Gulf War Illness: C-Reactive Protein is Associated with Reduction of the Volume of Hippocampus and Decreased Fractional Anisotropy of the Fornix

Page 6 of 15 Gulf War Illness: C-Reactive Protein is Associated with Reduction of the Volume of Hippocampus and Decreased Fractional Anisotropy of the Fornix Peka Christova1,2, Lisa M. James1,2,3, Adam F. Carpenter1,4, Scott M. Lewis1,4, Rachel A. Johnson1, Brian E. Engdahl1,2,5, Apostolos P. Georgopoulos1,2,3,4* 1Brain Sciences Center, Department of Veterans Affairs Health Care System, Minneapolis, MN, 55417, USA 2Department of Neuroscience, University of Minnesota Medical School, Minneapolis, MN 55455, USA 3Department of Psychiatry, University of Minnesota Medical School, Minneapolis, MN 55455, US 4Department of Neurology, University of Minnesota Medical School, Minneapolis, MN 55455, USA 5Department of Psychology, University of Minnesota Medical School, Minneapolis, MN 55455, USA


Introduction
Nearly thirty years after the 1990-1991 Persian Gulf War, Gulf War veterans continue to report a variety of chronic, often disabling symptoms, commonly referred to as Gulf War Illness (GWI) 1,2 . The cause of GWI has not been conclusively identified; however, mounting evidence suggests GWI is best characterized as a neuroimmune disorder 3,4 , with several studies documenting various structural and functional brain abnormalities [5][6][7] and evidence of elevated inflammatory markers such as C-reactive protein (CRP) in veterans with GWI [8][9][10][11] . Mood and memory impairments are particularly prominent among veterans with GWI 2 , suggesting hippocampal involvement; however, the effect of CRP on hippocampal alterations in GWI has not previously been investigated.
Several hippocampal abnormalities have been associated with GWI including reductions in total hippocampal and/or hippocampal subfield volume in Gulf War veterans exposed to organophosphates [12][13][14][15] , alterations in hippocampal cerebral blood flow following cholinergic challenge 16 , and reduced hippocampal N-acetyl apartate/creatinine ratio, indicative of neuronal and/or axonal loss or dysfunction 17 . Research using animal models of GWI have reported similar hippocampal alterations [18][19][20] , and have provided evidence that systemic and hippocampal inflammation underlie memory and mood impairment associated with GWI 21 , highlighting the likely influence of systemic inflammation on hippocampal macro-and microstructure.
The use of diffusion weighted imaging (DWI), has contributed to fundamental advances in human brain anatomy by providing a window to white matter microstructure 22,23 . DWI measures water molecule diffusion along different directions including in axons where water is constrained (by the myelin sheath) to diffuse mainly along the long axis of the axon. Basser and colleagues 24,25 mathematically described the diffusion processes by a diffusion tensor, starting the field of a diffusion tensor imaging (DTI). DTI is the widely used for studying white matter changes in the healthy brain across the lifespan [26][27][28][29] , as well as in conditions affecting the brain [30][31][32][33][34][35] .
With regard to GWI, research using DTI has been limited and findings have been inconsistent. For example, Rayhan and colleagues suggested that increased axial diffusivity in the right inferior fronto-occipital fasciculus may be a biomarker of GWI 36 ; however, that finding was not replicated in a subsequent study that reported increased axial diffusivity elsewhere 37 . A study of hippocampal gray matter microstructure in GW veterans with and without suspected in-theater exposure to organophosphates found that mean diffusivity predicted suspected in-theater organophosphate exposure 15 . The authors did not find evidence of differences in fractional anisotropy (FA) or gray matter density between the two exposure groups though they noted that FA is better suited to study microstructural changes in white matter than in gray matter.
One of the major white matter outputs of the hippocampus is the fornix, a bundle of fibers that connects the hippocampus to subcortical structures in the basal forebrain and diencephalon. This hippocampal-fornix circuitry is a part of the Papez circuit 38 . The fornix and hippocampus affect each other such that the damage to the integrity of the fornix causes disruption of hippocampal function and vice versa [39][40][41] . Hippocampal functional and structural abnormalities have been shown to be associated with fornix white matter integrity disruption in various conditions shown to be associated with fornix white matter disruption in various conditions, including conditions with prominent CNS inflammation such as multiple sclerosis and Alzheimer's disease [41][42][43][44][45] .
Systemic inflammation influences blood-brain barrier (BBB) permeability 46 , and the hippocampus is particularly sensitive to BBB breakdown 47 . GWI has been associated with inflammation [8][9][10][11] , and, separately, inflammation has been linked to reduced hippocampal volume in GW veterans with posttraumatic stress disorder 48 . Finally, CRP is associated with cerebral microstructural integrity alterations as evidenced by reduced FA 49 . FA shows the degree of anisotropy and restriction of water diffusion by macromolecules and cell membranes. It varies in magnitude ranging from 0 to 1, depending on the characteristics of tissue microstructure 50 . Values close to 1 represent an almost unidirectional anisotropic diffusion, while lower values of FA can be the result of compromised white matter integrity.

