Aberrant patterns of default‐mode network functional connectivity associated with metabolic syndrome: A resting‐state study

Abstract Introduction Metabolic syndrome (MetS) is a clustering of three or more cardiovascular risk factors (RF), including hypertension, obesity, high cholesterol, or hyperglycemia. MetS and its component RFs are more prevalent in older age, and can be accompanied by alterations in brain structure. Studies have shown altered functional connectivity (FC) in samples with individual RFs as well as in clinical populations that are at higher risk to develop MetS. These studies have indicated that the default mode network (DMN) may be particularly vulnerable, yet little is known about the overall impact of MetS on FC in this network. Methods In this study, we evaluated the integrity of FC to the DMN in participants with MetS relative to non‐MetS individuals. Using a seed‐based connectivity analysis approach, resting‐state functional MRI (fMRI) data were analyzed, and the FC measures among the DMN seed (isthmus of the cingulate) and rest of the brain voxels were estimated. Results Participants with MetS demonstrated reduced positive connectivity between the DMN seed and left superior frontal regions, and reduced negative connectivity between the DMN seed and left superior parietal, left postcentral, right precentral, right superior temporal and right superior parietal regions, after accounting for age‐ and sex‐effects. Conclusions Our results suggest that MetS is associated with alterations in FC between the DMN and other regions of the brain. Furthermore, these results indicate that the overall burden of vascular RFs associated with MetS may, in part, contribute to the pathophysiology underlying aberrant FC in the DMN.


| INTRODUC TI ON
Metabolic syndrome (MetS) is defined as a clustering of three or more cardiovascular risk factors (RF) that includes hypertension, abdominal obesity, high levels of fasting glucose (i.e., hyperglycemia), and low levels of high density lipoprotein cholesterol (HDL-C) and high levels of triglycerides (i.e., dyslipidemia) (Grundy, 2005). With overall prevalence continuing to increase every year (Beltrán-Sánchez, Harhay, Harhay, & McElligott, 2013), MetS is now considered as a substantial threat for the development of vascular-related cognitive impairment (Van den Berg, Biessels, Craen, Gussekloo, & Westendorp, 2007;Kim & Feldman, 2015;Yaffe, Weston, Blackwell, & Krueger, 2009) and neurodegenerative conditions such as vascular dementia Solfrizzi et al., 2011) and Alzheimer's disease (AD) (Misiak, Leszek, & Kiejna, 2012;Raffaitin et al., 2009). Prevalence rates have been estimated to approach 35% of the general U.S. population and increase to 54.7% of older adults over the age of 60, suggesting that older adults are disproportionally affected by the syndrome (Shin, Kongpakpaisarn, & Bohra, 2018). Moreover, given the fact that MetS is highly prevalent in middle age (Aguilar, Bhuket, Torres, Liu, & Wong, 2015;Arai et al., 2010;Grundy, 2008), there is a great need for early detection and intervention in order to prevent or delay cognitive and functional decline. Thus, it is critical to understand the relationship between the factors of MetS and the risk of brain degeneration.
The shared underlying pathophysiological mechanisms of the co-occurring RFs in MetS interrupt the cerebrovascular mechanism, which may result in disrupted structural and functional integrity in the brain. While researchers have utilized advanced neuroimaging techniques to measure effects of MetS on brain structure (Schwarz et al., 2018) and function (Haight et al., 2015;Kenna et al., 2013), the specific functional domains across the brain and their regional functional connectivity (FC) (Greicius, Krasnow, Reiss, & Menon, 2003;Van Den Heuvel & Pol, 2010) targeted by MetS are still poorly explored. Moreover, while investigators have examined individual RF's contribution to disrupted brain functions and their connectivity within isolation (Chen et al., 2014;Cui et al., 2015;Garcia-Garcia et al., 2013;Hoogenboom et al., 2014;Musen et al., 2012;Son et al., 2015;Xia et al., 2015), the shared contribution of RFs in MetS needs to be further studied in order to comprehensively understand the "comorbidity" impact.
Resting state fMRI provides important measures of brain function and can increase our knowledge of how MetS affects the brain. Exploring the default mode network (DMN) is important due to prior evidence suggesting its role as a biomarker of cognitive function and age-related decline (Greicius, Srivastava, Reiss, & Menon, 2004;Sorg et al., 2007;Zhou et al., 2010). Indeed, Zhou and colleagues have presented substantial evidence that the DMN is a robust correlate of pathological brain aging (Zhou et al., 2010).
While there have been a number of studies examining FC in association with individual risk factors, to the best of our knowledge, no studies have exclusively investigated differences in the DMN functional connectivity in those with and without MetS. In this work, we explored the combined effect of the RFs that is causing the altered FC more so than just an individual RF in isolation.
The present cross-sectional study therefore aimed to examine the FC between DMN and other brain networks, in participants with MetS. We hypothesized that compared to individuals with less than three RFs (i.e., non-MetS), individuals with MetS will demonstrate disrupted resting state connectivity between DMN and whole-brain functional networks. Results from our analyses will provide valuable information on the mechanisms by which comorbid vascular risk, a strikingly common problem facing middle and older aged adults, influence the integrity of default-mode network and its functional connectivity.
Furthermore, jointly the HC and pre-MetS individuals are referred to as the non-MetS group in the context of the group analyses. drug abuse or dependence. Exclusion criteria also included any contraindication to magnetic resonance imaging (MRI), such as a pacemaker or other metal implant.) The study was approved and monitored by the Institutional

