Correlations Between Peripheral Levels of Inflammatory Mediators and Frontolimbic Structures in Bipolar Disorder: An Exploratory Analysis

Abstract

Introduction

Altered peripheral immune/inflammatory system and brain volumetric changes have been implicated in the pathophysiology of bipolar disorder (BD). This study aimed to evaluate how peripheral levels of cytokines are related to volumetric brain changes in euthymic patients with BD.

Methods

Euthymic patients with BD (n=21) and healthy controls (n=22) were enrolled in this exploratory study. Blood samples were collected on the same day of clinical assessment and neuroimaging. Cytokines were measured through cytometric bead array (CBA) method. Neuroimaging data were acquired using a sagittal three-dimensional Magnetic Resonance Imaging (MRI) T1-weighted fast field echo sequence and was processed using FreeSurfer.

Results

Compared to controls, BD patients had significantly lower volumes in the cingulate, medial-orbitofrontal (MOF) and parahippocampal regions. We found a negative correlation between right MOF volume and interferon-gamma (IFN-γ) levels (β −0.431, p=0.049) and a positive correlation between IL-10 levels and left posterior cingulate volume (β 0.457, p=0.048).

Conclusion

Our results support the involvement of inflammatory pathways in structural brain changes in BD.

Background

The pathophysiology of BD is complex, involving genetic, epigenetic and environmental factors^1,2. Current theoretical frameworks postulate that these factors underlie neuronal changes (e.g. hyperexcitability), further leading towards: a) circuitry and brain matter changes (e.g., respectively, default mode network alterations and gray matter loss); b) systemic consequences (e.g. increased inflammation and oxidative stress); and c) behavioral correlates (e.g. changes in energy and activity levels)³. Together these pathways determine functional outcome in BD, commonly marked by cognitive deficits and poor psychosocial adjustment^3–5.

Over the years, BD has been linked to peripheral and central nervous system immune and/or inflammatory mechanisms^6–8. Cytokines – molecules responsible for immune cells interactios – have been reported to be increased in patients with BD, especially during mood episodes^9–12, including IL-1β, IL-4, IL-6, IL-10, TNF-α, and IFN-γ^8,13–15.

Regarding structural neuroimaging findings, reductions in brain grey matter (GM) volumes and thickness in several regions have been consistently reported in BD^16–18. Specific cortical areas found with low GM were frontal regions, including orbitofrontal cortex, inferior frontal cortex, pars orbitalis, middle frontal cortex, right ventral prefrontal cortex, right fronto-insular cortex,^19–22; occipitotemporal regions, including lingual gyrus, planum polare, temporal pole, and temporal gyrus,^19,20,22,23, and other cortical regions, like parahippocampal gyrus and cingulate cortex^22,24,25, and subcortical regions, namely insula, claustrum, hippocampus, and thalamus^20,26.

Despite the evidence pointing out the involvement of both peripheral inflammation and structural brain changes in BD, only a few studies have evaluated the potential link or association between these two different features of BD pathophysiology. Apart from few studies showing a link between volumetric alterations and IL-6^19,27, there is a dearth of research investigating potential associations between inflammatory biomarkers and volumetric changes in BD²⁸. Our study aimed at exploring potential correlations between peripheral inflammatory biomarkers and frontolimbic volumes in patients with BD.

Methods

Study Sample

This study was performed at the UT Center of Excellence on Mood Disorders, Houston, TX. Twenty-one patients with BD type I were recruited from the associated outpatient clinic. Healthy controls (n=22) were recruited through flyers placed in the community.

The inclusion criteria for cases and control were a) Age 18–65 years old; b) any race/ethnicity; c) fluency in English; d) have the capacity to give informed consent. Additional specific inclusion criteria for the patient group were a) Bipolar I disorder; b) euthymic mood state, identified as lack of any active symptoms of severe depression or mania or psychosis (based on the clinical scales mentioned below). The inclusion criteria for controls incorporated a) no lifetime diagnosis of major psychiatric and neurologic disorders; b) no family history of any axis-I psychiatric disorder and/or hereditary neurologic disorders in first-degree relatives. For both cases and controls, the exclusion criteria were a) any current major medical problems that could potentially affect brain anatomy or function; b) history of any central nervous system diseases; c) family history of hereditary neurological disorders; d) pregnancy; e) substance abuse or dependence; f) any contraindications to magnetic resonance imaging (e.g., floating metallic objects in the body). The Structured Clinical Interview SCID for DSM-IV²⁹ was administered to confirm the diagnosis of BD among patients and the absence of a psychiatric diagnosis in controls. Written informed consent was obtained from patients and healthy controls. The UT Health Institutional Review Board approved the study protocol.

