Inflammatory Markers and Brain Volume in Patients with Post-traumatic Stress Disorder

Abstract

Objective

Posttraumatic stress disorder (PTSD) is characterized by increased inflammatory processing and altered brain volume. In this study, we investigated the relationship between inflammatory markers and brain volume in patients with PTSD.

Methods

Forty-five patients with PTSD, and 70 healthy controls (HC) completed clinical assessments and self-reported psychopathology scales. Factors associated with inflammatory responses including brain-derived neurotrophic factor and four inflammatory biomarkers (C-reactive protein, cortisol, Interleukin-6, and homocysteine) and T1-magnetic resonance imaging of the brain were measured.

Results

In the PTSD group, cortisol level was significantly lower (t = 2.438, p = 0.046) than that of the HC. Cortisol level was significantly negatively correlated with the left thalamus proper (r = −0.369, p = 0.035), right thalamus proper (r = −0.394, p = 0.014), right frontal pole (r = −0.348, p = 0.039), left occipital pole (r = −0.338, p = 0.044), and right superior occipital gyrus (r = −0.397, p = 0.008) in patients with PTSD. However, these significant correlations were not observed in HC.

Conclusion

Our results indicate that increased cortisol level, even though its average level was lower than that of HC, is associated with smaller volumes of the thalamus, right frontal pole, left occipital pole, and right superior occipital gyrus in patients with PTSD. Cortisol, a major stress hormone, might be a reliable biomarker to brain volumes and pathophysiological pathways in patients with PTSD.

INTRODUCTION

Posttraumatic stress disorder (PTSD) is a complex and severe mental disorder [1]. It results from directly experiencing, witnessing, or being exposed to unexpected extreme traumatic events including combat, terrorist attack, physical illness, and sexual abuse [2,3]. According to the Diagnostic and Statistical Manual of Mental disorders 5th Edition (DSM–5) [4], PTSD is characterized by re-experiencing the traumatic event, increased physical arousal, the avoidance of thoughts, feelings, activities related to the trauma, and alterations in mood and cognition. Fur-thermore, PTSD is pertinent to other health problems, among which dysfunctional inflammatory responses are noteworthy [5].

According to previous studies, individuals with PTSD showed increased inflammatory immune activities and high levels of inflammatory cytokines. For instance, increased Interleukin-6 (IL-6) levels predicted the development of PTSD at 6 months [6]. In addition, brain-derived neurotrophic factor (BDNF) [7,8] and other inflammatory response markers, including high sensitive C-reactive protein (hs-CRP) [9,10] and homocysteine [11,12] have been repeatedly reported to be associated with pathology of PTSD. Furthermore, elevated levels of cortisol in individuals with PTSD, except for a few phenomenon such as cortisol resistance, might result in decreased immune reactions and insufficient immune regulations in general [13]. However, several studies found relatively low levels of cortisol [14], specifically in women with PTSD [15] and individuals with severe dissociative symptoms [16].

Brain volumetric anomalies were observed in whole or specific regions among PTSD patients. Meta-analyses of structural magnetic resonance imaging (MRI) studies have consistently identified reductions in brain volume, most prominently in the hippocampus in patients with PTSD [17]. Regarding the whole-brain volume level, PTSD patients showed significant reductions, particularly in the frontal and the occipital regions in comparison to healthy individuals [18]. Furthermore, the change of brain volume has been related to high concentrations of glucocorticoid receptors and dysregulation of hypothalamic- pituitary-adrenal (HPA) activity [19]. Therefore, reduced brain volume might be associated with an inflammatory circle which could result in more severe atrophy in some brain regions.

