Investigating obesity‐associated brain inflammation using quantitative water content mapping

There is growing evidence that obesity is associated with inflammation in the brain, which could contribute to the pathogenesis of obesity. In humans, it is challenging to detect brain inflammation in vivo. Recently, quantitative magnetic resonance imaging (qMRI) has emerged as a tool for characterising pathophysiological processes in the brain with reliable and reproducible measures. Proton density imaging provides quantitative assessment of the brain water content, which is affected in different pathologies, including inflammation. We enrolled 115 normal weight, overweight and obese men and women (body mass index [BMI] range 20.1‐39.7 kg m‐2, age range 20‐75 years, 60% men) to acquire cerebral water content mapping in vivo using MRI at 3 Tesla. We investigated potential associations between brain water content with anthropometric measures of obesity, body fat distribution and whole‐body metabolism. No global changes in water content were associated with obesity. However, higher water content values in the cerebellum, limbic lobe and sub‐lobular region were detected in participants with higher BMI, independent of age. More specifically, the dorsal striatum, hypothalamus, thalamus, fornix, anterior limb of the internal capsule and posterior thalamic radiation showed the strongest relationship with BMI, independent of age. In a subgroup with available measurements (n = 50), we identified visceral adipose tissue to be the strongest tested link between higher water content values and obesity. Individuals with metabolic syndrome had the highest water content values in the hypothalamus and the fornix. There is accumulating evidence that inflammation of the hypothalamus contributed to obesity‐associated insulin resistance in that area. Whether brain inflammation is a cause or consequence of obesity in humans still needs to be investigated using a longitudinal study design. Using qMRI, we were able to detect marked water content changes in young and older obese adults, which is most likely the result of chronic low‐grade inflammation.


| INTRODUC TI ON
From 1975 onwards, obesity has tripled worldwide to 650 million, 1 increasing the risk of metabolic syndrome, type 2 diabetes (T2D), coronary heart disease and certain forms of cancer. 2,3 Despite tremendous efforts, an effective cure and the prevention of obesity and T2D have remained elusive. This is partly the result of a the multitude of factors that substantially contribute to the pathophysiology of obesity and T2D, including impaired insulin secretion, insulin resistance, inflammation and disproportionate body fat distribution. 4 Particularly central adiposity, a prominent trait of metabolically unhealthy obesity 4 and metabolic syndrome, is a significant source of inflammation. 5 Recent accumulating evidence shows that chronic low-grade inflammation prompts inflammatory processes in the brain, which involves non-neuronal populations such as astrocytes and microglia. 6,7 Moreover, rodent models clearly show that dietary excess can trigger brain inflammation, causing weight gain. 7,8 In humans, less is known about whether obesity is related to inflammation in the brain because it is challenging to detect subclinical brain inflammation in vivo. Thaler et al 8 were the first to describe hypothalamic gliosis in individuals with obesity using T 2 -weighted magnetic resonance imaging (MRI). Subsequently, as a result of technological advances, quantitative MRI has emerged as a tool for characterising pathophysiological processes in the whole brain in vivo with reliable and reproducible measures. 9,10 The initial evidence points to water content alteration contributing to the alterations found in individuals with obesity. 11 Proton density imaging provides a quantitative assessment of the brain water content as a marker of inflammatory processes. The higher water content (ie, increased uptake of fluid can result in local swelling) may compress the distinct local microenvironments, which may finally result in dysfunctional states, accompanying various processes that damage cells. 10,[12][13][14][15] In the present study, we investigated potential associations between measures of obesity and brain water content using proton density imaging to investigate potential inflammation of the entire brain. We hypothesise that individuals with obesity, especially those with visceral adiposity and metabolic syndrome, will show increased brain water content particularly in the hypothalamus. Based on our cross-sectional design, we cannot differentiate between diet-induced or chronic inflammation.

| Participants
The study sample consisted of 115 normal weight, overweight and obese adult participants (body mass index [BMI] range 20.1-39.7 kg m -2 , 69 men). The local ethics committee approved the protocol and informed written consent was obtained from all participants, who were recruited using broadcast emails at the University of Tübingen or through local newspaper advertisement. Participants underwent a thorough medical examination and did not suffer from psychiatric or neurological diseases. To rule out T2D, participants underwent a 75-g oral glucose tolerance test. Peripheral insulin sensitivity, 16 blood pressure and lipid profiles were additionally assessed. Metabolic syndrome was diagnosed in accordance with the International Diabetes Federation criteria. 17 It is defined by central obesity (waist circumference > 94 cm for men, > 80 cm for women) plus any two of the following risk factors: raised triglycerides, reduced high-density lipoprotein cholesterol, raised blood pressure and raised fasting glucose. Participant characteristics are summarised in Table 1 and the Supporting information (Table S1).

| Data acquisition
Studies were conducted after an overnight fast of at least 10 hours.
Whole-brain MRI was obtained by using a 3 Tesla scanner (PRISMA; Siemens, Munich, Germany) with a 20-channel head coil for signal reception and the body coil for excitation. Quantitative cerebral free water (FW) content measurements were estimated based on MR-visible-proton density (PD) with an acquisition time the dorsal striatum, hypothalamus, thalamus, fornix, anterior limb of the internal capsule and posterior thalamic radiation showed the strongest relationship with BMI, independent of age. In a subgroup with available measurements (n = 50), we identified visceral adipose tissue to be the strongest tested link between higher water content values and obesity. Individuals with metabolic syndrome had the highest water content values in the hypothalamus and the fornix. There is accumulating evidence that inflammation of the hypothalamus contributed to obesity-associated insulin resistance in that area. Whether brain inflammation is a cause or consequence of obesity in humans still needs to be investigated using a longitudinal study design. Using qMRI, we were able to detect marked water content changes in young and older obese adults, which is most likely the result of chronic low-grade inflammation.

K E Y W O R D S
cerebral oedema, hypothalamus, inflammation, obesity, quantitative MRI of 14 minutes. 10,15 For this purpose, the water content mapping protocol was implemented using the Siemens multi-slice, multiecho radio frequency (RF)-spoiled gradient recalled echo sequence (GRE). This method was previously validated in multiple cohorts (ranging from healthy controls to patients) using 1.5-T and 3-T MRI. 15,18 Two proton-density-weighted (PD-w) GRE images, with interleaved concatenations (32 slices each), were acquired with an acceleration factor of 2 and 24 auto-calibration lines, TR = 1800 ms, TE = 5.8 ms, FA = 40° and bandwidth = 210 Hz per pixel. This led to 64 gap-free transverse slices of 2-mm slice thickness and 1-mm in-plane resolution measured in 6 minutes.
To quantify tissue water content, the PD-w images need to be corrected for (i) the RF field inhomogeneities; (ii) the T 2 * -contrast; and (iii) the residual T 1 -contrast.

TA B L E 1 (Continued)
The correction for the T 2 * contrast was achieved by measuring the Correction (c) required quantification of T 1 relaxation time.
This was performed using the two-point technique. 21,22 This requires an extra T 1 -weighted GRE scan, which was acquired with TR = 500 ms, TE = 5.8 ms, FA = 90° and BW per pixel = 210 Hz.
Sixty-four transverse slices were acquired in two concatenations in 1.7 minutes.
For all sequences, the same orientation and field of view were used. The details about the parameters and the processing steps are provided in Abbas et al. 15 The free water content was estimated using in-house matlab (Mathworks, Natick, MA, USA) algorithms.
Additionally, to identify obesity-associated difference in cerebral blood flow arterial spin labelling was used, as described previously. 23 Accordingly, pulsed arterial spin labelling images were obtained with a PICORE-Q2TIPS (proximal inversion with control for off-resonance effects-quantitative imaging of perfusion by using a single subtraction) sequence using a frequency offset corrected inversion pulse and echo-planar imaging readout for acquisition. 24

| Estimation of the relaxation parameters and total free water content
Following the T 1 , T 2 * and transmit/receiver profiles corrections, the PD-weighted scan cannot yet be treated as a quantitative water content map because it requires further correction for receiver bias profile maps. 9,10 This is achieved by exploiting the linear relationship between corrected PD-weighted contrast and T 1 relaxation time in certain brain regions (60 ms > T 2 * > 50 ms). 9,15 Finally, the calibration of the bias-free water content map was performed using a robust and reliable approach. 9,15 It uses the regions within cerebrospinal fluid based on T 1 , T 2 * thresholds and its stability in terms of transmit profile. The calibration factor was then computed via the weighted average across the stable regions.
All quantitative free water content and T 1 maps were normalised using SPM12.

| Cerebral blood flow quantification
Image preprocessing was performed using asltbx 24 with spm12 (Wellcome Trust Centre for Neuroimaging, London, UK). 23 We used the general kinetic model for absolute perfusion quantification, as reported previously. 25 Perfusion images were generated by calculating the control-tag differences by using surround subtraction. For accurate CBF quantification (mL × 100g −1 × min −1 ), we used an M0 map to quantify the perfusion on each voxel.

| Body fat assessment by whole-body MRI
On a separate day, a subgroup (n = 50) underwent whole-body MRI to assess body fat distribution of the participants. These MRI examinations were performed on a 1.5-T whole-body imager (Magnetom Sonata; Siemens Healthineers, Erlangen, Germany). A whole-body imaging protocol was used to record a set of 90-120 T 1 -weighted axial slices. This approach enabled quantification of body volume, total adipose tissue (TAT) and total mass of specific fat depots such as subcutaneous (s.c. adipose tissue of the lower extremities (SCAT LE ) ranging from feet to femoral heads, and visceral adipose tissue (VAT) via an automated segmentation algorithm, applying fuzzy clustering and orthonormal snakes. 26,27

| Statistical analysis
Region-of-interests (ROIs) were selected from the Wake Forest  Table S2). In total, 66 ratios were created to compare the mean quantitative FW values within the ROIs with total grey matter or white matter tissue as a control region.
Correlation analysis was performed between the 66 ratios and BMI,

| Higher water content with increasing BMI in grey and white matter regions
In our sample of 115 adults, no significant associations of global water content of the brain with measures of obesity were observed.
Instead, we observed regional specific associations ( Figure 1). Of the different brain lobes, the cerebellum and sub-lobular region (which includes striatal regions and hypothalamus) showed higher cerebral free water content values with increasing BMI independent of age and sex (cerebellum: r sp = 0.434, P < 0.0007; sub-lobular: r sp = 0.420, P < 0.0007). Further region of interests, namely the lateral hypothalamus (r sp = 0.439, P < 0.0007) (Figure 2), dorsal striatum (r sp = 0.394, P < 0.0007) and thalamus (r sp = 0.441, P < 0.0007), showed a significant positive relationship with BMI independent of age and sex ( Table 2). The BMI-correlated increase in FW amounted to 0.006 (p.u.) per BMI point (calculated slope). Furthermore, FW values of white matter tracts showed a significant positive association with BMI independent of age and sex, most significantly the fornix (r sp = 0.419, P < 0.0007) (Figure 2), left anterior limb of the internal capsule (r sp = 0.334, P < 0.0007) and right posterior thalamic radiation (r sp = 0.346, P < 0.0007) ( Table 2).

| Individuals with metabolic syndrome display increased hypothalamic and thalamic water content values
We divided the sample of 115 individuals into three groups: individu-  Table S3). After adjusting for sex and age, correlations did not remain significant (P > 0.05, uncorrected).

| No significant association between baseline cerebral blood flow and obesity
Based on the arterial spin labelling measure, we identified no significant differences between cerebral blood flow of the different brain regions and BMI, presence of the metabolic syndrome or MR-based body fat distribution (P > 0.05) (data not shown). The mean ± SD cerebral blood flow value of the total grey matter was 33.12 ± 0.79 mL × 100 g −1 × min −1 .

| D ISCUSS I ON
Obesity and its associated comorbidities, such as insulin resistance and dyslipidaemia, are known to individually associate with changes in brain structure and function, 31,32 although the underlying cause of the observed effects remains inconclusive. In the present study, we used whole-brain proton density imaging to quantify brain water content in 115 normal weight, overweight and obese individuals. We found obesity-associated measures to positively correlate with brain water content, mainly in the subcortical and cerebellar regions and the white matter tracts connecting these regions. Individuals with obesity, particularly those with visceral obesity and altered lipid profiles, showed an enhanced water content in spatially discrete brain regions and white matter tracts, including the cerebellum, hypothalamus, striatum, thalamus, fornix, anterior limb of internal capsule and posterior thalamic radiation. In accordance with our hypothesis, the hypothalamus and white matter tracts surrounding the hypothalamus displayed the most prominent alterations in water content. However, our results additionally show that brain inflammation is not restricted to the hypothalamic area and also affects the surrounding brain regions.
This coincides with structural MRI studies showing that individuals with obesity and metabolic syndrome exemplify grey matter atrophy primarily in the cerebellum and subcortical regions 31 and show reduced white matter integrity in fibres connecting these regions. 11,32,33 Although the underlying cause still remains elusive, genetic factors, as well as cellular and cerebrovascular mechanisms, contribute to this deteriorating brain health seen in obesity. 32 In this context, inflammation has gained particular notoriety as a potential cellular mechanism leading to neuronal atrophy and compromised white matter integrity. 34 However, inflammation in the brain does not replicate the usual process seen in the periphery with the recruitment of peripheral immune cells. This is largely a result of the presence of

TA B L E 2 (Continued)
the blood-brain barrier limiting the entry of immune cells to the brain in the healthy state. 35 Instead, inflammation in the brain involves activation of glial cells (microglia and astrocytes), which are able to produce inflammatory mediators. 6,7,36 Microglial cells are considered resident immune cells of the brain, whereas astrocytes are the most numerous cells in the brain performing many functions, including the modulation of the blood-brain barrier. Animal studies using diet-induced obesity models show increased blood-brain barrier permeability and glial activation in the hypothalamus, hippocampus, amygdala, brainstem, cerebellum and cortex with increased inflammatory markers. 35 Moreover, neurones are also directly affected by hyperphagia with decreased neurogenesis 37 and less dendritic complexity. 38 These models in rodents confirm that diet-induced hypothalamic inflammation is causally related to hyperphagia and weight gain, making it a model of obesity pathogenesis. 39 In addition to diet-induced hyperphagia, chronic low-grade inflammation, caused by obesity and unhealthy fat distribution, can promote inflammatory processes in numerous tissues including the brain. 35 To detect brain inflammation in humans, we take advantage of the fact that the response of glia cells results in increased water uptake with local swelling. This can damage the cell by compressing distinct microenvironments. 18 Quantitative MRI techniques are sensitive with respect to detecting local changes in brain water content, 18,40 located in multiple tissues, including neurones, axons, myelin sheaths, extracellular space, blood vessels and glial cells. Most abundantly, water diffusion, as assessed by diffusion tensor imaging (DTI) measurements, has been used to evaluate obesity-associated changes in white matter structures. These studies reveal a negative impact of obesity on white matter microstructure regardless of age, particularly in tracts of the limbic system and those connecting temporal and frontal lobes. 33 However, it is difficult to disentangle the contribution of changes in water and myelin content in DTI metrics. To explore specific brain tissue properties, quantitative MRI techniques are implemented to specifically investigate the contribution of water content to the MRI signal. 11 Within the white matter, the initial evidence points to an increase in water rather than a decrease myelin in young adults with obesity. 11 However, very little is known about changes in water content within the grey matter. Other proposed mechanisms to explain structural and function changes in obesity include cerebrovascular mechanisms. Core features of the metabolic syndrome, such as insulin resistance and dyslipidaemia, can lead to endothelial dysfunction, along with vascular reactivity and cerebral blood flow. 32 In our present sample, however, we did not observe any changes in regional or global cerebral blood flow in individuals with obesity. This emphasises that it is most likely inflammation, rather than cerebrovascular mechanisms, that led to our current findings.
A limitation of the method used in our cohort was the presence of motion artefacts in MR images, especially in obese patients, as a result of breathing. However, the MR images were visually inspected and the datasets with no visible motion artifacts were selected for the analysis. Because of the cross-sectional design of the study, no cause-effect relationship can be established between obesity and brain inflammation, such that we cannot distinguish between brain inflammation induced by dietary excess or chronic low-grade inflammation; a sequel of obesity and excess adipose tissue. Further longitudinal studies are required to investigate whether this inflammatory process can be manipulated acutely through diet, weight loss and pharmaceutical interventions, with potential benefits for further health.
In conclusion, we postulate that chronic low-grade inflammation, as observed in obesity, metabolic syndrome and diabetes, also affects the brain, which may facilitate further weight gain and brain insulin resistance. 50 In 115 volunteers, we observed an increase in local water content in individuals with obesity. Remarkably, this local increase is specifically found in subcortical regions; further supporting the idea that inflammation in the brain may be a cause of altered brain function and structure.

ACK N OWLED G EM ENTS
This study was partly supported by a grant from the German Federal

Ministry of Education and Research (BMBF) to the German Center for
Diabetes Research (DZD e.V. 01GI0925) and the Helmholtz Alliance

ICEMED-Imaging and Curing Environmental Metabolic Diseases.
Open access funding enabled and organized by Projekt DEAL.

CO N FLI C T O F I NTE R E S T S
The authors declare that they have no conflicts of interest.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/jne.12907.

DATA AVA I L A B I L I T Y
The data that support the findings of this study are available on reasonable request from the corresponding author. The data are not publicly available as a result of privacy or ethical restrictions.