Cingulum White Matter Microstructure and Treatment
Success in Veterans with Posstraumatic Stress Disorder –
A Tractography Study
L. van Soolingen (10209530)
MSc in Brain and Cognitive Sciences, University of Amsterdam, Cognitive Neuroscience
Keywords: Posttraumatic Stress Disorder (PTSD); veterans; trauma; trauma-‐focused treatment; DTI; tractography; cingulum
Supervisor: Mitzy Kennis Co-‐assessor: Elske Salemink
Abstract
Posttraumatic stress disorder (PTSD) is a trauma-‐ and stressor-‐related psychiatric condition that has been associated with whiter matter alterations in the cingulum. However, the neurobiological effects on the cingulum of available trauma-‐focused treatment for PTSD are still not fully understood. The aim of this study is to investigate the effect and location of the effect of trauma-‐focused treatment in PTSD on the white matter of the cingulum by using diffusion tensor imaging (DTI). We used a longitudinal design in which PTSD patients received treatment as usual. Pre-‐ and post-‐treatment, DTI scans were obtained from veterans with (n=41) and without PTSD (n=20), on which tractography was applied to reconstruct the cingulum. Each cingulum bundle was divided into approximately 2mm-‐long segments. Twenty-‐three patients responded to treatment (>30% CAPS reduction), and 18 patients did not respond (<30% CAPS reduction). A linear mixed-‐effect model was used to test whether trauma-‐focused treatment as a function of time has a different effect on FA values for Responders and Non-‐Responders compared to combat control-‐group. A trend towards an interaction was found for group-‐by-‐time for FA values in the left parahippocampal cingulum, which was driven by changes in the Responders over time. Our results indicate that increased integrity of the white matter microstructure in the left parahippocampal cingulum is a possible effect of trauma-‐focused treatment in Responders that develops during treatment. These findings highlight the importance of considering the spatial distribution of effects in tractography studies; in previous studies the exact localization of the effect of trauma-‐focused treatment in PTSD had not been studied. In conclusion, tractography studies can shed more light on the neurobiological mechanisms of available trauma-‐focused treatments.
Introduction
Posttraumatic stress disorder (PTSD) is a trauma-‐ and stressor-‐related psychiatric condition that can develop after trauma-‐exposure (DSM-‐5, American Psychiatric Association, 2013). In Dutch Military personnel deployed to Afghanistan, prevalence of PTSD has been reported to be 5.6-‐8.9% after deployment (Reijnen et al., 2015). Common symptoms of PTSD are re-‐experiencing of the traumatic event (e.g. nightmares, flashbacks), avoidance of reminders of the traumatic event and emotional numbing, and increased arousal (e.g., hyper vigilance, insomnia). Moreover, dysfunction of coping, affective reactions, beliefs, attention, memory, cognitive-‐affective reactions and social support has been identified as well in patients with PTSD (DSM-‐IV, American Psychiatric Association, 1994). All these symptoms cause impairment in daily functioning (Hoge et al., 2006; Van Ameringen et al., 2008). To help these PTSD patients, it is of great importance to understand the neurobiology of this disorder. Neuroimaging techniques have been used to unravel the neurobiology of PTSD. Previous neuroimaging studies have revealed structural and functional alterations in the amygdala, hippocampus, anterior cingulate cortex (ACC) and ventromedial prefrontal cortex (vmPFC) (e.g., Brunetti et al., 2010; Pitman
et al., 2012; Shin & Liberzon, 2010). Unravelling the neurobiology of PTSD, such as identifying
biomarkers that could predict treatment response, is of great importance to improve treatment of PTSD (Garfinkel and Liberzon, 2009).
Effective treatment for PTSD can be Eye Movement Desensitization and Reprocessing (EMDR) and/or Cognitive Behavior Therapy (CBT) with exposure (Harvey et al., 2003; Bradley et al., 2005; Ponniah & Hollon 2009). One of the key points of EMDR and CBT is exposure of the patient to trauma-‐ related stimuli, which weakens the trauma-‐related memories and leads to extinction of the PTSD patients fear and symptoms (Rothbaum & Davis, 2003; Vermetten et al., 2012). However, despite the
relative success of trauma-‐focused treatment, 27 – 83% of the PTSD patients do not respond to EMDR (i.e., do not show reduction of symptoms) (Bradley et al., 2005). Investigating the different brain alterations between Responders and Non-‐Responders as a result of treatment will illuminate the biological mechanisms of available trauma-‐focused treatments and increase the possibility to match treatments to different patients (Etkin et al., 2005; Garfinkel and Liberzon, 2009).