Review Board of the Veterans Administration Boston Healthcare
System (VA), Jamaica Plain, MA, USA. All participants provided informed consent prior to study procedures.

| Risk factor measurement
Fasting blood (12 hr) was drawn and processed for analysis of serum RF levels including triglycerides, HDL-C, and fasting plasma glucose.
Systolic and diastolic blood pressure (BP) were recorded in a seated position after five minutes of rest with the arm at rest, at the level of the heart, using a standard sphygmomanometer. A second measurement was obtained five minutes later and the average of two values was recorded. Waist circumference measurement was taken while standing, to the nearest centimeter, with measuring tape placed around the abdomen at the level of the umbilicus. Medications taken to treat hypertension, diabetes, or abnormal cholesterol were reported by participants and recorded by an examiner.

| Imaging data acquisition
For 70 participants, neuroimaging data were acquired on a 3-Tesla Siemens, Erlangen, German Prisma Fit 60 cm bore (RF coil ID) using a transmission body coil and a 32-channel reception head coil. To achieve a T1 steady state, the scanner was set to automatically discard the first three volumes from the acquired data. Prior to scanning, participants were instructed to keep their eyes open and stay awake.

| Imaging data preprocessing
A model of each subject's cortical surface was reconstructed from the T1 -weighted MRI volume using FreeSurfer as described previously Lindemer, Salat, Leritz, McGlinchey, & Milberg, 2013). The surface was then anatomically parcellated into 34 distinct ROIs (which included the seed region) using the Desikan-Killiany atlas (Desikan et al., 2006;Fischl et al., 2004). Functional neuroimaging data were processed using a combination of FreeSurfer (Fischl, Sereno, Tootell, & Dale, 1999), AFNI (Cox, 1996), and FSL ( were excluded (0.5 mm/TR). Data were sampled to and smoothed on the surface, and each brain was warped to a surface-based template (fsaverage) . Seed regions were derived from surface-based parcellation of the cortex (Fischl et al., 2002).
Previous studies have shown the isthmus cingulate region to be a reliable seed to study the default network (Poole et al., 2016;Robinson et al., 2015;Seibert & Brewer, 2011). Thus, the bilateral superior third of the isthmus of the cingulate, as defined within each participant's native space, was used as a seed region. Following the FreeSurfer FSFAST processing stream, the vertex-wise partial correlation to the DMN seed was computed and used for further group-level analyses.
Briefly, to estimate FC to the DMN seed, the mean time series of the DMN seed was first correlated with all other voxels' time series in the brain, and then the measures were transformed onto the cortical surface and represented them as vertex-wise partial correlation (i.e., at each vertex over the cortical surface, more details can be found in FS-FAST processing stream: http://frees urfer.net/fswik i/FsFast). Group differences and associations (see Table 2) were computed at each vertex over the cortex using multiple linear regression using

| Demographics
The MetS group was significantly older (p = 0.007) and had a larger proportion of males (p = 0.011) relative the HC group (Table 1A).
Furthermore, the MetS group was significantly older (p = 0.014) than the pre-MetS group, although no significant difference in sex was observed (p = 0.1).     Abbreviations: cm = centimeter; mg/dL = milligrams per deciliter; HDL-C = high-density lipoprotein cholesterol; BP = blood pressure; mm Hg = millimeters of mercury.

| Additional validation: no RFs versus three or more RFs
In an effort to further validate our findings, and to examine the two groups while balancing the sample size, we conducted an additional GLM analysis in which we compared FC between the MetS group and in a subset of HC participants with zero RFs (HC N0 = 25, age (mean ± SD): 59.24 ± 8.83, 11 females) (Figure 3).
Even after excluding the participants with one and two RFs from the HC group, between-group differences in FC were observed in the same clusters as from our primary GLM analysis. Moreover, stronger group difference in FC was captured in the left postcentral cluster, where patients with MetS showed less negative connectivity to the seed (isthmus cingulate) compared to that of the HC group with no RFs.

