/39 msec/90°) and T2 (TR/TE/flip angle ϭ 10 sec/48 msec/120°) sequences were used for registration of Psychophysiological Measures individual functional data. Functional MRI images (i.e., BOLD) As previously described (40,51,54), SCR for each CS trial was were acquired with gradient echo T2*-weighted sequence (TR/ calculated by subtracting the mean skin conductance level TE/flip angle ϭ 3 sec/30 msec/90°) (55). The T1, T2, and during the 2 sec before CS onset (during which the context alone gradient-echo functional images were collected in the same was being presented) from the highest skin conductance level plane (45 coronal oblique slices parallel to the anterior-posterior during the 6-sec CS duration. Thus, SCRs to the CSϩE, CSϩU, commissure line, tilted 30° anterior) with the same slice thickness and CSϪ reflected changes in skin conductance level beyond (3 mm ϫ 3mmϫ 3 mm). The same scanning procedure was any change in SC level produced by the context. The magnitude conducted on Day 2. of extinction retention (recall) was quantified as follows: each subject’s SCR to the first four CSϩ trials of the extinction Recall Functional MRI Data Analysis phase was divided by their largest SCR to a CSϩ trial during the Functional MRI data were analyzed with the Freesurfer Func- Conditioning phase and then multiplied by 100, yielding a tional Analysis Stream (http://surfer.nmr.mgh.harvard.edu). All percentage of maximal conditioned responding. This in turn was functional runs were motion-corrected with the Analysis of subtracted from 100% to yield an “extinction retention index.” Functional Images motion correction tool, spatially smoothed The purpose of calculating the extinction retention index was to (full-width-at-half-maximal ϭ 5 mm) with a three-dimensional normalize each subject’s SCR during extinction recall to that Gaussian filter, and intensity-normalized to the low-level base- exhibited during the conditioning phase. This index is important, line. Images were manually inspected for motion artifact, and because it adjusts the SCR during extinction recall for differences subjects with Ͼ2-mm total vector motion were excluded. Sub- in CR magnitude during acquisition. Unless otherwise specified, jects’ functional runs were then individually registered to their all data are presented as means Ϯ SEM. Analysis of variance anatomical volumes with FLIRT (Functional Magnetic Resonance (ANOVA) and Student t tests were performed to test for statisti- Imaging of the Brain Linear Image Registration Tool, http:// cally significant differences between means, with appropriate www.fmrib.ox.ac.uk/fsl/flirt/index.html), and the registrations Bonferroni corrections when required. were visually inspected for accuracy. Estimates of the stimulus

www.sobp.org/journal 1078 BIOL PSYCHIATRY 2009;66:1075–1082 M.R. Milad et al. effects at each voxel were made with an event-related design and A. by convolving the functional signal for each event with a 0.3 canonical hemodynamic response function. The analysis in- cluded a linear correction to account for low-frequency drift. 020.2 Statistical parametric maps were calculated according to a general linear model for the contrasts of interest across the time 0.1 window (56). The contrast used for the Stimulus factor during the SCR (sqrt) Extinction Learning phase was the last 12 CSϩE versus the last 12 0.0 CSϪ trials in the Extinction Learning phase. Note that no US was CS+ CS- CS+ CS- delivered during this phase. The contrast used for the Stimulus -0.101 PTSDTENC factor during Extinction Recall phase was the first four CSϩE versus the first four CSϩU trials. These specific trials were B. PTSD vs. TENC CS+ > CS- (late exncon learning) selected for three reasons. First, their use minimizes the con- found introduced by additional extinction learning that might take place during this phase and be especially reflected in responses to the latter trials. Second, electrophysiological data from rodents indicate that the vmPFC signals extinction recall only during the early portion of extinction recall (57). Third, we found that this contrast revealed the most robust activation of the vmPFC in our previous studies in healthy humans. CFPmv aladgymA Group ϫ Stimulus interactions (i.e., PTSD vs. TENC contrasts C. PTSD on the Stimulus contrast maps) were analyzed separately for the 0.4 0.2 TENC Extinction Learning and Recall phases. Functional regions of 0.1 interest (ROIs) were empirically defined as clusters of contiguous 0.2 0.1 voxels exceeding the a priori statistical threshold in the following 0.0 text. The BOLD signal values were extracted from these ROIs to 0.0