Volumetric brain changes in multiple regions have been related to inflammation factors, especially in stress-related disorders [20]. According to several previous studies, there has been a positive correlation between inflammatory factors and brain volume [21,22]. This correlation means that elevated inflammatory factors were associated with increased brain volume in patients. However, these findings were controversial. Other studies have found a negative correlation between inflammatory factors and brain volume [23,24], which means that increased inflam-matory factors were associated with reduced brain volume in patients. Especially, it is important to find out how inflammation factors affect volumetric changes of the brain in stress-related disorders since it is known that stress might significantly affect inflammation. Therefore, it is needed to investigate whether these inconsistent results might be attributed to types of inflammation factors or brain volume region. Additionally, despite closely interrelated relationships, few studies have focused on the relationship between inflammatory factors and the several regional brain volumes in patients with PTSD.

Therefore, this study aimed to investigate the relationship between inflammatory factors and brain volume in PTSD patients and healthy controls (HC). First, we compared the levels of inflammatory factors and the brain volume between PTSD and HC. Furthermore, we examined the correlations between inflammatory factors and brain volume in each group. We hypothesize that the inflam-matory factors would be negatively correlated with the brain volume of the main pathology in patients with PTSD.

METHODS

Participants

A total number of 130 participants (PTSD: n = 50, HC: n = 80; male/female n = 42/88) were recruited from the Psychiatry Department of Inje University Ilsan Paik Hos-pital (PTSD group) and the local community by distributing flyers and posters (HC group). The diagnosis of PTSD was based on the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5) by trained psychiatrists [25]. Patients were excluded if they had a severe brain injury, which could be detected from computed tomography (CT) or MRI. The patients with PTSD were taking medications such as escitalopram (n = 24), vortioxetine (n = 8), paroxetine (n = 4), desvenlafaxine (n = 4), sertraline (n = 5) zyprexa (n = 8), quetiapine (n = 10), lorazepam (n = 12), clonazepam (n = 5), diazepam (n = 5), and alprazolam (n = 25). Moreover, in that case of PTSD group, comorbid diagnosis was MDD (n = 16), and panic disorder (n = 8). Other psychiatric diagnosis was not permitted in this study. Additionally, the participants in the HC did not satisfy any type of psychiatric disorder criteria and had no history of mental disorder or major head trauma. Furthermore, 15 outliers (PTSD: n = 5, HC: n = 10) were excluded from the analyses, which are defined as more than 3 standard deviations above or below the mean of each inflammatory factor, which had been widely used in several studies [26-28]. It could prevent false positive correlation and obtain more reliable findings since correlation analysis is sensitive to outliers [29]. Specifically, there were outliers in hs crp (PTSD: n = 2, HC: n = 3), cortisol (PTSD: n = 2, HC: n = 3), BDNF (PTSD: n = 0, HC: n = 2), IL-6 (PTSD: n = 1, HC: n = 1), homocysteine (PTSD: n = 0, HC: n = 1). Then, 115 data were analyzed in the present study. However, the results from all 130 participants including outliers were also presented in Figures 1-3. Each participant signed a written form of informed consent before the experiment. This study was approved by the Institutional Review Board (IRB no. 2015-07-025) at Inje University Ilsan Paik Hospital on human experimentation.

Psychological Measures

CAPS-5

The CAPS-5 was used to diagnose PTSD based on the DSM-5 diagnosis for PTSD. It is a structured diagnostic interview by a psychiatrist, which is used to assess the frequency and severity of PTSD symptoms [25]. It consists of 30 items that measured frequency of symptoms, intensity of symptoms using dichotomous scores (“Yes” or “No”) and severity of symptoms using a five-point Likert scale, ranging from 0 (“absent”) to 4 (“extreme/incapacitating”). The coefficient alpha of the CAPS-5 score was 0.75 in this study.

Post-traumatic Stress Disorder Checklist-5

The Korean version of the Post-traumatic Stress Disor-der Checklist-5 (PCL-5) was used to assess the severity of the PTSD symptoms [30]. It is a self-reporting rating scale, which is used to assess screening, diagnostic evaluation, changes in PTSD symptoms [31]. It consists of 20 items that are measured using a five-point Likert scale, ranging from 0 (“not at all”) to 4 (“extremely”). The coefficient alpha of the PCL-5 score was 0.90 in this study.

Blood Sample Analysis

BDNF and 4 inflammatory factors such as hs-CRP, cortisol, IL-6 and homocysteine were measured since these factors have commonly reported to be significantly associated with brain volume in PTSD patients [22,32,33]. In addition, these factors have strong evidence that they have the significant association with brain volume in Korean participants [34,35]. Participants provided a blood sample between 6:00 AM and 10:00 AM. The obtained blood samples were centrifuged at 3,017 revolutions per minute at 10°C, for 10 minutes. Each inflammatory factor was analyzed as follow. First, high sensitivity C-reactive protein analysis was performed by the Ilsan Paik Laboratory Medi-cine Department using a turbidimetric immunoassay (TIA, Cobas 8000 Roche Diagnostics) with a coefficient of variation of 2.3% (measurement range 0.015−2.0 mg/dl). Second, serum cortisol analysis was performed by Ilsan Paik Laboratory Medicine Department using an electrochemiluminescence immunoassay (ECLIA, Cobas 8000 Roche Diagnostics) with a coefficient of variation of 3.8% (measurement range 0.054−63.4 mg/dl). Third, brain-derived neurotrophic factor analysis was performed by the Eone Laboratories using an enzyme-linked immunosorbent assay (ELISA, SpectraMax 190 Molecular Devices) with a coefficient of variation of 5.7% (measurement range 0.0625−4 pg/L). Fourth, interleukin-6 analysis was performed by the Eone Laboratories using an enzyme-linked immunosorbent assay (ELISA, SpectraMax 190 Molecular Devices) with a coefficient of variation of 3.9% (measurement range 3.13−300 pg/ml). Lastly, homocysteine analysis was per--formed by Ilsan Paik Laboratory Medicine Department using chemiluminescent microparticle immunoassay (CMIA, ARCHITECT i2000SR Abbott Diag-nostics) with a coefficient of variation of 3.3% (mea-surement range 1.00−50.00 mmol/L).

Image Acquisition, Processing, and Extraction of Regional Brain Volume

MRI was performed using high-resolution T1-weighted scans on a 1.5-Tesla scanner (Magneton, Avanto, Siemens). According to image pre-processing procedures, the T1- weighted MR images at the anterior commissure (AC) were set, and then the alignment by way of the mutual information affine registration with SPM12 tissue probability maps was approximated. The structural T1 images were affine regularization with an the Interna-tional Con-sortium for Brain Mapping space East Asian brain template and spatially normalized using the high-dimensional Diffeomorphic anatomical registration th-rough exponentiated lie algebra registration algorithm [36]. Jacobian- transformed tissue probability maps were conducted to estimate volume differences in gray matter and modulate the images using the computational anatomy toolbox for SPM (CAT12; developed by Christian Gaser, University of Jena, http://www.neuro.uni-jena.de/cat/), pro-vided in SPM12 (Wellcome Department of Cognitive Neurology, London, UK, https://www.fil.ion.ucl.ac.uk/spm) software and implemented in MATLAB (Mathworks Inc, https://kr.mathworks.com) platforms. Among 142 brain regions defined according to the Neuromorphometrics atlas, ventricles, brain white matter, and sub-regions were excluded. The 114 regions-of-interest of brain gray matter were estimated.

Statistical Analysis

Normality tests were conducted using skewness and kurtosis. The skewness over 2.0 and kurtosis over 7.0 were considered non-normal [37]. All variables in our results were established to be normally distributed when excluded outliers.

Demographic variables, levels of inflammatory factors, and brain volumes were analyzed by using χ² tests and independent ttests between the PTSD and HC. Covariates including sex, age, education were controlled. Analyses related to the difference of inflammatory factors or brain volume were statistically adjusted using 5,000-bootstrap resampling techniques for multiple tests [38]. The bootstrap test might be a weaker method than the Bonferroni test or false discovery rate for addressing the multiple comparison problem. However, the stability and robustness of the bootstrap test have been demonstrated by several previous studies [39,40]. Moreover, the bootstrap test has been widely used in brain volume analysis [41-45].