Demographic data, including age, sex, weight, height, and education were obtained. Any potential confounding effects of cognitive deficits were ruled out through our exclusionary criteria and cognitive assessment with the MoCA³⁰. Only patients performing above the cutoff points in the general cognitive assessment scale were recruited for the study.

Clinical scales

To assess clinical variables, the following scales were used – a) The Clinical Global Impressions (CGI) Scale with subscales of CGI-1 = severity of mania, CGI-2 = severity of depression, CGI-3 = severity of overall illness; b) Montgomery–Åsberg Depression Rating Scale (MADRS) for depressive symptoms; c) Young Mania Rating Scale (YMRS) for manic symptoms d) Behavioral Inhibition System (BIS) scale for aversive motives; and e) Functioning Assessment Short Test (FAST) for functional impairment.

Blood work

Blood samples were collected in vacuum tubes, centrifuged twice at 1,800g, and samples were stored in −80° C freezer for further processing. Plasma levels of interleukins (IL-1β, 2, 4, 6, 8,10), TNF-α, IFN-γ were assessed through cytometric bead array (CBA). The concentrations were measured in pg/ml. All imaging and plasma analyses were performed blindly to the clinical status of the participants.

Neuroimaging

Based on current literature, frontolimbic regions were assessed to compare between BD patients and controls. All structural neuroimaging scans were acquired using a 3.0 T magnetic resonance imaging (MRI) Philips scanner (Philips Medical System, Andove, MA) using a sagittal three-dimensional T1-weighted fast field echo sequence with the following parameters: Repetition time (TR) = 8.0 ms, echo time (TE) = 3.68 ms, flip angle = 6°, field of view (FOV) = 288 mm, slice thickness =0.88 mm, matrix size = 288×288 and 180 slices. Cortical thickness and grey matter volumes measures were acquired using FreeSurfer suite version 6.0^31–33. FreeSurfer is a fully automated software that performs motion correction, intensity normalization, automated topology correction, and atlas-based cortical segmentation and labeling of MRI images³⁴.

Analyses

The statistical packages STATA³⁵ and SPSS³⁶ were used for the statistical analyses. Student t-test and chi-square test were used for the statistical inference regarding the comparison of both groups on the variables of interest. Linear regression was used for the correlation analysis between volumetric data and cytokines, with age as a covariate, and further to assess the correlation between clinical data with significant volumetric and cytokine levels.

Results

Both groups had similar demographic profiles and, as expected, all psychopathological scores were significantly higher in BD patients compared to controls (Table 1). At the time of the study, 23% of the BD group were using lithium, 27% were taking anticonvulsants, 50% were on antidepressants, 5% were on atypical and 14% were on typical antipsychotics.

Table 1:

Demographic and clinical comparison between patients with BD and controls

	Cases (n=21)	Controls (n=22)	P
Demographic comparison
Age (Mean ± SD) in years	33.9 ± 9.17	33.91 ± 9.99	0.999
Sex [Male n (%)]	7 (33.3)	9 (40.9)	0.607
Education (Mean ± SD) in years	14.8 ± 2.84	13.45 ± 1.97	0.086
BMI (Mean ± SD)	27.86 ± 7.01	30.17 ± 7.29	0.349
Clinical comparison	Mean ± SD	Mean ± SD
CGI-1	2 ± 0.95	1 ± 0	< 0.001
CGI-2	2.52 ± 1.12	1 ± 0	< 0.001
CGI-3	2.86 ± 0.96	1 ± 0	< 0.001
MADRS	12.1 ± 9.76	0.14 ± 0.47	< 0.001
YMRS	6.00 ± 6.29	0.05 ± 0.213	< 0.001
BIS	54.43 ± 11.04	37.18 ± 10.67	< 0.001
FAST	29.84 ± 15.3	3.5 ± 4.67	< 0.001

Footnotes:

CGI- Clinical Global Impressions, CGI-1 = severity of mania, CGI-2 = severity of depression CGI-3 = severity of overall illness, MADRS- Montgomery–Åsberg Depression Rating Scale, YMRS - Young Mania Rating Scale, BIS-Behavioral Inhibition System scale, FAST- Functioning Assessment Short Test; the Student t-test and the chi-square test were used for the statistical inference; a significance level of 0.05 was adopted

No significant differences were found between BD patients and controls with regards to any of the assessed cytokine levels (Table 2).

Table 2:

Serum cytokine level differences between patients with BD and healthy controls

Serum cytokine levels (pg/mL)	Cases (n=21) Mean ± SD	Controls (n=22) Mean ± SD	p
IFN-γ	0.708 ± 0.269	0.766 ± 0.447	0.609
IL-1b	8.319 ± 2.428	7.856 ± 2.913	0.575
IL-2	2.000 ± 0.567	2.274 ± 1.050	0.297
IL-4	2.957 ± 1.059	3.048 ± 1.850	0.847
IL-6	5.350 ± 2.946	4.303 ± 1.727	0.161
IL-8/CXCL8	6.974 ± 2.452	6.300 ± 2.478	0.375
IL-10	12.897 ± 6.274	13.100 ± 7.805	0.925
TNF-α	9.716 ± 2.155	9.040 ± 2.587	0.359

Footnotes:

IFN-γ: interferon-gamma, IL: interleukin, CXCL8: C-X-C Motif Chemokine Ligand 8, TNF-α: tumor necrosis factor-alpha

With respect to the volumes of the brain structures analyzed, BD patients had significantly lower GM volumes in i) left Caudal-Anterior-Cingulate; ii) left Pars orbitalis; iii) left Posterior-Cingulate; iv) right Medial-Orbito-Frontal; v) right Parahippocampal; and had lower WM left Pars orbitalis volume (Table 3). Effect size for all the significantly lower volumes were at least medium (p < 0.05, d > 0.5). No other volumetric differences were found in frontal, hippocampus or amygdala regions between BD patients and healthy controls.

Table 3:

Frontolimbic differences between patients with BD and healthy controls

Structural regions (volume in mm3)	Cases (n=21) Mean ± SD	Controls (n=22) Mean ± SD	p	Cohen’s d
LH Caudal-Anterior-Cingulate	1.109 ± 0.208	1.295 ± 0.303	0.03	0.716
RH Caudal-Anterior-Cingulate	1.521 ± 0.365	1.516 ± 0.364	0.96	0.014
LH Pars Orbitalis	1.397 ± 0.261	1.583 ± 0.261	0.02	0.713
RH Pars Orbitalis	1.752 ± 0.277	1.853 ± 0.299	0.26	0.35
LH Posterior-Cingulate	2.001 ± 0.272	2.251 ± 0.447	0.03	0.676
RH Posterior-Cingulate	2.241 ± 0.346	2.183 ± 0.447	0.63	0.145
LH Medial-Orbito-Frontal	3.354 ± 0.457	3.470 ± 0.459	0.41	0.253
RH Medial-Orbito-Frontal	3.204 ± 0.355	3.491 ± 0.564	0.05	0.609
LH Parahippocampal	1.541 ± 0.271	1.635 ± 0.256	0.25	0.357
RH Parahippocampal	1.417 ± 0.200	1.583 ± 0.211	0.01	0.807
WM LH Parsorbitalis	0.566 ± 0.120	0.668 ± 0.153	0.02	0.742

Footnotes:

LH – left hemisphere, RH- right hemisphere, WM- white matter the Student t-test was used for the statistical inference; a significance level of 0.05 was adopted

With a regression model adjusted for age as a covariate in regions significantly different between cases and controls, a significant negative correlation was found between the RH Medial-Orbito-Frontal volume and IFN-γ (β −0.431, p=0.049), as well as a significant positive correlation between IL-10 and LH Posterior-Cingulate volume (β 0.457, p=0.048) in BD patients (figure 1, figure 2). No significant correlations were reported between any other cytokine and other brain structures in BD patients or healthy controls.

Figure 1: Correlation between IFN-γ and medial orbitofrontal gray matter volume between patients with BD and controls.

Footnotes:

Volumes in mm³, BD- bipolar disorder, IFN-γ - interferon-gamma

Linear regression was used for the statistical inference, with age as a covariate. A significance level of 0.05 was adopted

Figure 2: Correlation between IL-10 and medial posterior cingulate gray matter volume between patients with BD and controls.

Footnotes:

Volumes in mm³, BD- bipolar disorder, IL-10- Interleukin 10

Linear regression was used for the statistical inference, with age as a covariate. A significance level of 0.05 was adopted

Discussion

In our exploratory study, we found that peripheral immune mediators correlate with frontolimbic changes implicated in the pathophysiology of BD. Specifically, we identified negative correlation between medial-orbito-frontal volumes and levels of IFN-γ, a pro-inflammatory cytokine, while posterior cingulate volumes were positively correlated with the levels of the anti-inflammatory cytokine IL-10.

The relationship between immune–inflammatory pathways and mood disorders has been extensively investigated^37–39. While previous studies have shown increased levels of pro-inflammatory cytokines in BD, this evidence has been more consistent in patients with active bipolar depression, mania/hypomania or mixed mood states^9–12. For example, high levels of IL-6 have been reported during mania/hypomania, with IL-6 returning to baseline levels after treatment with mood stabilizers^40–42. We did not find significant differences between patients and controls in the levels of citokynes. Actually, this lack of difference might reflect the fact that our sample comprised only clinically stable euthymic outpatients.

As expected, there were significant differences between patients and controls in volumes of frontolimbic regions implicated in emotion processing and BD pathophysiology^43,44. Even with small sample size in this study, the effect size showed significant difference in BD patients. Chen et al. reported volumetric differences in regions like orbitofrontal cortex, lingual gyrus, inferior frontal cortex, middle frontal cortex, and planum polare, which seem to be significantly smaller in patients with BD compared to controls¹⁹. In the same study, the volume of certain structures (i.e. orbitofrontal cortex, middle frontal cortex, inferior frontal cortex, cingulate cortex, lingual gyrus, superior temporal cortex and Heschl’s gyrus) were negatively correlated with IL-6 receptor levels, suggesting that high levels of pro-inflammatory mediators might affect brain structure¹⁹. Bai et al. also reported that levels of pro-inflammatory mediators (IL-6 and sTNF-R1) are associated with greater reductions of 12 GM volumes (R. cerebellar lobule, R. putamen, L. putamen, R. superior frontal gyrus, L. lingual gyrus, L. precentral gyrus, R. fusiform gyrus, L. calcarine, R. precuneus, L. inferior temporal gyrus, L. hippocampus, L. superior frontal gyrus) in BD patients⁴⁵. Our finding of a negative correlation between right medial-orbito-frontal volumes and the levels of IFN-γ is in line with these previous reports.

Our results also show positive correlation between left posterior cingulate cortex volumes and IL-10 levels in BD patients. IL-10 is an anti-inflammatory cytokine and has been found to be increased in the early stages of BD^15,46. Actually, the increase of IL-10 follows a heightend inflammatory response as a compensatory mechanism⁴⁷. While the involvement of cingulate cortex in emotional processing and pathophysiology of mood disorders has been well-described, there is no clear consensus about the function of posterior cingulate cortex (PCC) within system-level models of brain function⁴⁸. One hypothesis is that PCC plays a central role in supporting internally directed cognition as it shows increased activity when individuals retrieve autobiographical memories or plan for the future⁴⁹. PCC also plays a direct role in regulating the focus of attention and the balance between internally and externally focused thoughts⁵⁰.

Altogether all results suggest that inflammation, as evidenced by pro- and anti-inflammatory cytokines, might be involved in progressive structural changes in brain areas implicated in emotion processing and pathophysiology of BD. Interestingly, an ‘allostatic load’ theory has been proposed to explain the progressive changes in BD, also called ‘neuroprogression’, based on the physiological ‘wear and tear’ in response to multiple stressors, and associated with increased peripheral inflammation among other biological processes⁵¹.

Our study had several limitations. First, all patients were medicated at the time of their participation in the study. The lack of significant differences in cytokine levels between patients and controls can also be attributed to different effects of medication on cytokines^52–54.

Literature from studies in rodents and human cell lines have showed effect of psychotropic medications on brain volume changes in BD. While lithium and valproate have shown neurotrophic effects in vitro, antipsychotics; especially first-generation ones have been associated with brain volume loss⁵⁵. Specifically, ACC region has been found with significantly lower volume in earlier phases of BD⁵⁶. Given our sample size and euthymic status of our patients, it was not possible to assess impact of psychotropic medications in our patients.

Second, the small sample size is another limitation of our study, preventing us from carrying out stratified analyses. Third, the cross-sectional nature of study limited our ability to infer causal relationships between inflammatory biomarkers and GM volumes. Taking into consideration that the inflammatory cytokines per se could be influenced by many factors including body weight and other medical comorbidities, it will be important to follow the trajectory of both inflammatory biomarkers and GM volumes in drug-naïve and drug-free patients.

In summary, our findings corroborate the importance of the complex interactions between inflammation and structurcal brain changes in the pathophysiology of BD. Future studies including patients in different mood states and adopting longitudinal designs are of pivotal importance to further elucidate this complex relationship and its possible implications for a better understanding of BD, its pathophysiology, and ultimately its management.