To date, a few structural and functional neuroimaging studies showed an associating between pre-‐treatment structure and activity of the cingulate cortex and responsiveness of trauma-‐focused treatment. Poor improvement in PTSD patients after CBT was associated with pre-‐treatment greater activity in the ventral anterior cingulate in response to fearful and neutral facial expressions (Bryant et
al., 2008a), greater activity in the parahippocampal cingulate when completing a Go/No-‐Go task
(Falconer et al., 2013) and larger rostral anterior cingulate cortex volume in general (Bryant et al., 2008b). PTSD patients who did not respond to EMDR showed lower grey matter density in bilateral dorsal cingulate cortex while they were exposed to trauma-‐related pictures (Nardo et al., 2010). These results suggest that there are possible potential differences in the brain of PTSD patients who respond to treatment compared to those that do not respond. Moreover, these results suggest the possibility of using biomarkers that could predict treatment response.
These studies only investigated biological differences between Responders and Non-‐ Responders prior to the trauma-‐focused treatment and did not control for trauma exposure. Two longitudinal studies where the PTSD patients underwent scanning prior and after treatment showed that poor improvement to exposure therapy was associated with higher activity in the rostral anterior cingulate during fear processing over time (Felmingham et al., 2007) and higher activity in the dorsal anterior cingulate in Non-‐Responders in response to negative pictures over time (van Rooij et al.,
2015). Overall, there seems to be some evidence to indicate that poor improvement to trauma-‐focused treatment is associated with higher activity in different segments of the cingulate cortex.
However, all the studies above did not assess possible alterations in white matter over time after trauma-‐focused treatment, despite previous research that showed differences in integrity of the cingulum in PTSD patients vs. controls (Abe et al., 2006; Kim et al., 2006; Schuff et al., 2011; Fani et al., 2012; Zhang et al., 2012; Sanjuan et al., 2013). The cingulum bundle interconnects the prefrontal cortex with the entire temporal limbic complex (Wakana et al., 2004). The prefrontal cortex is involved in extinction (Shin & Liberzon 2010), which is one of the key processes in EMDR and CBT (Vermetten et
al., 2012). The temporal limbic complex contains the amygdala and hippocampus. These regions are
associated with generating and maintenance of emotional responses (Davis 1992), and encoding and retrieval of episodic/autobiographical memories (Eichenbaum 2000). These processes are impaired in PTSD and may also be implicated in the responsiveness to EMDR treatment (Shapiro, 1999; Corrigan, 2004). Because the cingulum bundle connects the prefrontal cortex with these regions, it can be suggested that the white matter track of the cingulum bundle plays an important role in the development of PTSD. Therefore, it would be of great interest to assess the effect of trauma-‐focused treatment on the microstructure of white matter in PTSD.
In recent years, diffusion tensor imaging (DTI) has emerged as a non-‐invasive, in vivo method for characterizing microstructural changes of white matter connections (Jones & Leemans 2010). DTI measures the diffusion of water molecules, which is anisotropic in white matter, because the movement of the water molecules is constraint by physical boundaries (i.e., the myelin around the axon repels water molecules). Therefore, the water molecules diffuse more freely along the main orientation of the white matter than across them (Moseley et al., 1990). A common DTI metric that
quantifies the diffusion is Fractional Anisotropy (FA), which is an index of the microstructure of white matter and provides information of myelination, axonal density and coherence of fiber orientation (Le Bihan et al., 2001). FA is highest in white matter (nearing 1) and lower (nearing 0) in grey matter and cerebrospinal fluid (Smith et al., 2006).
A few studies showed differences in white matter microstructure in different segments of the cingulum in PTSD. Previous studies have reported a decreased FA in PTSD in anterior cingulum (Kim et
al., 2006; Schuff et al., 2011), in the posterior cingulum (Fani et al., 2012), and in the dorsal cingulum
(Sanjuan et al., 2013). Although increased FA in PTSD has also been found, as in the left anterior cingulum (Abe et al., 2006), and in the parahippocampal cingulum (Zhang et al., 2012). A longitudinal study by Zhang et al. (2012) showed increased FA in the left parahippocampal cingulum in PTSD over time. So far, only one study has investigated the effect of trauma-‐focused treatment for patients with PTSD on the white matter microstructure of the cingulum bundle, where an increase in FA in the dorsal cingulum in PTSD over time was found over the course of unsuccessful trauma-‐focused treatment (Kennis et al., 2015). Moreover, Kennis et al. (2015) found an interaction between group and time for FA in the left hippocampal cingulum and left dorsal cingulum. These previously published studies on the localization of the effect of trauma-‐focused treatment in PTSD on the white matter of the cingulum are not consistent. Because of all these findings in different segments of the cingulum, the neurobiological effects of available trauma-‐focused treatment for PTSD are still not fully understood. The aim of this study is to investigate the effect and localization of the effect of trauma-‐focused treatment for patients with PTSD on the white matter microstructure of the cingulum. To examine this, we will compare the FA values of segments of the cingulum in PTSD patients prior and after trauma-‐ focused treatment. In this way, we are able to localize the effect of trauma-‐focused treatment for
patients with PTSD on the white matter microstructure of the cingulum. We will compare 41 PTSD patients with a trauma-‐exposed control group (combat control group), consisting of 20 veterans who did not develop PTSD. MRI-‐scans were obtained at two time-‐points for the PTSD group and combat control group: Prior to trauma-‐focused treatment (T0) and after six to nine months (T6). At both time points interviews and questionnaires (to measure relevant covariates, early trauma etc.) were collected. PTSD patients received trauma-‐focused treatment between T0 and T6. We will systematically investigate segments of the cingulum bundle, by subdividing the cingulum in approximately 2mm segments. To assess the microstructure of the white matter of these segments, FA will be obtained prior and after trauma-‐focused treatment. By mean of this, changes in the microstructure of the white matter of the segments after treatment can be detected. It is expected that PTSD patients will have pre-‐treatment lower FA values in the cingulum compared to controls, and that these FA values will increase, so be more similar to trauma-‐controls after successful treatment.