nal Change -0.202 calculate percent signal change. These values were then used for -0.1 regression analyses with the extinction retention index. Coordi- % Sign -0.4 -0.1 nates for the peak voxels in each region were specified in terms -0.6 -0.2 of the Talairach atlas (58) to allow comparison with results of previous studies. We focused our fMRI data analysis a priori on the vmPFC, amygdala, hippocampus, and dACC, areas within Figure 2. Responses during late extinction learning (last 12 of 16 trials). (A) Ͻ Skin conductance responses (SCRs) to the conditioned stimulus (CS) that which we employed a threshold of uncorrected, one-tailed p was previously paired with shock (CSϩ, dark shading) versus the CS that was .001. We used a more stringent threshold of p Ͻ .0001 for never paired with shock (CSϪ, light shading) in posttraumatic stress disor- activations and deactivations in remaining brain regions. der (PTSD) (red) versus trauma-exposed non-PTSD control (TENC) (black) subjects. (B) Group ϫ Stimulus interaction in ventromedial prefrontal cortex (vmPFC) and amygdala, Talairach coordinates: x ϭϪ15, y ϭ 34, z ϭϪ21 for Results vmPFC and x ϭ 25, y ϭϪ6, z ϭϪ24 for amygdala. Image is masked to show only the activation in this hypothesized brain region. Threshold for display- Psychophysiological Responses During Fear Conditioning ing the images is set at p ϭ .01. (C) Percent signal change extracted from the (Acquisition) amygdala and vmPFC functional region of interest shown in (B). An ANOVA revealed a significant Stimulus main effect (F ϭ 19.6, p Ͻ .001), with greater responses to the CSϩ (combined marginally significant, showing deactivation to the CSϩE relative across the first four to-be CSϩE and to-be CSϩU trials) than to to the CSϪ in PTSD relative to TENC subjects (t ϭϪ3.28, p ϭ the CSϪ (combined across the first four trials) in the PTSD (.28 Ϯ .0015; Figure 2B). Extracted percent BOLD signal changes from .07 ␮S vs. .07 Ϯ .05 ␮S) and in TENC (.15 Ϯ .04 ␮S vs. Ϫ.08 Ϯ the amygdala and vmPFC functional ROIs are shown in Figure .05 ␮S) groups. Importantly, there were no group differences in 2C. These data indicate that, during extinction learning, amyg- conditioning, as evidenced by the absence of a significant Group dala activation (to CSϩ relative to CSϪ) was observed in PTSD main effect (F ϭ 2.8, p ϭ .10) or Group ϫ Stimulus interaction subjects, and amygdala deactivation was observed in TENC (F ϭ .13, p ϭ .72). Functional MRI analysis was not conducted for subjects. The opposite pattern was observed in the vmPFC, (i.e., this phase. deactivation in PTSD and activation in TENC).

Psychophysiological and fMRI Responses During Extinction Psychophysiological and fMRI Responses During Extinction Learning Recall An ANOVA for the late extinction SCR data (last 12 CSϩE vs. An ANOVA for the early extinction recall SCR data (first four last 12 CSϪ trials) revealed no significant main effect of Stimulus CSϩE vs. first four CSϩU trials) revealed a significant Group ϫ (F ϭ 1.06, p ϭ .31) or Group (F ϭ 1.62, p ϭ .21) and no Stimulus interaction (F ϭ 4.99, p ϭ .03). Whereas the TENC significant Group ϫ Stimulus interaction (F ϭ 2.13, p ϭ .16), group exhibited smaller SCRs to the stimulus that had been suggesting that comparable extinction learning had been extinguished during the previous extinction learning phase achieved in both groups (Figure 2A). There was, regarding the compared with the stimulus that had not been extinguished fMRI data, a significant Group ϫ Stimulus interaction in right (.12 Ϯ .07 ␮S for CSϩE vs. .30 Ϯ .1 ␮S for CSϩU, F ϭ 5.14, p ϭ amygdala, which was more reactive to the CSϩE relative to the .03), the PTSD group did not (.40 Ϯ .11 ␮S for CSϩE vs. .37 Ϯ .10 CSϪ in PTSD relative to TENC subjects (t ϭ 3.71, p ϭ .00025; ␮S for CSϩU, F ϭ 1.1, p ϭ .3), suggesting impaired recall of Figure 2B). The Group ϫ Stimulus interaction in vmPFC was extinction memory in the PTSD group (Figure 3A). Consistent www.sobp.org/journal M.R. Milad et al. BIOL PSYCHIATRY 2009;66:1075–1082 1079