The residualized values of inflammatory factors were calculated using linear regression with age, sex, and education as covariates. Moreover, the residualized values of brain volume were calculated using the same method controlling for age, sex, education, body mass index (BMI), total intracranial volume as covariates. These variables were expected to influence inflammatory factors or brain volume [46,47]. Additionally, education was also selected as covariates [48,49], because it showed a significant difference between two groups. Pearson correlation analyses were performed to evaluate the relationship between the residualized values of inflammatory factors and brain volume. Then, the 5,000-bootstrap resampling techniques were used to statistically correct for multiple correlations. All analyses were conducted using IBM SPSS 21 (IBM Co.) and all significant levels were set at p < 0.05.

RESULTS

Demographic Characteristics

Table 1 shows the demographic characteristics of patients with PTSD and HC. There was a significant difference in education levels among the two groups, with higher education in HC compared to the PTSD patients (12.13 ± 3.35 vs. 14.37 ± 2.94 years; t = 3.77, p < 0.001, d = 0.71). In addition, there was a marginally significant age difference between the two groups, with higher age in HC compared to the PTSD patients (40.58 ± 12.99 vs. 45.47 ± 14.21 years; t = 1.86, p = 0.07, d = 0.35). The PCL-5 scores were significantly lower in HC than patients with PTSD (48.91 ± 13.15 vs. 11.54 ± 10.21, t = −16.96, p < 0.001, d = 3.17). The CAPS-5 scores were signifi-cantly lower in HC than patients with PTSD (CAPS-severity: 39.40 ± 8.57 vs. 6.32 ± 7.68, t = −15.14, p < 0.001, d = 4.07; CAPS-number of symptoms: 13.33 ± 2.59 vs. 2.10 ± 2.91, t = −14.83, p < 0.001, d = 4.08).

Table 1.

Demographic information of post-traumatic stress disorder (PTSD) and healthy controls (HC)

Variable	PTSD (n = 45)	HC (n = 70)	t or χ2	pvalue
Age (yr)	40.58 ± 12.99	45.47 ± 14.21	1.86	0.07
Sex			0.05	0.83
Male	15 (33.3)	22 (31.4)
Female	30 (66.7)	48 (68.6)
BMI	22.37 ± 3.12	23.10 ± 2.65	−0.50	0.62
Education (yr)	12.13 ± 3.35	14.37 ± 2.94	3.77	0.00
Duration of illness (yr)	225.04 ± 192.68	-	-	-
PCL-5	48.91 ± 13.15	11.54 ± 10.21	−16.96	0.00
CAPS-severity	39.40 ± 8.57	6.32 ± 7.68	−15.14	0.00
CAPS-number of symptoms	13.33 ± 2.59	2.10 ± 2.91	−14.83	0.00
Brain volume (mm3)
TIV	1,500.85 ± 133.35	1,525.32 ± 148.34	0.90	0.37

Values are presented as mean ± standard deviation or number (%).

BMI, body mass index; PCL-5, Post-traumatic Stress Disorder Checklist-5; CAPS, Clinician-Administered PTSD Scale for the Diagnostic and Statistical Manual of Mental disorders 5th Edition; TIV, total intracranial volume.

Inflammatory Factors and Brain Volume

There was a significant difference in the basal cortisol levels between PTSD and HC. The basal cortisol levels were significantly lower in the PTSD than in the HC (10.11 ± 4.40 vs. 12.16 ± 3.83 ug/dl; t = 2.438, p = 0.046, d = −0.497). BDNF and other inflammation factors were not (Fig. 1). Additionally, differences in brain volume between individuals with PTSD and HC were not significant. In the case of including outliers, there was also the significant difference in only cortisol, as same as results excluding outliers, between the two groups (10.40 ± 5.58 vs. 12.40 ± 4.34 ug/dl; t = 2.28, p = 0.036, d = −0.4; Supplementary Fig. 1; available online).