Methods & Materials
Participants
In total, 41 male veterans with PTSD and 20 male veterans without PTSD (combat-‐control) were included in this study. All PTSD patients were recruited from one of the outpatient clinic of the Military Mental Healthcare, Ministry of Defence, The Netherlands. PTSD patients were included when current PTSD was diagnosed according to DSM-‐IV criteria (American Psychiatric Association 1994). Severity of the PTSD symptoms was assessed with the CAPS (the clinician administered PTSD scale) (Blake et al., 1995), and comorbid psychiatric disorders were assessed with the Structural Clinical interview for DSM-‐IV axis I disorders (SCID) (First et al., 1997). The combat control-‐group was recruited via advertisements, where participants were included when they had no clinical PTSD symptoms (CAPS≤15) and no current psychiatric disorder. Furthermore, for both groups inclusion criteria were no alcohol or substance dependency, and no neurological disorder. MRI-‐scans were obtained at two time-‐ points for all participants: Prior to trauma-‐focused treatment (T0) and one after six to nine months (T6). At both time points interviews were collected. Between T0 and T6, all PTSD patients received ‘treatment as usual’, consisting of CBT and/or EMDR (Vermetten et al., 2012). Response of the PTSD patients to treatment was defined as a reduction of at least 30% of total CAPS score at T6 (Brady et al., 2000; Davidson et al., 2001), which resulted in a division of the PTSD patients into 23 Responders and 18 Non-‐Responders. Research was approved by the University Medical Centre Utrecht ethics committee and performed in line with the Declaration of Helsinki (World Medical, 2013), and all participants have written informed consent.
Diffusion MRI data pre-‐processing
At T0 and T6, diffusion MRI data were obtained using a 3.0 Tesla magnetic resonance imaging scanner (Philips Medical System, Best, The Netherlands). Diffusion MRI data included 30 diffusion weighted images wit a b-‐value of 1000 s/mm2 and one diffusion un-‐weighted image with a b-‐value of 0 s/mm2. In order to correct for distortions, two sets of diffusion MRI scans were obtained using reversed phase-‐ encoding blips with the following sequence parameters: TE = 68 ms, TR = 7057 ms, matrix 128 x 99, resolution 1,875 x 1,875 x 2, no gap, SENSE factor 3, EPI factor 35, FOV = 240 mm, 75 slices, slice thickness 2 mm. The obtained scans were corrected for eddy current induced geometric distortions, echo-‐planar-‐imaging (EPI) distortions, and subject motion (Leemans & Jones, 2009). A tensor model was fitted the tensor image, which was used for tractography and to calculate fractional anisotropy (FA). FA is an index of the microstructural integrity of white matter and provides information of myelination, axonal density and coherence of fibre orientation (Le Bihan et al., 2001). To increase the signal to noise ratio, these scalars were smoothed (FWHM 8 mm). At last, the FA values obtained by the two sets of scans were averaged.
Tractography
First, to reconstruct the cingulum bundle in the brain, whole brain tractography has been performed with Explore DTI (Leemans & Jones, 2009). In order to obtain a complete anatomical cingulum bundle on each side of the brain, four subdivisions are reconstructed on the left and right side (based on Jones
et al., 2013). The fibre tract pathways were extracted by defining a set of AND and NOT regions of
interest (ROIs) based on a representative single participant. The first ROI was placed vertical at the rostral end of the cingulum bundle, right under the most posterior part at the anterior flexure of the
corpus callosum. The second ROI was placed horizontal at the most anterior point of the cingulum bundle, perpendicular to the fibres of the cingulum bundle. The third ROI was placed superior in the middle of the cingulum bundle, perpendicular to the placed ROI. The fourth ROI was placed horizontal at the most caudal point of the cingulum bundle, perpendicular to the placed ROI. The last ROI was placed in line with the most superior ROI used for the reconstruction of the posterior-‐ and anterior dorsal cingulum. See figure 1 for all the placed ROIs.
After tracing these four subdivisions on a representative single participant, the fibre tract pathways of the cingulum bundle were automatically reconstructed for all data sets (Lebel et al., 2008), and stats of the fibre tracts were calculated using Explore DTI automatic reconstruction (Leemans & Jones, 2009). The tract segment between two consecutive AND ROIs has been used. After obtaining these tracts, a filter has been used to cut off tracts that are leaving and re-‐enter the segmented region (length range = ± 3 x SD and kappa = 3). Each constructed subdivision of the cingulum bundle was divided into segments of approximately 2mm (voxel size) by performing uniform tract resampling and FA values has been calculated for all the segments derived from the four subdivisions.