Participants
Sixty-three GWI patients (59 men and 4 women) were studied after providing an informed consent in accordance with the Declaration of Helsinki. GWI status was determined using a self-report symptom checklist that permits classification as GWI case or control according to the Center for Disease Control 1 and the Kansas criteria 2 .
The Center for Disease Control definition requires one or more symptoms in at least two domains that include fatigue, pain, or mood and cognition. The more restrictive Kansas criteria requires that veterans report moderate to severe symptoms in at least 3 of 6 domains: fatigue, pain, neurological/cognitive/mood, skin, gastrointestinal, and respiratory. All GWI veterans in the present study met both case definitions. Consistent with the Kansas criteria case definition, veterans were excluded from the study if they reported medical or psychiatric conditions that could account for GWI symptoms or impair reporting 2 . Individuals with traumatic brain injury were also excluded from the study. The study was approved by the University of Minnesota and Minneapolis VA Health Care System Institutional Review Boards.

Body Mass Index (BMI)
BMI was computed using the height and weight of the participant (BMI = kg/m 2 ).

CRP
Non-fasting peripheral venous blood samples were collected for evaluation of high sensitivity C-reactive protein and analyzed using standard procedures by the Minneapolis VAHCS Clinical Laboratory.

Diffusion weighted images (DWI, DTI_medium_iso_E)
consisted of a single-shot echo-planar imaging sequence (EPI, TR=11.023 s, TE = 55 ms, Acquisition matrix 112 x 112, 70 slices with 2 mm thickness without gap, in-plane resolution = 0.875 mm × 0.875 mm). Images were acquired in the axial plane with diffusion gradients applied in 15 non-collinear directions with a b-value of 1000 s/mm 2 and one non-diffusion weighted image with a b-value of 0 s/ mm 2 . In advance of each acquisition a capsule of Vitamin E was taped to the participant's right temple to determine orientation in the imaged data.

Structural MRI
A 704-core High Performance Computing system (CentOS 6.5 Linux, Rocks 6.1.1) with Matlab R2016 (64 bit), Human Connectome Project (HCP humanconnectome. org) pipeline with FreeSurfer (FS; http://surfer.nmr.mgh. harvard.edu) HCP version (freesurfer-hpc) was used for data processing, as we describe in detail elsewhere 7 . Briefly, we used a modified version of FS, implemented in the structural HCP pipeline, which utilizes both T1-w and T2-w images to eliminate uncertainty and improve pial surface reconstruction 51 . FS segments subcortical structures by using a volume-based analysis pipeline. In this analysis subcortical regions were automatically labeled 52,53 . The volumes of left and right hippocampi were averaged to obtain the average hippocampal volume that was used in subsequent analyses.