A. * 0.6 * 0.4 rt) q

0.2 SCR (s

0.0 CS+E CS+U CS+E CS+U

B. PTSD vs. TENC CS+E > CS+U

L-vmPFC R-vmPFC Hippocampus Dorsal ACC C. 0.6 0.6 0.6 0.6 PTSD 0.4 0.4 0.4 0.4 TENC

ange 020.2 020.2 020.2 020.2 0.0 0.0 0.0 0.0 -0.2 -0.2 -0.2 -0.2

% Signal Ch -0.4 -0.4 -0.4 -0.4

Figure 3. Responses during early extinction recall (ﬁrst four trials). (A) The SCRs to the stimulus that was previously extinguished on day 1 (CSϩE, dark shading) versus the CS that was not extinguished on day 1 (CSϩU, light shading) in PTSD (red) vs. TENC (black) subjects. (B) Group ϫ Stimulus interactions. Talairach coordinates: left (L)-vmPFC: x ϭϪ10, y ϭ 43, z ϭϪ11; right (R)-vmPFC: x ϭ 2, y ϭ 45, z ϭϪ12; hippocampus, x ϭ 32, y ϭϪ9, z ϭϪ27; dorsal anterior cingulate cortex (dACC): x ϭϪ2, y ϭ 37, z ϭ 18. All images were masked to only show activations/deactivations in hypothesized brain regions. Threshold for displaying the images is set at p ϭ .01. (C) Percent signal change extracted from the functional regions of interest shown in (B).*p Ͻ .05. Abbreviations as in Figure 2. with this, the extinction retention index was significantly smaller analysis revealed that the Group ϫ Stimulus interaction remained in the PTSD than the TENC group (46% vs. 85%, t ϭ 2.9, p Ͻ .01). significant for the SCR data during extinction recall (F ϭ 5.39, p ϭ Moreover, within the PTSD group, total CAPS score was nega- .02). Moreover, the percent extinction retention between the two tively correlated with extinction retention index (r ϭϪ.71, p ϭ groups remained statistically significant (52% for PTSD vs. 85% .01). With respect to the fMRI data during extinction recall, the for TENC, t ϭ 2.18, p ϭ .037). Regarding the fMRI data, reanalysis same contrast was used (first four CSϩE vs. first four CSϩU of the main contrast during extinction recall (CSϩE vs. CSϩU) trials). There were significant Group ϫ Stimulus interactions in revealed that the deactivation in the bilateral vmPFC in the PTSD right hippocampus (t ϭ 4.27, p ϭ .0001), right vmPFC (t ϭ 3.54, relative to the TENC group was marginally significant (t ϭ 3.25, p ϭ .0007), left vmPFC (t ϭ 3.41, p Ͻ .001), and left dACC (t ϭ p ϭ .0015 for both right and left vmPFC), whereas the hippocam- 3.41, p Ͻ .001) (Figure 3B). Extracted percent BOLD signal pal difference between groups remained significant (t ϭ 4.05, changes from these functional ROIs are shown in Figure 3C. The p ϭ .00025). The increased activation in the dACC in the PTSD TENC subjects showed activation in left and right vmPFC and relative to the TENC group became more significant (t ϭ 4.20, hippocampus and deactivation in dACC, in response to the CSϩE p ϭ .00015). The reduced significance level regarding the vmPFC relative to the CSϩU. The PTSD subjects showed the opposite activation is most likely due to reduced power. Thus, this patterns. subanalysis revealed that comorbidity in the PTSD sample ana- To test for relationships between activations or deactivations lyzed in this study is unlikely to have accounted for the differ- in these brain regions during extinction recall and extinction ences observed between groups with regard to either the psy- memory, we conducted analyses correlating percent BOLD chophysiological or the fMRI data. signal changes with extinction retention index across all subjects (Figure 4). These analyses revealed significant positive correla- tions between activation in vmPFC (bilaterally) and hippocam- Discussion pus and extinction retention as well as a trend toward a negative correlation between dACC activation and extinction retention. The psychophysiological and fMRI data obtained in the TENC Activations/deactivations outside the a priori hypothesized group show intact fear extinction memory (or recall), manifest in brain regions are shown in Table 2. lower SCRs to a previously extinguished compared with a previously unextinguished CS that is associated with vmPFC and Subanalyses with Comorbidity-Free PTSD Subjects hippocampal activation during extinction recall, thereby replicat- The key results were subjected to reanalysis excluding six ing our previous report (49). In contrast, the psychophysiological PTSD subjects with current comorbid Axis I disorders. This data obtained in the PTSD group show impaired extinction

www.sobp.org/journal 1080 BIOL PSYCHIATRY 2009;66:1075–1082 M.R. Milad et al.

CFPmv-L supmacoppiH

150 r = 0.49* 150 r = 0.45* PTSD p = 0.005 P = 0.01 100 100 TENC 50 50 0 0 xt. Retenon xt. Retenon

-50 Ex -50 % % Ex -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 % Signal Change % Signal Change

CFPmv-R lasroD CCA r = 0.45* r = -0.33 150 150 PP001 = 0.01 pp0055 = 0.055 100 100 50 50

0 0

% Ext. Retenon -50 % Ext. Retenon -50 -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 % Signal Change % Signal Change

Figure 4. Regression plots between percent extinction (Ext.) retention and percent blood-oxygen-level dependent signal change during extinction recall extracted from the functional regions of interest shown in Figure 2, collapsed across groups. All p values listed in the ﬁgures are below the Bonferroni correction threshold. Abbreviations as in Figures 2 and3. retention, manifest in no difference between SCRs to the extin- learning in the PTSD group in the absence of vmPFC activation guished and unextinguished CSs, replicating another of our is consistent with animal studies. For example, it has been previous reports (41). In addition, the present data suggest that previously shown that lesions or pharmacological manipulations this deficient extinction retention in PTSD might be the result of of the vmPFC do not interfere with extinction learning per se dysfunctional responding in brain regions previously reported to (28). Rather, single neurons recorded from this brain region be implicated in the recall of fear extinction in healthy subjects. increase their neural activity to the extinguished CSϩ only during Specifically, we found less activation in hippocampus and bilat- extinction recall (57). Thus, the data gathered from the current eral vmPFC but more activation in dACC during extinction recall study provide a translational link between rodent and human data in PTSD compared with TENC subjects. The amount of extinction indicating that vmPFC function is not necessary for initial extinction retention across all subjects was positively correlated with acti- learning but is critical for extinction recall. In other words, the vation in both vmPFC and hippocampus and nearly significantly present data suggest that dysfunctional brain activation in the PTSD negatively correlated with activation in dACC, thereby replicating group (i.e., greater activity in amygdala, and lesser activity in vmPFC prior fMRI results in an independent sample of healthy subjects compared with the TENC group) during extinction learning might and extending them to PTSD (48,49,59). contribute to PTSD patients’ failure to consolidate extinction mem- The greater activation in the amygdala in PTSD patients ory. The present data further suggest that failure to activate vmPFC during extinction learning replicates a recent report (16). How- and hippocampus during recall contribute to deficient expression of ever, despite their greater amygdala activation and their lesser extinction memory in PTSD. As noted in the introduction, PTSD vmPFC activation, the PTSD group displayed extinction learning patients’ failure to activate these brain regions has also been found that was comparable to the TENC group. Normal extinction in other neuroimaging tasks.

Table 2. Signiﬁcant Activations in Regions Outside the A Priori Regions of Interest

Area of Activation Talairach Coordinates p

Late Extinction Learning Contrast: PTSD Ͼ TENC, Late CSϩE vs. CSϪ Regions outside the a priori areas Superior temporal cortex (R) 61, Ϫ11, Ϫ4 3.1 ϫ 10Ϫ5 Superior temporal cortex (R) 64, Ϫ9, 0 4.3 ϫ 10Ϫ5 Superior temporal cortex (L) Ϫ51, 5, Ϫ8 7.9 ϫ 10Ϫ5 Extinction Recall Contrast: PTSD Ͼ TENC, Early CSϩE vs. CSϩU Regions outside the a priori areas Cerebellar cortex (R) 22, Ϫ46, Ϫ14 4.0 ϫ 10Ϫ6 Cerebellar cortex (R 9, Ϫ59, Ϫ40 1.2 ϫ 10Ϫ5 Cerebellar cortex (R) 40, Ϫ67, Ϫ34 2.8 ϫ 10Ϫ5 Medial parietal cortex (R) Ϫ4, Ϫ17, 61 3.0 ϫ 10Ϫ5 Occipital cortex (L) Ϫ48, Ϫ69, Ϫ1 4.7 ϫ 10Ϫ5 Threshold for peak voxel p Ͻ 10Ϫ4, two-tailed, uncorrected. CS, conditioned stimulus; E, extinction; PTSD, posttraumatic stress disorder; TENC, trauma-exposed non-PTSD control subject; U, unextinguished. www.sobp.org/journal M.R. Milad et al. BIOL PSYCHIATRY 2009;66:1075–1082 1081

The dACC has traditionally been implicated in conflict mon- dance of the Advisory Board Meeting on the Anatomy and itoring, attention, and pain (60–62). One caveat when compar- Neuroscience of anxiety and depression. Note that no direct ing the results of those studies and the data presented in the conflict is anticipated. All relationships are listed for full disclo- present study is that the term “dACC” has been used to refer to a sure. The remaining authors reported no biomedical financial broad area of the anterior cingulate. In a recent meta-analysis, interests or potential conflicts of interest. Vogt et al. (60,61) identified a subregion of the dACC (termed the anterior midcingulate [aMCC]) that was specifically activated by 1. Liberzon I, Sripada CS (2008): The functional neuroanatomy of PTSD: A fear-inducing stimuli. Importantly, the dACC region that showed critical review. Prog Brain Res 167:151–169. activation during extinction recall in our PTSD subjects seems to 2. Etkin A, Wager TD (2007): Functional neuroimaging of anxiety: A meta- overlap with aMCC. Moreover, we have previously shown that analysis of emotional processing in PTSD, social anxiety disorder, and dACC thickness and function are positively correlated with specific phobia. Am J Psychiatry 164:1476–1488. 3. Rauch SL, Shin LM, Phelps EA (2006): Neurocircuitry models of posttrau- conditioned responding during fear acquisition in healthy con- matic stress disorder and extinction: Human neuroimaging research— trol subjects, suggesting that this brain region might be involved past, present, and future. Biol Psychiatry 60:376–382. in promoting the fear response (37). A recent neuroimaging 4. Milad MR, Rauch SL (2007): The role of the orbitofrontal cortex in anxiety study reported increased activation of the dACC region in PTSD disorders. Ann N Y Acad Sci 1121:546–561. patients (11). All of these findings support a role for the dACC in 5. Shin LM, Orr SP, Carson MA, Rauch SL, Macklin ML, Lasko NB, et al. (2004): the pathological expression of conditioned fear in PTSD. Regional cerebral blood flow in the amygdala and medial prefrontal cortex during traumatic imagery in male and female Vietnam veterans We observed, in addition to the a priori ROIs, increased with PTSD. Arch Gen Psychiatry 61:168–176. cerebellar activation in PTSD patients relative to control subjects 6. Shin LM, Shin PS, Heckers S, Krangel TS, Macklin ML, Orr SP, et al. (2004): during extinction recall (Table 1). The meaning of this finding is Hippocampal function in posttraumatic stress disorder. Hippocampus unclear, given that we previously observed cerebellar activation 14:292–300. during extinction recall in a healthy cohort (49). In addition to 7. Rauch SL, Whalen PJ, Shin LM, McInerney SC, Macklin ML, Lasko NB, et al. the well-documented role of this brain region in movement and (2000): Exaggerated amygdala response to masked facial stimuli in motor coordination, the cerebellum has been reported to be posttraumatic stress disorder: A functional MRI study. Biol Psychiatry 47:769–776. involved in the processing of fear memories (63,64) and in 8. Bremner JD, Vythilingam M, Vermetten E, Southwick SM, McGlashan T, extinction of eye-blink conditioning (65). Further studies are Staib LH, et al. (2003): Neural correlates of declarative memory for emo- needed to clarify its role in emotional learning and memory, tionally valenced words in women with posttraumatic stress disorder including fear extinction, in general, and in PTSD. related to early childhood sexual abuse. Biol Psychiatry 53:879–889. It has been hypothesized that fear extinction and its retention 9. Pissiota A, Frans O, Fernandez M, Von KL, Fischer H, Fredrikson M (2002): are deficient in PTSD due to failure to activate brain extinction Neurofunctional correlates of posttraumatic stress disorder: A PET symptom provocation study. Eur Arch Psychiatry Clin Neurosci 252: circuitry, including hippocampus and vmPFC (31,42). With 68–75. positron emission tomography, Bremner et al. (16) were the first 10. Vermetten E, Schmahl C, Southwick SM, Bremner JD (2007): Positron to examine fear conditioning and extinction learning in PTSD. tomographic emission study of olfactory induced emotional recall in The authors reported increased amygdala and decreased vmPFC veterans with and without combat-related posttraumatic stress disor- activity in PTSD relative to control subjects, which is consistent der. Psychopharmacol Bull 40:8–30. with this hypothesis. In the current study, the link here between 11. Shin LM, Bush G, Whalen PJ, Handwerger K, Cannistraro PA, Wright CI, et al. (2007): Dorsal anterior cingulate function in posttraumatic stress deficient, psychophysiologically measured extinction recall in disorder. J Trauma Stress 20:701–712. PTSD and failure to activate vmPFC and hippocampus during 12. Shin LM, Whalen PJ, Pitman RK, Bush G, Macklin ML, Lasko NB, et al. extinction recall provide direct data in support of this model. The (2001): An fMRI study of anterior cingulate function in posttraumatic present results also provide neurobiological evidence that the stress disorder. Biol Psychiatry 50:932–942. pathologically elevated and persistent conditioned fear clinically 13. Bryant RA, Felmingham KL, Kemp AH, Barton M, Peduto AS, Rennie C, et observed in PTSD is at least in part due to failure to activate al. (2005): Neural networks of information processing in posttraumatic vmPFC and hippocampus as well as to hyperactivation of dACC stress disorder: A functional magnetic resonance imaging study. Biol Psychiatry 58:111–118. and amygdala. 14. Shin LM, McNally RJ, Kosslyn SM, Thompson WL, Rauch SL, Alpert NM, et al. (1999): Regional cerebral blood flow during script-driven imagery in The work was supported in all aspects, including design, data childhood sexual abuse-related PTSD: A PET investigation. Am J Psychi- atry 156:575–584. collection, and preparation, by a grant from the National 15. Bremner JD, Narayan M, Staib LH, Southwick SM, McGlashan T, Charney Institute of Mental Health (1R21MH072156-1) to SLR and a DS (1999): Neural correlates of memories of childhood sexual abuse in Young Investigator Award from the National Alliance for Re- women with and without posttraumatic stress disorder. Am J Psychiatry search on Schizophrenia and Depression to MRM. We would like 156:1787–1795. to thank Dr. Clas Linnman for helpful comments on the manu- 16. Bremner JD, Vermetten E, Schmahl C, Vaccarino V, Vythilingam M, Afzal script. N, et al. (2005): Positron emission tomographic imaging of neural corre- Scott Rauch’s financial disclosures are as follows: he received lates of a fear acquisition and extinction paradigm in women with childhood sexual-abuse-related post-traumatic stress disorder. Psychol funded research through Massachusetts General Hospital (MGH) Med 35:791–806. for Brain Stimulation Therapy from Medtronics; funded research 17. Britton JC, Phan KL, Taylor SF, Fig LM, Liberzon I (2005): Corticolimbic through MGH for vegal nerve stimulation from Cyberonics; and blood flow in posttraumatic stress disorder during script-driven imag- funded research through MGH on anxiolytic action from Cepha- ery. Biol Psychiatry 57:832–840. lon. He also received honoraria from Novartis for consultation 18. Phan KL, Britton JC, Taylor SF, Fig LM, Liberzon I (2006): Corticolimbic on emerging treatments; Neurogen for his participation as a blood flow during nontraumatic emotional processing in posttraumatic stress disorder. Arch Gen Psychiatry 63:184–192. consultant on emerging trends in anxiety associated with insom- 19. Shin LM, Wright CI, Cannistraro PA, Wedig MM, McMullin K, Martis B, et nia; Sepracor for his consultation on fear/conditioning/extinc- al. (2005): A functional magnetic resonance imaging study of amygdala tion and from Primedia for his participation in developing a and medial prefrontal cortex responses to overtly presented fearful continuing education activity; and Medtronics for his atten- faces in posttraumatic stress disorder. Arch Gen Psychiatry 62:273–281.