Fig. 1.

Comparison of BDNF and four inflammatory factors between patients with post-traumatic stress disorder (PTSD) and healthy controls (HC).

Correlation between the Inflammatory Factors and Brain Volume

The level of cortisol was significantly correlated with brain volume in the PTSD group, but not in HC. For individuals with PTSD, the cortisol level all showed a statistically significant negative correlation with the volumes of the left thalamus proper (r = −0.369, p = 0.035), right thalamus proper (r = −0.394, p = 0.014), right frontal pole (r = −0.348, p = 0.039), left occipital pole (r = −0.338, p = 0.044), and right superior occipital gyrus (r = −0.397, p = 0.008). Figures 2 and 3 showed correlation scatter plots created based on residualized values of cortisol and five regions of brain volume (i.e., left thalamus proper, right thalamus proper, right frontal pole, left occipital pole, and right superior occipital gyrus) in the PTSD patients and HC, with the brain region visualized using brainstorm toolbox [50]. When it included the outliers, there were also significant negative correlation between cortisol levels and brain volumes in patients with PTSD (Supplementary Figs 2, 3; available online).

Fig. 2.

Correlation between residualized value of cortisol controlling for age, sex, education, and residualized value of brain volumes controlling for age, sex, education, body mass index, and total intracranial volume. In the post-traumatic stress disorder (PTSD) group, the cortisol level significantly negatively correlated with (A) the left thalamus proper, (B) right thalamus proper, and (C) right frontal pole volumes, respectively. In the healthy group (HC), all correlations were not significant. All pvalues were adjusted using bootstrapping. Images of the brain were acquired using the Brainstorm toolbox.

Fig. 3.

Correlation between residu-alized value of cortisol adjusted for age, sex, education and residualized value of brain volumes adjusted for age, sex, education, body mass index, and total intracranial volume. In the post-traumatic stress disorder (PTSD) group, the cortisol level significantly negatively correlated with (A) the left occipital pole and (B) right superior occipital gyrus volume, respective-ly. In the healthy group (HC), all cor-relations were not significant. All pvalues were adjusted using boot-strapping. Images of the brain were acquired using the Brainstorm tool-box.

DISCUSSION

In this study, we investigated the relationship between inflammatory factors and brain volume in patients with PTSD compared to the HC. There was a significant difference in cortisol levels between patients with PTSD and HC. The PTSD group showed significantly lower cortisol levels compared to the HC. Furthermore, there were significant negative correlations between the cortisol levels and brain volume, including the left thalamus proper, right thalamus proper, right frontal pole, left occipital pole, and right superior occipital gyrus only among the PTSD patients. These correlations remained significant even after controlling for several covariates, such as age, sex, education, BMI, and total intracranial volume.

The findings indicate that the PTSD patients, compared to HC, showed lower basal cortisol levels. This result was also persistent when the outliers were included. This hypocortisolism among PTSD has been found in several previous studies using serum or plasma cortisol [51,52]. These results suggest that prolonged stress might relate to adrenal depletion or increased feedback inhibition of the HPA axis, which could result in the symptoms of PTSD including hyperarousal, sensitization of fear or anxiety, and hypervigilance [53]. On the contrary, hypercortisolism has also been reported in patients with PTSD [54,55]. These inconsistent findings could be explained by several causes. First, the characteristic of comparison group, sex, and comorbidity with other psychiatric disorder could account for the mixed finding. Second, dissociative symptom, which were commonly observed in patient with PTSD, could a significant influencing factor of low cortisol level in patients with PTSD [16]. Previous studies reported that the dissociative subtype of PTSD exhibit a symptom like derealization or depersonalization, which could be attributed to low cortisol levels of patients with PTSD [56,57].