DTI
Diffusion data were analyzed using the default parameter settings in the diffusion MR toolbox ExploreDTI, version 4.8.6 (www.exploredti.com 54 ). Anatomical T1-w images for each participant were linearly transformed into Montreal Neurological Institute (MNI) 152 space using AFNI (Analysis of Functional NeuroImages, afni.nimh.nih.gov). Diffusion weighted images were corrected for head movement and eddy current induced geometric distortions using the procedure described elsewhere 54 and corrected for EPI/ susceptibility distortion 55 . Then the image was warped to an 'undistorted' T1-w modality, in the MNI-152 template using the ELASTIX approach 56 with non-rigid registration. The above steps were applied to the data obtained from each participant and done in one interpolation to reduce blurring effects, resulting in streamlined files. The quality of each processing step was visually inspected. The validity of co-registrations was checked by overlaying the respective images for each participant.
ExploreDTI provides the Mori standard labeled atlas in the same MNI standard space (ICBM-DTI-81 57,58 ). By warping the atlas template and transforming the associate labels to each individual data set, the mean values of diffusion metric fractional anisotropy (FA) of the body and column of the fornix (called "fornix" thereafter) was obtained for each participant. The following formula was used for calculating FA: (1) where is the mean of the eigenvalues of the diffusion tensor.

Data analysis
Standard statistical methods were used to analyze the data, including descriptive statistics and linear regression. Regression coefficients were compared, as needed, using the following formula 59 : where z is the normal deviate, and denote regression coefficients to be compared, and SE are their standard errors. All statistical analyses were done using the IBM-SPSS statistical package (version 23).

Distributions of data and descriptive statistics
Age. The frequency distribution of age is shown in Figure 1; the mean ± SEM age was 55.2 ± 1.12 y.
CRP. The frequency distribution of CRP values (mg/dl) is shown in the left panel of Figure 2 It can be seen that it is skewed to the right, deviating appreciably from a normal distribution, as evidenced by the probability-probability plot in right panel of Figure 2. A logarithmic transformation was applied to bring the distribution close to normal ( Figure 3): A detailed analysis of this distribution showed the Volume of left hippocampus. The frequency distribution of the volume of the left hippocampus is shown in the left panel of Figure 4; the mean ± SEM was 4205.32 ± 60.76 mm 3 (N = 63).
Volume of right hippocampus. The frequency distribution of the volume of the right hippocampus is shown in the right panel of Figure 4; the mean ± SEM was 4284.30 ± 64.05 mm 3 (N = 63).    distribution of the FA of the fornix is shown in Figure 5; the mean ± SEM was 0.433 ± 0.012.

Associations
Effect of total intracranial volume (eTIV). eTIV did not have a significant effect on the volumes of left, right, and average hippocampus (P > 0.1 for all; linear regression); therefore, eTIV was not used in subsequent analyses.
Effect of gender. Gender did not have a significant effect on the volumes of left, right, and average hippocampus, and the FA of fornix (P > 0.5 for all); therefore, gender was not used in subsequent analyses.    Since age had a significant effect, it was added as a covariate in all subsequent analyses below. Finally, age and ln(CRP) were not significantly correlated (r = −0.146, P = 0.254).
Effect of BMI. BMI did not have a significant effect on the volumes of left, right, and average hippocampus, and the FA of fornix (P > 0.3 for all); therefore, BMI was not used in subsequent analyses.
Effect of CRP. CRP had a negative effect on the volumes of left, right, and average hippocampus, and the FA of fornix. This effect did not reach statistical significance for the left hippocampal volume but was significant for the other measures. The regression equations were as follows.

Comparison of regression coefficients for age
A regression coefficient (slope, beta) indicates the rate by which the dependent variable changes per unit change of the independent variable. When estimated under different conditions, a comparison of coefficients between conditions could provide information regarding the possible effect of the condition on the coefficient. Here we compared the coefficients for age between the left and right hippocampal volumes to find out whether the effect of age differed significantly between the left and right hippocampi. The regression coefficients for age (and their standard errors in parentheses) were −20.92 (6.45) and −20.64 (6.87) for the left and right hippocampus, respectively. Applying the formula of Equation 2 above, we get z = 0.03, indicating that the two regression coefficients do not differ significantly.
Therefore, age affects the volumes of the left and right hippocampus in the same way.
We also compared the regression coefficients for age obtained in the absence or presence of ln(CRP) as an additional covariate. A significant difference of the coefficients between the two cases would indicate a moderating effect of ln(CRP) on the age effect. In the absence of ln(CRP) as a covariate, the coefficients for age (and their standard errors in parentheses) were −20.92