Statistical analysis
To test whether trauma-‐focused treatment as a function of time has a different effect on FA values for Responders and Non-‐Responders compared to combat control-‐group, a linear mixed-‐effect model (lme4 package; Bates, Maechler & Bolker, 2012) was used. As fixed effects, we entered group and time into the model. For the left and right cingulum bundle, the linear mixed-‐effect model was performed separately for all segments derived from the four subdivisions. To correct for multiple comparisons, permutation testing was used with a statistical threshold of p < 0.05 (Nichols & Holmes, 2001). However, because analyses were performed for the left and right cingulum bundle separately, a Bonferroni correction was applied, which gives a statistical threshold of p < 0.025 (p= 0.05 / 2 = 0.025) (Bland & Altman, 1995). In these permutation tests, the group where the participants are assigned to was shuffled. Then the linear mixed-‐effect model was performed 1000 times to acquire a distribution under the null hypothesis of the maximum length of consecutive ranges of segments showing a significant per-‐segment effect. This was used to define the criterion for when a range of per-‐segment significant segments had a significant length. This approach thus controls for the multiple testing problem, without automatically becoming highly conservative when the number of segments is high. The linear mixed-‐effect model as well the permutation-‐test was performed in R-‐studio (http://www.rstudio.org/).
Results
Participants
Descriptive statistics of the clinical and demographical information are shown in table 1. At the second clinical assessment, 23 patients were Responders (>30% CAPS reduction), and 18 patients were Non-‐ Responders (<30% CAPS reduction). The combat controls, Responders and Non-‐Responders did not differ in age (F(2,58) = 1.27, p = 0.29), number of times they were deployed (X2(14) = 12.59, p = 0.56), time since last deployment (F(2, 57) = 0.49, p = 0.61), and time between scans (F(2, 57) = 0.61, p = 0.55).
Table 1 Descriptive statistics of clinical and demographical information of the different groups
Responders (mean ± SD) Non-‐Responders (mean ± SD) Combat control (mean ± SD) Test-‐value (df) Sig. Two-‐ tailed Demographical information N Age (range: 21-‐57) 23 33.52 (±8.56) 18 38.44 (±10.30) 20 36.40 (±11.60) F(2,58) = 1.27 p = 0.29 Deployment information prior T0
Number of times deployed (1 / 2 / 3 / >3) Time since last deployment (in years)
(9 / 6 / 3 / 5) 7.70 (±7.78) (6 / 2 / 7 / 2) 6.12 (±7.60) (6 / 5 / 4 / 5) 5.65 (±5.65) X2(14) = 12.59 F(2, 57) = 0.49 p = 0.56 p = 0.61 Treatment information prior T0
EMDR treatments before T0 Numbers of sessions (<5 /5-‐10/>10)
CBT treatments before T0
Numbers of sessions (<5 /5-‐10/>10)
ECL treatments before T0
Numbers of sessions (<5 /5-‐10/>10)
Structured treatments before T0 Numbers of sessions (<5 /5-‐10/>10)
0.50 (±1.24) (3 / 1 / -‐) 0.10 (±0.45) (1 / -‐ / -‐) 0.00 (±0.00) (-‐ / -‐ / -‐) 0.15 (±0.49) (2 / -‐ / -‐) 0.20 (±0.78) (1 / -‐ / -‐) 2.13 (±4.41) (1 / 1 / 2) 0.27 (±1.03) (1 / -‐ / -‐) 0.00 (±0.00) (-‐ / -‐ / -‐) t(33) = -‐.83 t(33) = 1.78 t(14) = 1.00 t(19) = -‐1.37 p = 0.42 p = 0.10 p = 0.25 p = 0.19
Table 1 continued Responders (mean ± SD) Non-‐Responders (mean ± SD) Combat control (mean ± SD) Test-‐value (df) Sig. Two-‐ tailed Medication prior T0 SSRI/SARI Benzodiazepines Antipsychotics Other 4 7 2 7 3 -‐ -‐ X2(2) = 4.19 X2 (1) = 1.04 X2(1) = 1.65 -‐ p = 0.12 p = 0.31 p = 0.20 -‐
Clinical scores prior T0 Caps total score
Current comorbid disorder baseline (SCID) Mood disorder Anxiety disorder Somatoform disorder 70.13 (±3.07) 13 4 1 69.78 (±2.69) 8 8 2 t(39) = -‐0.08 X2(1) = 0.59 X2 (1) = 0.06 X2(1) = 0.41 p = 0.93 p = 0.54 p = 0.09 p = 0.41 Scan information between T0 – T6
Time between scans (in months) 6.57 (±0.99) 6.35 (±0.86) 6.30 (±0.57) F(2, 57) = 0.61 p = 0.55
Treatment information between T0-‐T6 CBT treatments between scans Number of sessions (<5 /5-‐10/>10)
EMDR treatments between scans Number of sessions (<5 /5-‐10/>10)
ECL treatments between scans Numbers of sessions (<5 /5-‐10/>10)
Structured treatments between scans Numbers of sessions (<5 /5-‐10/>10)
0.05 (±0.22) (1 / -‐ / -‐) 5.60 (±4.32) (6 / 7 / 4) 1.00 (±4.47) ( / / 1) 3.20 (±4.40) (9 / 4 / 2) 4.38 (±5.33) (2 / 3 / 4) 4.31 (±4.14) (5 / 4 / 2) 0.81 (±2.29) (-‐ / 2 / -‐) 0.25 (±1) (1 / -‐ / -‐) t(15.04) = 3.25 t(34) = -‐0.91 t(34) = -‐0.15 t(21.42) = -‐0.15 p = 0.00 p = 0.37 p = 0.88 p = 0.01 Medication T6 SSRI/SARI Benzodiazepines Antipsychotics Other 5 5 2 -‐ 10 1 2 2 X2 (2) = 9.45 X2 (1) = 2.09 X2(1) = 0.07 X2 (1) = 2.73 p = 0.01 p = 0.15 p = 0.79 p = 0.10 Clinical scores T6
Caps total score
Current comorbid disorder baseline (SCID) Mood disorder Anxiety disorder Somatoform disorder 29.52 (±3.53) 3 2 0 66.94 (±3.95) 3 6 2 t(39) = 7.06 X2(1) = 0.16 X2(1) = 4.32 X2 (1) = 2.85 p = 0.00 p = 1.00 p = 0.05 p = 0.17
Furthermore, no differences were found between the responders group and Non-‐Responders group in the numbers of EMDR sessions (t(33) = -‐.83, p = 0.42), CBT sessions (t(33) = 1.78, p = 0.10), ECL treatments (t(14) = 1.00, p = 0.25), and structured treatments (t(19) = -‐1.37, p = 0.19) before T0. The total numbers of EMDR sessions between T0 and T6 did not differ between the Responders and Non-‐ Responders (t(34) = -‐0.91, p = 0.37), neither did the total numbers of ECL treatments (t(34) = -‐0.15, p = 0.88). However, the Non-‐Responders group had more CBT sessions between the two scans compared to the Responders group (t(15.04) = 3.25, p = 0.00) whereas the Responders group had more structured treatments (t(21.42) = -‐0.15, p = 0.01). The CAPS score prior to treatment was not different among the Responder-‐ and Non-‐Responder group (t(39) = -‐0.08, p = 0.93). However, CAPS score at the second clinical assessment was higher for the Non-‐Responders group (t(39) = 7.06, p = 0.00). The Non-‐ Responders group used more SSRI/SARI compared to Responders post treatment (X2(2) = 9.45, p = 0.01). No differences were found prior treatment between the Responders group and Non-‐Responders group in SSRI/SARI (X2(2) = 4.19, p = 0.12), benzodiazepines (X2(1) = 1.04, p = 0.31) or antipsychotics (X2(1) = 1.65, p = 0.20). Neither did the two groups differs post treatment in benzodiazepines (X2(1) = 2.09, p = 0.15), antipsychotics (X2(1) = 0.07, p = 0.79) or other medication (X2(1) = 2.73, p = 0.10). However, they did differ in SSRI/SARI (X2(2) = 9.45, p = 0.01) post treatment, in which 10 participants of the Non-‐ Responders group against 5 participants of the Responders group used SSRI/SARI medication. At baseline, mood-‐ (X2(1) = 0.16, p = 1.00) and somatoform disorder was equally prevalent among the Responders-‐ and Non-‐Responder group (X2(1) = 2.85, p = 0.17). A trend was observed for comorbidity of anxiety disorder (X2(1) = 4.32, p = 0.05), where 2 participants of the Responders group and 6 participants of the Non-‐Responders group had an anxiety disorder.
Tractography Analysis
Right cingulum. The right cingulum was divided into 102 segments of approximately 2mm
(voxel size). No group-‐by-‐time interaction (p = 0.659) and no group-‐effect (p = 1) was found for the 102 FA values of the right cingulum.
Left cingulum. The left cingulum was divided into 97 segments of approximately 2 mm (voxel
size). A trend was found for the group-‐by-‐time interaction for the 97 FA values (p = 0.034) of the left cingulum. This trend significant effect was driven by the Responders group over time (p = 0.041), where we found a maximum range of five consecutive, significant segments (see figure 2). For these five consecutive segments, the three different groups did not differ in FA values at T0 (F(2,29) = 1.231, p = 0.307). At T6, a trend was found for differences between the groups (F(2,29) = 3.283, p = 0.052). Responders showed a significant effect of time, where they had higher FA values after treatment (t(10)=-‐ 6.790, p=0.000) compared to their own FA values prior treatment (see figure 3). No significant effect of time was found for the Non-‐Responders (t(8)=0.841, p=0.425) and combat control (t(11)=0.964, p=0.356). In the Responders group, however, no correlation was found between symptom improvement and delta FA (i.e. the differences in FA value on T6 and T0) (r = 0.160, N = 11, p = 0.64).