www.sobp.org/journal 1082 BIOL PSYCHIATRY 2009;66:1075–1082 M.R. Milad et al.

20. Lanius RA, Williamson PC, Hopper J, Densmore M, Boksman K, Gupta 42. Quirk GJ, Mueller D (2008): Neural mechanisms of extinction learning MA, et al. (2003): Recall of emotional states in posttraumatic stress and retrieval. Neuropsychopharmacology 33:56–72. disorder: An fMRI investigation. Biol Psychiatry 53:204–210. 43. Likhtik E, Popa D, Apergis-Schoute J, Fidacaro GA, Pare D (2008): Amyg- 21. Kim MJ, Chey J, Chung A, Bae S, Khang H, Ham B, et al. (2008): Diminished dala intercalated neurons are required for expression of fear extinction. rostral anterior cingulate activity in response to threat-related events in Nature 454:642–645. posttraumatic stress disorder. J Psychiatr Res 42:268–277. 44. Herry C, Ciocchi S, Senn V, Demmou L, Muller C, Luthi A (2008): Switch- 22. Lanius RA, Williamson PC, Boksman K, Densmore M, Gupta M, Neufeld ing on and off fear by distinct neuronal circuits. Nature 454:600–606. RW, et al. (2002): Brain activation during script-driven imagery induced 45. Vidal-Gonzalez I, Vidal-Gonzalez B, Rauch SL, Quirk GJ (2006): Micro- dissociative responses in PTSD: A functional magnetic resonance imag- stimulation reveals opposing inﬂuences of prelimbic and infralimbic ing investigation. Biol Psychiatry 52:305–311. cortex on the expression of conditioned fear. Learn Mem 13:728–733. 23. Zubieta JK, Chinitz JA, Lombardi U, Fig LM, Cameron OG, Liberzon I 46. Corcoran KA, Quirk GJ (2007): Activity in prelimbic cortex is necessary for (1999): Medial frontal cortex involvement in PTSD symptoms: A SPECT the expression of learned, but not innate, fears. J Neurosci 27:840–844. study. J Psychiatr Res 33:259–264. 47. Burgos-Robles A, Vidal-Gonzalez I, Quirk GJ (2009): Sustained condi- 24. Liberzon I, Taylor SF, Amdur R, Jung TD, Chamberlain KR, Minoshima S, tioned responses in prelimbic prefrontal neurons are correlated with et al. (1999): Brain activation in PTSD in response to trauma-related fear expression and extinction failure. J Neurosci 29:8474–8482. stimuli. Biol Psychiatry 45:817–826. 48. Phelps EA, Delgado MR, Nearing KI, Ledoux JE (2004): Extinction learn- 25. Quirk GJ, Russo GK, Barron JL, Lebron K (2000): The role of ventromedial ing in humans: Role of the amygdala and vmPFC. Neuron 43:897–905. prefrontal cortex in the recovery of extinguished fear. J Neurosci 20: 49. Milad MR, Wright CI, Orr SP, Pitman RK, Quirk GJ, Rauch SL (2007): Recall 6225–6231. of fear extinction in humans activates the ventromedial prefrontal cor- 26. Rothbaum BO, Foa EB (2002): Post-traumatic stress disorder and sleep. tex and hippocampus in concert. Biol Psychiatry 62:446–454. N Engl J Med 346:1334–1335. 50. Milad MR, Quinn BT, Pitman RK, Orr SP, Fischl B, Rauch SL (2005): Thick- 27. Rothbaum BO, Davis M (2003): Applying learning principles to the treat- ness of ventromedial prefrontal cortex in humans is correlated with ment of post-trauma reactions. Ann N Y Acad Sci 1008:112–121. extinction memory. Proc Natl Acad SciUSA102:10706–10711. 28. Corcoran KA, Quirk GJ (2007): Recalling safety: Cooperative functions of 51. Milad MR, Orr SP, Pitman RK, Rauch SL (2005): Context modulation of the ventromedial prefrontal cortex and the hippocampus in extinction. memory for fear extinction in humans. Psychophysiology 42:456–464. CNS Spectr 12:200–206. 52. Rauch SL, Milad MR, Orr SP, Quinn BT, Fischl B, Pitman RK (2005): Orbito- 29. Sotres-Bayon F, Cain CK, Ledoux JE (2006): Brain mechanisms of fear frontal thickness, retention of fear extinction, and extraversion. Neuro- extinction: Historical perspectives on the contribution of prefrontal cor- report 16:1909–1912. tex. Biol Psychiatry 60:329–336. 53. Hermans D, Craske MG, Mineka S, Lovibond PF (2006): Extinction in 30. Ledoux JE (2000): Emotion circuits in the brain. Annu Rev Neurosci 23: human fear conditioning. Biol Psychiatry 60:361–368. 155–184. 54. Pitman RK, Orr SP (1986): Test of the conditioning model of neurosis: 31. Myers KM, Davis M (2007): Mechanisms of fear extinction. Mol Psychiatry Differential aversive conditioning of angry and neutral facial expres- 12:120–150. sions in anxiety disorder patients. J Abnorm Psychol 95:208–213. 32. Fredrikson M, Hugdahl K, Ohman A (1976): Electrodermal conditioning 55. Kwong KK, Belliveau JW, Chesler DA, Goldberg IE, Weisskoff RM, Ponce- to potentially phobic stimuli in male and female subjects. Biol Psychol let BP, et al. (1992): Dynamic magnetic resonance imaging of human 4:305–314. brain activity during primary sensory stimulation. Proc Natl Acad SciUSA 33. Soares JJ, Ohman A (1993): Backward masking and skin conductance 89:5675–5679. responses after conditioning to nonfeared but fear-relevant stimuli in 56. Dale AM, Fischl B, Sereno MI (1999): Cortical surface-based analysis. I. fearful subjects. Psychophysiology 30:460–466. Segmentation and surface reconstruction. Neuroimage 9:179–194. 34. Orr SP, Metzger LJ, Pitman RK (2002): Psychophysiology of post-trau- 57. Milad MR, Quirk GJ (2002): Neurons in medial prefrontal cortex signal matic stress disorder. Psychiatr Clin North Am 25:271–293. memory for fear extinction. Nature 420:70–74. 35. Labar KS, Ledoux JE, Spencer DD, Phelps EA (1995): Impaired fear con- 58. Talairach J, Tournoux P (1998): Co-Planar Stereotaxic Atlas of the Human ditioning following unilateral temporal lobectomy in humans. J Neuro- Brain. 3-D Proportional System: An Approach to Cerebral Imaging. New sci 15:6846–6855. York: Thieme Publishers. 36. Knight DC, Smith CN, Cheng DT, Stein EA, Helmstetter FJ (2004): Amyg- 59. Kalisch R, Korenfeld E, Stephan KE, Weiskopf N, Seymour B, Dolan RJ dala and hippocampal activity during acquisition and extinction of (2006): Context-dependent human extinction memory is mediated by a human fear conditioning. Cogn Affect Behav Neurosci 4:317–325. ventromedial prefrontal and hippocampal network. J Neurosci 26:9503– 37. Milad MR, Quirk GJ, Pitman RK, Orr SP, Fischl B, Rauch SL (2007): A role for 9511. the human dorsal anterior cingulate cortex in fear expression Biol Psy- 60. Vogt BA (2005): Pain and emotion interactions in subregions of the chiatry 62:1191–1194. cingulate gyrus. Nat Rev Neurosci 6:533–544. 38. Orr SP, Pitman RK, Lasko NB, Herz LR (1993): Psychophysiological assess- 61. Vogt BA, Berger GR, Derbyshire SW (2003): Structural and functional ment of posttraumatic stress disorder imagery in World War II and dichotomy of human midcingulate cortex. Eur J Neurosci 18:3134–3144. Korean combat veterans. J Abnorm Psychol 102:152–159. 62. Bush G, Vogt BA, Holmes J, Dale AM, Greve D, Jenike MA, et al. (2002): 39. Milad MR, Goldstein JM, Orr SP, Wedig MM, Klibanski A, Pitman RK, et al. Dorsal anterior cingulate cortex: A role in reward-based decision mak- (2006): Fear conditioning and extinction: Inﬂuence of sex and menstrual ing. Proc Natl Acad SciUSA99:523–528. cycle in healthy humans. Behav Neurosci 120:1196–1203. 63. Sacchetti B, Scelfo B, Strata P (2005): The cerebellum: Synaptic changes 40. Orr SP, Metzger LJ, Lasko NB, Macklin ML, Peri T, Pitman RK (2000): De and fear conditioning. Neuroscientist 11:217–227. novo conditioning in trauma-exposed individuals with and without 64. Sacchetti B, Sacco T, Strata P (2007): Reversible inactivation of amygdala posttraumatic stress disorder. J Abnorm Psychol 109:290–298. and cerebellum but not perirhinal cortex impairs reactivated fear mem- 41. Milad MR, Orr SP, Lasko NB, Chang Y, Rauch SL, Pitman RK (2008): ories. Eur J Neurosci 25:2875–2884. Presence and acquired origin of reduced recall for fear extinction in 65. Robleto K, Thompson RF (2008): Extinction of a classically conditioned PTSD: Results of a twin study. J Psychiatr Res 42:515–520. response: Red nucleus and interpositus. J Neurosci 28:2651–2658.