Discussion
In this study, we examined the effect of inflammation, as assessed by CRP, on fornix microstructural integrity and gray matter hippocampal volume in GWI veterans. The findings demonstrated that higher CRP levels are associated with greater white matter fornix alterations and reduced hippocampal gray matter volume. The results further highlight the role of inflammation in GWI and document specific effects of inflammation on the hippocampus and fornix in GWI.
While the brain was historically thought to be immuneprivileged, it is now widely recognized that the immune system and brain interact and profoundly influence each other 60 . Under healthy conditions, the brain's resident immune cells, microglia, influence learning, memory, neural plasticity, and neurogenesis; however, chronic immune system activation impairs these processes and contributes to mood alterations [61][62][63] . GWI commonly involves memory and mood impairments 2, and is increasingly viewed as a neuroimmune condition 3,4 . Several studies have documented inflammation in GWI [8][9][10][11] with mounting evidence suggesting that pathogen exposure in those lacking immunogenetic protection contributes to GWI and associated inflammation [64][65][66][67][68][69][70] . Here we documented that peripheral inflammatory signaling is associated with white matter abnormalities and reduced gray matter volume in GWI. It has been shown that lowgrade systemic inflammation alone may be enough to lead to neurological dysfunction 71 . Systemic inflammation may also directly contribute to microcirculatory dysfunction and cerebrovascular permeability 72 and increased proinflammatory signaling within the central nervous system 73 .  Inflammation affects the brain by producing proinflammatory cytokines such as interleukin 6 (IL-6), interleukin-1β (IL-1β), tumor necrosis factor-α (TNFα), and reactive oxygen species 61,74 . Pro-inflammatory cytokines are measured in the serum in the periphery, but they are found as mediators of many central nervous system effects 75 and have been shown to increase with age, stress, and fatigue 76 . Notably, a recent study found no elevation of peripheral pro-inflammatory cytokines in GWI compared to healthy controls despite evidence of neuroinflammation 77 . Others have also failed to find evidence of increased cytokines in GWI but have reported elevated CRP 9,10 , which had historically been thought to result from increased IL-6. Reevaluation of the causality between the inflammatory cytokines such as IL-6 and CRP found that they are interrelated and that they play both pro-and anti-inflammatory roles 78 , and that each can influence the other 74 . Using a more sensitive cytokine immunoassay, a recent study found increased IL-6 and CRP in veterans with GWI 79 .
Here we found CRP levels to be significantly and negatively correlated with fornix FA values and hippocampal volume, highlighting the influence of inflammation on both fornix microstructural integrity and hippocampal macrostructure. Notably, fornix degeneration has been shown to be correlated with hippocampal volume reductions across various conditions 43,45,[80][81][82] , though the direction of the effects is unclear. For healthy adults the integrity of the fornix has been shown to predict hippocampal volume, but only for the oldest age group 44 . A longitudinal study of individuals at-risk for Alzheimer's disease found that baseline hippocampal volume predicted fornix microstructure 2 years later but the reverse -FA predicting hippocampal volume -was not supported 42 . The current study documented significant negative associations between an inflammatory marker (CRP) and hippocampal measures of volume and white matter integrity. Given the known closely interwoven associations between inflammation and neurodegeneration 83 , it is possible that the effects we observed could be due to neurodegeneration, secondary to chronic neuroinflammation. Furthermore, because FA is a fairly non-specific indicator of microstructural architecture 84 , the exact nature of the microstructural alterations in GWI remain to be elucidated.

Limitations of the study
A limitation of the study concerns the possible partial volume effect of CSF contamination that may have affected the FA measurement of the fornix. In addition, our study did not permit differentiation of inflammation and neurodegeneration, especially given their intricate interrelations 83 .