Figure 2. Tract of the left parahippocampal cingulum. P-‐values of each voxel-‐sized segment are plotted (p ≥
0.05 are plotted in blue, whereas p < 0.05 are plotted in a range from red through blue, with p = 0.00 is red and p= 0.05 is blue). The red part at the bottom of the tract is the maximum found range of five consecutive, significant voxel-‐size segments.
Figure 3. Group by time interaction effect in mean Fractional Anisotropy (FA) values for the five consecutive, significant voxel-‐size segments in the left parahippocampal cingulum. Mean FA values at T0 and T6 are
presented for the combat controls (blue line), Non-‐Responders (green line), Responders (red line). The error bars represent the standard deviations.
0,3 0,35 0,4 0,45 0,5 0,55 T0 T6 Me an FA
LeV cingulum
Group by Wme interacWon effect
Combat Non-‐responders Responders
Discussion
To investigate the effect and location of the effect of trauma-‐focused treatment in PTSD on the white matter of the cingulum, diffusion tensor imaging (DTI) has been obtained prior (T0) and after (T6) treatment. Twenty-‐three patients responded to treatment (>30% CAPS reduction), and 18 patients did not respond (<30% CAPS reduction). In summary, we found a trend towards an interaction between group-‐by-‐time for FA values in the left parahippocampal cingulum, which was driven by the Responders-‐group over time. Specifically, this trend significant effect was found in five consecutive, significant segments in the left parahippocampal cingulum, where Responders showed a significant increase in FA values over time. Additionally, no differences were found in FA values prior to treatment between the different groups. However, a trend was found for differences in FA between the different groups after treatment: The FA values for the Responders was higher compared to the Non-‐ Responders and combat control.
These results suggests that trauma-‐focused treatment has an effect on the white matter microstructure of the cingulum bundle in PTSD patients who respond after treatment. The results of this study seems to contradict our hypothesis that PTSD patients will have lower pre-‐treatment FA values in the cingulum compared to combat controls and be more similar to combat controls after successful treatment. However, our results appear to support the hypothesis that in the Responders group the FA values will increase over time. This indicates that increased integrity of the white matter microstructure in the left parahippocampal cingulum is a possible effect of successful trauma-‐focused treatment that develops over time.
Previous findings of Kim et al. (2006), Schuff et al. (2011), Fani et al. (2012), and Sanjuan et al. (2013) showed reduced FA values in PTSD prior to treatment, in contrast to our study where no
differences were found in FA values between the different groups at T0. This contradicting finding is possibly due to differences in used control groups between these mentioned studies and this current study. Some of these studies did not include a control group at all, or included a non-‐deployed control group, whereby they did not control for trauma-‐exposure.
Results of the current study are in line with a previous study of Zhang et al. (2012), which also showed an increase in FA in PTSD in the parahippocampal cingulum. In this current study it has been shown that this increase in FA, located in the parahippocampal cingulum, was specific to PTSD patients who respond to trauma-‐focused treatment. We found a trend significant difference between these heightened FA values of the Responders compared with the Non-‐Responders and combat controls. In line with our results, a few structural neuroimaging studies showed an associating between structure changes over time and responsiveness of trauma-‐focused treatment. Lindauer et al. (2005) and Levy-‐ Gigi et al. (2013) demonstrated that PTSD patients had smaller hippocampal volumes compared to trauma-‐exposed controls prior to treatment. After successful psychotherapy, the hippocampal volumes of PTSD patients were increased and therefore no volume changes were found between the PTSD patients and trauma-‐exposed controls. Besides, Levy-‐Gigi et al. (2013) showed that improvement in PTSD symptoms was associated with increased hippocampal volume. However, so far, no studies explore the activation of the hippocampus prior-‐ and post successful treatment. Therefore, future research should assess hippocampal activity prior-‐ and post successful treatment, and the possible relationship of the hippocampal activity over time and FA in the left parahippocampal cingulum. Moreover, if this relationship would be confirmed in future research, one possible treatment for PTSD could be deep-‐brain stimulation during trauma-‐focused treatment by means of deep-‐brain stimulation
in areas such as the hippocampus, to induce increased FA values in the left parahippocampal cingulum, which may result in successful treatment.
We note some limitations of the current study. Firstly, as result of missing values, there is a lack of power in the current study. Secondly, many segments were used in the current study (in total 199 segments). The kind of correction provided by the permutation testing technique does not result in extremely conservative tests due to the number of segments, but it is unclear what the optimal number of segments is in terms of noise. However, due to the use of relatively many segments, we were able to localize precisely the effect of successful trauma-‐focused treatment on the cingulum. Further research could recruit more participants which will result in an increase in power of the analysis. Moreover, to assess whether increased integrity of white matter microstructure of the cingulum is a biological mechanism that underlies responsiveness of trauma-‐focused treatment, further research should perform a longitudinal study whereby the development of recently traumatized subjects are assessed to determine whether alterations in integrity of white matter microstructure develops prior to the onset of PTSD or perhaps is acquired after the onset of PTSD and in that case when exactly it is acquired. These further studies are needed to unravel possible biomarkers that could predict treatment response. Thirdly, some PTSD patients were taking medication at T0 and T6. Specifically, the Non-‐Responders were taken significant more SSRI/SARI compared with the Responders group. Besides, we included PTSD patients with comorbid disorders, whereby at both prior-‐ and after-‐treatment the Non-‐Responders group had near significant more comorbid anxiety disorder compared to Responders. A study by Tarrier et al. (2000) showed that poorer outcome after treatment was associated with more comorbid anxiety disorder. Taken these and our results together, it seems that PTSD patients with more comorbid anxiety disorders shows a poorer response to
treatment compared to PTSD patients with less comorbid anxiety disorders. However, in this current study, post-‐hoc analyses demonstrated that there is no correlation between number of comorbidity anxiety disorders prior nor after treatment and delta FA values. Therefore, it is unlikely that in this study comorbid anxiety disorders confounds our results. However, further studies are needed to unravel the exact mechanism whether and which medication and comorbid disorders influences white matter microstructure. Finally, the PTSD patients received ‘treatment as usual’. As results show, the Non-‐Responders group received significant more CBT sessions compared to the Responders-‐group in between the two scans. This contrasts with the structured treatment sessions (not protocolized treatment sessions), of which the Responders received significant more. Post-‐hoc analysis revealed there was a trend significant correlation between CAPS improvement and number of structured treatment sessions as well as a trend significant correlation between delta FA and number of sessions of EMDR and structured treatment sessions. Taken together, these results suggest that it is possible that the effects on FA reflected effects of receiving different treatments. Additionally, one limitation in this current study is the duration of the trauma-‐focused treatment that the PTSD patients received. Namely, it could be the case that 6-‐8 months of trauma-‐focused treatment is not long enough for each participant to acquire increased integrity of the white matter microstructure of the left parahippocampal cingulum. Besides, not all participants were at the end of their treatment program at T6. Therefore, future studies should follow PTSD patients for a longer time till they show a response to trauma-‐focused treatment. This will allow to unravel the neurobiological effects and the onset of these effects of available trauma-‐focused treatment for PTSD.
Conclusion
The current study indicates that the white matter microstructure of the left parahippocampal cingulum may be affected by trauma-‐focused treatment in PTSD. The findings highlight the importance of considering the spatial distribution of effects in tractography studies. In conclusion, tractography studies can selectively assess the effect of trauma-‐focused treatment in PTSD, and shed more light on the neurobiological mechanisms of available trauma-‐focused treatments. However, longitudinal studies which assess the participants immediately after trauma-‐exposure and until they cease receiving any trauma-‐focused treatment anymore are needed. In conclusion, the integrity of the white matter microstructure of the left-‐parahippocampal cingulum appear to be potentially an important factor in the outcome of trauma-‐focused treatment in PTSD.
Literature
Abe, O., Yamasue, H., Kasai, K., Yamada, H., Aoki, S., & Iwanami, A. et al. (2006). Voxel-‐based diffusion tensor analysis reveals aberrant anterior cingulum integrity in posttraumatic stress disorder due to terrorism. Psychiatry research, 146, 231-‐242.
American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 5th ed. (2013). Arlington: American Psychiatric Association.
Bates, D.M., Maechler, M., & Bolker, B. (2012). lme4: Linear mixed-‐effects models using S4 classes. R packages version 0.999999-‐0.
Bland, J.M, & Altman, D.G. (1995). Multiple significance tests: The Bonferroni method. British Medical Journal, 310, 170.
Bradley, R., Greene, J., Russ, E., Dutra, L., & Westen, D. (2005). A multidimensional meta-‐analysis of psychotherapy for PTSD. The American Journal of Psychiatry, 162, 214-‐227.
Brady, K., Pearlstein, T., Asnis, G.M., Baker, D., Rothbaum, B., & Sikes, C.R. et al. (2000). Efficacy and safety of sertraline treatment of posttraumatic stress disorder. Journal American Medical Association, 283, 1837-‐1844.
Bryant, R.A., Felmingham, K., Kemp, A., Das, P., Hughes, G., & Peduto, A. et al. (2008a). Amygdala and ventral anterior cingulate activation predicts treatment response to cognitive behaviour therapy for post-‐traumatic stress disorder. Psychological medicine, 38, 555-‐561.
Bryant, R.A., Felmingham, K., Whitford, T.J., Kemp, A., Hughes, G., & Peduto, A. et al. (2008b). Rapport sommaire rostral anterior cingulate volume predicts treatment response to cognitive-‐behavioural therapy for posttraumatic stress disorder. Journal of Psychiatry & Neuroscience, 33, 142-‐146.
Corrigan, F.M. (2004). Psychotherapy as assisted homeostasis: Activation of emotional processing mediated by the anterior cingulate cortex. Medical hypotheses, 6, 968-‐973.
Davis, M. (1992). The role of the amygdala in fear and anxiety. Annual review of neuroscience, 15, 353-‐375.
Etkin, A., Pittenger, C., Polan, H.J., & Kandel, E.R. (2005). Toward a neurobiology of psychotherapy: Basic science and clinical applications. Journal of Neuropsychiatry and Clinical Neuroscience, 17, 145-‐158.
Falconer, E., Allen, A., Felmingham, K.L., Williams, L.M., & Bryant, R. (2013). Inhibitory neural activity predicts response to cognitive-‐ behavioral therapy for posttraumatic stress disorder. The Journal of clinical psychiatry, 74, 895-‐901.
Fani, N., King, T.Z., Jovanovic, T., Glover, E.M., Bradley, B., Choi, K. et al. (2012). White matter integrity in highly traumatized adults with and without post-‐traumatic stress disorder. Neuropsychopharmacology, 37, 2740-‐2646.
Felmingham, K., Kemp, A., Williams, L., Das, P., Hughes, G., & Peduto, A. et al. (2007). Changes in anterior cingulate and amygdala after cognitive behavior therapy of posttraumatic stress disorder. Psychological science, 8, 127-‐129.
First, M.B., Spitzer, R.L, & Gibbon, M. (1997). Structured Clinical Interview for DSM-‐IV Axis I Disorders.
Harvey, A.G., Bryant, R., & Tarrier, N. (2003). Cognitive behaviour therapy for posttraumatic stress disorder. Clinical Psychology Review, 23, 501-‐522.
Hoge, C.W., Auchterlonie, J.L., & Milliken, C.S. (2006). Mental health problems, use of mental health services, and attrition from military service after returning from deployment to Iraq or Afghanistan. The Journal of the American Medical Association, 295, 1023-‐1032.
Jones, D.K., & Leemans, A. (2010). Diffusion Tensor Imaging. Magnetic Resonance Neuroimaging, 711, 127-‐144.
Kennis, M., Rademaker, A.R., van Rooij, S.J.H., Tromp, D.P.M., Fox, A.S., Rademaker, A.R., & Kahn, R.S. et al. (2015). Treatment outcome-‐ related white matter differences in veterans with posttraumatic stress disorder. Neuropsychopharmacology, advance online publication, 22 April 2015 (doi: 10.1038/npp.2015.94).
Kim, S.J., Jeong, D., Sim, M.E., Bae, S.C., Chung, A., & Kim, M.J. et al. (2006). Asymmetrically altered integrity of cingulum bundle in posttraumatic stress disorder. Neuropsychobiology, 54, 120-‐125.
Le Bihan, D., Mangin, J., Poupon, C., Clark, C.A., Pappata, S., & Molko, N. et al. (2001). Diffusion tensor imaging: Concepts and applications. Journal of Magnetic Resonance Imaging, 13, 534-‐546.
Lebel, C., Walker, L., Leemans, A., Phillips, L., & Beaulieu, C. (2008). Microstructural maturation of the human brain from childhood to adulthood. NeuroImage, 40, 1044-‐1055.
Leemans, A., and Jones, D.K. (2009). The B-‐matrix must be rotated when correcting for subject motion in DTI data. Magnetic Resonance in Medicine, 61, 1336-‐1349.
Levy-‐Gigi, E., Szabó, C., Kelemen, O., & Kéri, S. (2013). Association among clinical response, hippocampal volume, and FKBP5 gene expression in individuals with posttraumatic stress disorder receiving cognitive behavioral therapy. Biological Psychiatry, 74, 793-‐800.
Lindauer, R.J.L., Vlieger, E., Jalink, M., Olff, M., Carlier, I.V.E., & Majoie, C.B.L.M. et al. (2005). Effects of psychotherapy on hippocampal volume in out-‐patients with post-‐traumatic stress disorder: A MRI investigation. Psyhological Medicine, 35, 1421-‐1431.
Moseley, E., Wendland, F., Kucharczyk, J., & Tsumuda, J. (1990). Diffusion-‐weighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology, 176, 439-‐445.
Nardo, D., Högberg, G., Chee Leong Looi, J., Larrson, S., Hällström, T., & Pagani, M. (2010). Gray matter density in limbic and paralimbic cortices is associated with trauma load and EMDR outcome in PTSD patients. Journal of Psychiatric Research, 44, 477-‐485.
Nichols, T.E., & Holmes, A.P. (2001). Nonparametric permutation tests for functional neuroimaging: A primer with examples. Human Brain Mapping, 15, 1-‐25.
Ponniah, K., & Hollon, S.D. (2009). Empirically supported psychological treatments for adult acute stress disorder and posttraumatic stress disorder: A review. Depression and anxiety, 26, 1086-‐1109.