Moreover, we found no significant differences in brain volumes between patients with PTSD and HC. This result was against our assumption that brain volume of patients with PTSD would be reduced in general. However, many previous studies have found no significant reduction in brain volume of patients with PTSD [58-61]. These inconsistent results might be attributed to the various clinical characteristics of participants. Specifically, several demographical or pathophysiological factors like age of illness onset, course of illness, severity of symptoms, different PTSD triggers (combat, abuse, etc.) and treatment could influence regional brain volume [62].

The main results of the study were significant negative correlations between basal cortisol level and brain volumes including bilateral thalamus proper, right frontal pole, left occipital pole, and right superior occipital gyrus in patients with PTSD. Even though including the outliers, the negative correlation between cortisol and several brain volumes was also found in patients with PTSD. It is in accordance with several previous findings. A signifi-cant negative relationship was found between pre-bedtime cortisol levels and left ventral prefrontal cortex volumes in the youth with traumatic symptoms [24], left thalamus volumes [63], inferior occipital gyrus that is related to visual processing areas [64,65]. Our findings suggest that the underlying pathways of the relationship between cortisol and brain volume might be associated with stress response [14]. According to Heim and colleagues [14], hypocortisolism might be related to low adrenal activity or reactivity, which could lead to chronic HPA axis hypoactivity followed by hippocampal volume reduction [66]. It may seem paradoxical that the two biomarkers, hypocortisolism and brain volume atrophy, have opposite relationships. In general, during periods of stress, cortisol dysregulation can be identified, which produces volumetric alterations in the brain, so the two biomarkers have negative correlations in patients with PTSD [24,67]. How-ever, ironically, hypocortisolism was observed in patients with PTSD since it could be influenced by several causes including dissociative symptoms as we described [16, 56,57]. Therefore, the relationship between cortisol and brain volume should be more discussed in further studies considering these points.

Furthermore, our correlational results might relate to cognitive dysfunction. The brain regions of the correlational findings (i.e., thalamus, right frontal pole, left occipital pole, and right superior occipital gyrus) might have in common that they are associated with cognitive functions directly or indirectly. For instance, the thalamus is known to play an important role in memory processing [68,69]. It is also noted that the right frontal pole is related to verbal-auditory information processing for both working memory and long-term memory [70] and the superior occipital cortex might give an attention-based component to visual short-term memory [71,72]. Thus, volumetric reduction of such brain region could be related to cognitive dysfunction in PTSD, so it is necessary to investigate the association with cognition using cognitive tests.

Although most previous studies, which found the relationship between cortisol and brain volume, have focused on the hippocampal or amygdala volume, our study focused on not only the hippocampus but also other brain regions that are expected to be associated with symptoms of PTSD. In addition, despite overall reduced cortisol level in patients with PTSD compared to HC, cortisol level was negatively correlated with brain volumes in frontal, parietal, and occipital lobe in patients with PTSD. This implicated that PTSD is a disorder related to inflammatory processes in the prefrontal, parietal, and occipital lobe and thus, these areas could be vulnerable to inflam-mation.

Despite the implications of the results, our study has some limitations. First, we did not control the trauma type (e.g., emotional abuse, natural disaster, etc.) in patients with PTSD and demographic differences (age and education) between the two groups. Second, we used the bootstrapping method to solve the multiple comparison, but it is weaker than the conventional methods for controlling the family wise error rate such as Bonferroni corrections or False discovery rate. Third, the period of blood sampling was long to measure consistent cortisol level. Lastly, the study was conducted with a cross-sectional design and did not measure the longitudinal trajectories of the basal cortisol levels and regional brain volumes.

This study suggests that the morning basal cortisol levels were significantly lower in the PTSD group compared to HC, and it was negatively correlated with brain volume in PTSD group. Our results could demonstrate that hypocortisolism is a reliable marker in PTSD patients and that cortisol levels are vulnerability markers to brain regions in PTSD patients. It is recommended that clinicians should investigate the cortisol levels and brain volume in patients with PTSD when assessing their stress responses. Further studies with the longitudinal design may need to examine the relationship between inflammatory markers and brain volume to extend our results.