| D ISCUSS I ON
In this study, we performed a resting state seed based functional connectivity (FC) analysis to evaluate the patterns of connectivity  between the default mode network (DMN) and the voxels in the rest of the brain. Previous connectivity studies suggest that typically in HC individuals, the DMN is positively correlated within itself and mostly negatively correlated (i.e., anti-correlated) with other taskpositive brain networks (Fox et al., 2005). In accordance with these findings, our study found that when compared with the non-MetS Additionally, we have performed the group difference analyses for the whole sample following the same processing pipeline as the main analyses without global signal regression. Results for the group difference without global signal regression showed similar patterns as observed in the main analyses ( Figure 1). Furthermore, all of the regions from the main analyses showing group differences (i.e., cluster-wise corrected group difference as seen in Figure 1 with global signal regression) were also observed from the uncorrected group difference maps estimated without global signal regression, with regions just falling below the threshold. We speculate that, since global signal regression accounts for individual variability, without global signal regression, the data will contain greater variability across individuals. Therefore, for the current dataset, the differences observed between with and without global signal regression are due to lack of statistical power, and not influenced by the global signal.
Our novel findings demonstrate altered resting state connectivity in individuals with MetS. These results suggest that multiple co-occurring vascular RFs may disrupt fundamental brain networks, particularly in frontal, parietal, and temporal regions. This is critical, From a mechanistic standpoint, it is possible that the collective burden of vascular RFs that comprise the MetS syndrome impact underlying brain vasculature, thus globally disrupting the resting state signals within and across functional neural networks. This interpretation suggests that the damaging effects of MetS on brain function are, at least in part, explained by abnormalities in the brain's vascular system. Indeed, the DMN appears to be particularly vulnerable to neurovascular compromise, and studies have demonstrated reduced cerebrovascular function in the brain regions overlapping this network (Claus et al., 1998;Dai et al., 2009;Johnson et al., 2005), providing additional support for our speculation. Our findings, together with the previous evidence, indicate that disruption in DMN connectivity, a network of primary interest in aging, neurological disease, and psychiatric disorders, (Dunn et al., 2014;Sperling, 2007), may be in part due to underlying changes to the brain's vascular system. which tend to become weaker in neurodegenerative disorders (Hafkemeijer, Grond, & Rombouts, 2012). Although, future studies are required to fully disentangle the impact of disrupted network connectivity on cognitive tasks in individuals with MetS.
Our findings are consistent with previous studies demonstrating alterations in DMN in the context of one or more of these component RFs. For example, one study reported increased connectivity among anterior and decreased connectivity among posterior DMN networks in patients with diabetes (Cui et al., 2015). Decreased connectivity in posterior nodes was also associated with worse performance on tasks of memory and executive functioning. Animal studies have also observed alterations in the DMN in rats genetically predisposed to develop hypertension (Huang et al., 2016). Another study examined the association between dynamic functional network connectivity and metabolic risks using a sliding-window clustering approach, and found that metabolic risk was associated with the relative amount of time allocated to dynamic connectivity states (Viviano, Raz, Yuan, & Damoiseaux, 2017 that is more prevalent in older adults. It is possible that this represents an uncoupled synchronization of distinct neural networks at rest, which may play an essential role in cognitive performance or in switching between mental states (i.e., task-positive vs. tasknegative). Therefore, further studies examining the relationship between functional connectivity of brain networks and cognition, particularly within the cognitive domains that are vulnerable to both vascular RFs and aging, is warranted.
Our findings are also consistent with prior studies that demonstrate alterations in brain structure and function in the presence of MetS or its component RFs (Friedman et al., 2014).
Overwhelmingly, these previous studies have demonstrated lower white matter integrity (Salat et al., 2012;Williams et al., 2013), gray matter volume (Kharabian Masouleh et al., 2018) and cortical thickness (Schwarz et al., 2018;Tchistiakova, Anderson, Greenwood, & MacIntosh, 2014), white matter volume (Figley, Asem, Levenbaum, & Courtney, 2016;Karlsson et al., 2013), task related BOLD response (Hoth et al., 2011), and cerebral blood flow (Birdsill et al., 2013)  duration and/or higher number of fMRI volumes using a shorter TR in order to reliably interpret the findings. In order to identify how MetS impacts functional brain connectivity over time, it is necessary to examine samples from a longitudinal study design with information on duration of specific RFs. Moreover, given the well-documented health disparities across different ethnic groups in the context of metabolic syndrome, future studies should also disentangle differential impact on brain health in other ethnic groups. Finally, in this study we examined a single seed, the DMN, and explored its FC to other voxels in the brain. Further consideration of other networks and ROIs will be important to understanding the full impact of MetS on brain functioning.

| CON CLUS ION
In summary, MetS is associated with disrupted DMN functional connectivity, which include both positive and negative connectivity with frontal, parietal, and temporal regions. Moreover, even after only including HC participant with no RFs for group difference analyses, altered FC measures between the DMN seed and other brain regions were observed across similar regions as found in our main analysis.
This reinforces the fundamental hypothesis of MetS criteria, where a participant is characterized as MetS only if three or more vascular RFs are reported. Also, our results showed that the mean connectivity across MetS group is not distributed based on the number of RFs for a given participant. In other words, our findings do not indicate that having three (lowest possible number) or five (highest possible number) RFs will result in higher (or lower) mean FC. This may suggest that the disruption in FC between DMN and other brain regions may arise from the underlying pathophysiology of MetS, regardless of the number of RFs. The findings from this study may allow the development of functional connectivity based biomarkers as observed between the DMN and other critical brain regions in MetS participants, which could facilitate diagnosis, targeted intervention and, in some cases, prevention of the disease.

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests in relation to the work presented.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions (funding agency: