The relationship between interleukin-6 and functional connectivity in methamphetamine users

Abstract

Methamphetamine (MA) causes an increase in pro-inflammatory cytokines in animal models and in humans. Resulting activation of microglia and neuro-inflammation could, via effects on reward networks, mediate behavioral characteristics of addiction. We examined the relationship between interleukin-6 (IL-6) and corticolimbic and striatolimbic resting-state functional connectivity (RSFC). Thirty adults diagnosed with MA dependence and 20 control subjects underwent a resting-state functional magnetic resonance imaging (fMRI) scan and gave a blood sample for determination of plasma IL-6 levels. Seed-based RSFC analyses were performed to examine the interactive effect of group and IL-6 on ventral striatal and prefrontal connectivity. Within the MA group, IL-6 levels were positively related to striatolimbic RSFC but negatively related to corticostriatal RSFC. Our findings with IL-6 support the idea that inflammation may at least partly mediate the link among MA use disorder, RSFC, and behavior, possibly via effects on mesolimbic and mesocortical dopaminergic systems.

1. Introduction

Methamphetamine (MA) exposure leads to neurotoxicity and increases oxidative stress, inflammatory cytokine production and expression of factors associated with activated microglia (1–8). The innate neuroimmune response is characterized, in part, by proliferation and morphological changes of microglia and astrocytes (9). Activation of microglia results in the production of a number of pro-inflammatory markers including interleukin (IL)-1β, IL-6 and tumor necrosis factor-alpha (TNF-α), and the release of reactive oxygen and nitrogen species that cause neuronal damage (10).

Inflammatory factors target the striatum (11), and administration of inflammatory cytokines (12) and endotoxins (13) alters ventral striatal activation and corticostriatal resting-state functional connectivity (RSFC) (14), but there are no studies that link inflammation to functional connectivity deficits that are consistently seen in MA users (15–18). In nonhuman primates, chronic cytokine exposure decreases striatal dopamine (DA) release and striatal D2-type receptor availability, and the decrease in DA release is associated with reductions in the sensitivity to rewards (19). Inflammation may reduce the availability of DA precursors, as reductions in striatal DA caused by interferon-alpha (IFN-α) are restored after levodopa administration in rhesus monkeys (20). In humans, IFN-α treated patients show an increase in DA uptake and a decrease in turnover of the DA precursor [¹⁸F]-fluorodopa in the caudate and putamen (12).

Although MA renders the brain more vulnerable to inflammation and resultant neuropathology, very little is known whether systemic inflammation associated with MA promotes dopaminergic brain dysfunction in human MA users. To the extent that DA plays a role in reward processing and is linked to addiction-related brain and behavioral deficits (21, 22), it is possible that inflammation promotes such impairments through altered connectivity of dopaminergic pathways such as the corticostriatal and mesocorticolimbic DA system. This study therefore examined the relationship between IL-6 and RSFC of the ventral striatum and dorsolateral prefrontal cortex (DLPFC) in individuals with a history of MA dependence. We hypothesized that IL-6 would be positively related to RSFC between regions of the mesolimbic system and negatively related to RSFC of the corticostriatal system.

2. Materials and methods

2.1. Participants

Thirty volunteers diagnosed with MA dependence and 20 healthy controls provided written informed consent, as approved by the Veterans Affairs Portland Health Care System (VAPORHCS) and Oregon Health & Science University Institutional Review Boards (IRBs). The work was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki)(23). Participants in the MA group were recruited through word of mouth, IRB-approved advertisements, and referrals from VAPORHCS, the community, and from addiction treatment centers. Control participants were recruited by advertisement from the general public and from friends and families of those who attended the clinics associated with the study. Exclusion criteria for both groups, determined by medical history and laboratory blood tests were: systemic, neurological, cardiovascular, or pulmonary disease, head trauma with loss of consciousness, magnetic resonance imaging (MRI) contraindications, use of medications known to have dopaminergic mechanisms (e.g., antipsychotics, antidepressants, antiparkinsonian agents), sedative-hypnotics (barbiturates, benzodiazepines, zolpidem) or anticholinergics. Past or Current Axis I diagnoses, other than depression, posttraumatic stress disorder or nicotine dependence for either group, assessed with the the Mini-International Neuropsychiatric Interview (M.I.N.I.) (24), and alcohol use > 7 standard drinks per week for women and > 14 drinks per week for men were exclusionary. For inclusion in the MA group, participants met DSM-IV criteria for MA dependence and had been abstinent from MA for > 1 month and < 6 months (Note: Although the currently accepted terminology is methamphetamine use disorder, as per the DSM-5, the diagnostic category was previously termed MA dependence, and this term is used when referring to the research participants described in this paper). Urine testing on day of MRI scan verified abstinence from cocaine, methamphetamine, benzodiazepines, and opiates.

2.2. Blood sample collection and processing

Blood was drawn from non-fasting participants at rest by one-time venipuncture. Samples were collected in cell preparation tubes (BD Vacutainer Systems, Franklin Lakes, NJ, USA) containing 1 mL of 0.1M sodium citrate solution. The blood was then centrifuged at 1500 RCF for 20 minutes at room temperature (22–25° C). Plasma was separated, collected and immediately aliquoted in polypropylene tubes (Phenix Research Products, Hayward, CA, USA) and frozen at −80° C until assayed. Every effort was made to collect samples between 11 AM and 1 PM (average collection time was 11:36 AM); mean blood collection times by study group were as follows: Control group = 11:28 AM and MA group = 12:05 PM.

2.3. Multiplex immunoassay

Human plasma samples were analyzed in duplicate using a customized, high-sensitivity multiplex polystyrene bead-based immunoassay [Luminex Performance Assay (cat. no. FCST09-03); R&D Systems, Minneapolis, MN, USA] to measure peripheral immune biomarkers associated with inflammation, specifically IL-6 and IL-1β, and the anti-inflammatory cytokine, IL-10. The manufacturer-reported lower detection limits were 0.31 pg/ml for IL-6, 0.18 pg/ml for IL-1β, and 0.24 pg/ml for IL-10, but for a number of samples we were able to estimate values for samples below these limits. Our assay detection limits were 0.06 – 3,719 pg/ml for IL-6, 0.04 – 1,612 pg/ml for IL-1β, and 0.09 – 2,488 pg/ml for IL-10. In order to assess the reliability of the cytokine measurements, we calculated the intra- and inter-assay coefficients of variations (CVs) as indices of within- and between-assay precision, respectively. The average intra-assay % CVs for the cytokine measures were 8.8% for IL-6, 11.4% for IL-1β, and 9.9% for IL-10, which are comparable to, or lower than, similar multiplexed cytokine analysis (25, 26). The inter-assay % CVs (indices of plate-to-plate consistency) were calculated for low and high control samples with known cytokine concentrations. The average of the low and high % CVs were 14.5% for IL-6, 27.0% for IL-1β, and 8.2% for IL-10. This variation is comparable to previously reported inter-assay CVs using this method (25, 27).

2.4. MRI imaging acquisition

Imaging was performed on a 3 Tesla Siemens TIM Trio MRI scanner. A localizer scan was acquired in order to guide slice alignment during anatomical and functional scans. A T₂*- weighted echo-planar image (EPI) was acquired (24 slices, 4 mm thick, gap width = 1 mm, TR/TE/α = 2,000 ms/40 ms/80°, matrix = 128 × 128, FOV = 240 mm², 170 volumes, in-plane pixel size of 1.875 mm²) while subjects stared at a white cross on a black screen. One high-resolution T1-weighted anatomical magnetically prepared rapid acquisition gradient echo (MPRAGE; 144 slices 1 mm thick, TR/TE/TI/α = 2,300 ms/4.38 ms/1,200 ms/12°, FOV = 208 × 256 mm²) was acquired for co-registration with functional images and statistical overlay.

2.5. Resting-state fMRI analysis

Three participants in the MA group were excluded for incomplete scan acquisition due to technical difficulties. On the remaining subject data, image analysis was performed using FSL 5.0.2.1 (www.fmrib.ox.ac.uk/fsl). Images were realigned to compensate for motion (28), and high-pass temporal filtering (100 s) was applied. Data were skull-stripped and spatially smoothed (5-mm FWHM Gaussian kernel). Images were further pre-processed to include additional nuisance regressors: average signal of cerebrospinal fluid, average signal from white-matter, and two metrics of motion-related artifact, specifically motion scrubbing with frame-wise displacement and a combination of the temporal derivative of the time series and root-mean-squared variance over all voxels (29). Global signal regression was not applied. The EPI images were registered to the high-resolution MPRAGE image and then into standard Montreal Neurological Institute space, using a 12-parameter affine transformation. An anatomically-defined region of interest (30) from the Harvard-Oxford Subcortical atlas of the ventral striatum and a spherical ROI of the DLPFC (Montreal Neurological Institute coordinates: x = 30, y = 36, z = 20) (31) were used as seeds in two separate analyses. The seeds were transformed into each subject’s native space by applying the inverted transformation matrix of EPI to MPRAGE to standard space. In separate whole-brain, voxel-wise resting-state analyses, the mean time series across all voxels within the striatum seed and within the DLPFC seed from pre-processed images were entered as covariates.

For IL markers that showed significant group differences, between-group mixed-effects analyses were conducted with IL concentrations as regressors to test group interactions in whole-brain voxel-wise regression of striatum and DLPFC RSFC (see Section 2.6. for further description of statistical analyses). All whole-brain fMRI statistics were corrected for multiple comparisons by using cluster-correction with voxel height threshold of Z > 2.3 and cluster significance of P < 0.05.

2.6. Statistical analysis

Student’s t-tests and chi-squared (χ²) tests were used to compare groups in demographic and substance use variables. A linear regression was used to test for the main effects of group, age, and sex on differences in IL-6, IL-1β and IL-10 concentrations. In addition, given the non-normal distribution of IL-10 concentrations, a nonparametric Mann-Whitney test was conducted. Statistical analysis was conducted using SPSS 24 (Armonk, NY: IBM Corp).

3. Results

3.1. Demographics, substance use, and immune factors

Demographic and substance use variables are summarized in Table 1. There were no significant group differences in age, alcohol use, years of education or BMI, but there were significant differences in sex (χ² = 8.53, P = 0.003) and cigarette smoking status (χ² = 30.10, P < 0.001) (Table 1). Analysis of the pro- and anti-inflammatory cytokines detected no significant effects of group on IL-1β and IL-10 levels, but there was a significant main effect of group on IL-6 levels (P = 0.047), where individuals with a history of MA dependence exhibited higher levels than control participants (IL-6 mean pg/ml ± SD: MA group 0.61 ± 0.59; Control group 0.51 ± 0.39) (Table 1).

Table 1.

Demographic and clinical variables^a of research participants

	Control Group (n = 20)	MA Group (n = 30)
Age (years)	33.40 ± 11.11	37.62 ± 9.65
Sex (# male)*	9	26
Years of education	13.40 ± 2.23	12.20 ± 1.72
Body mass index	28.67 ± 4.75	26.94 ± 4.50
Alcohol use
Standard drinks per day	1.99 ± 2.44	3.43 ± 12.48
Tobacco Use (# smokers)*	2	24
Cigarettes per day (smokers)*	1.5 ± 4.89	9.21 ± 5.77
MA use
Years of use		12.04 ± 8.55
Average grams per day		1.06 ± 0.94
Months abstinent prior to MRI		4.03 ± 3.26
Immune factors, pg/ml
IL-1β	0.18 ± 0.15	0.11 ± 0.05
IL-6*	0.51. ± 0.39	0.64 ± 0.61
IL-10	0.53 ± 0.92	0.23 ± 0.20

^aData shown are means ± SD

^*Significant differences between control and MA groups (P < 0.05).

3.2. Relationship between RSFC and IL-6

Group differences in the relationship between IL-6 and RSFC are reported in Table 2. For group analyses using the ventral striatum seed, significant Group x IL-6 interactions were found (p < 0.05, whole-brain corrected, Fig. 1A), where individuals with MA dependence showed a greater positive relationship (r = 0.79) between IL-6 and RSFC between ventral striatum and amygdala, hippocampus and insula than controls (r = - 0.452). Significant Group x IL-6 interactions with the DLPFC seed showed greater negative relationships between IL-6 and RSFC of DLPFC and dorsal and ventral striatum and insula in individuals with MA dependence compared to controls (P < 0.05, whole-brain corrected, Fig. 1B).

Table 2.

Group differences in the relationship between IL-6 and RSFC

Brain region	Cluster size (voxels)	Xa	y	z	z-statistic
Greater positive relationships in the MA group with ventral striatal RSFC
Cluster #1b	1371
Occipital cortex (L/R) c		−32	−80	−10	4.67
Lingual gyrus (L/R)		−6	−78	−6	3.29
Cluster #2	1217
Middle Temporal gyrus (L/R)		54	−2	−18	4.45
Inferior Temporal gyrus (L/R)		44	−12	−20	4.25
Amygdala (R)		24	−6	−22	4.19
Hippocampus		26	−6	−24	3.26
Insula (R)		40	−4	−8	3.23
Greater negative relationship in the MA group with DLPFC RSFC
Cluster #1	1198
Lateral orbital frontal cortex		−44	44	−6	3.98
Caudate (L/R)		−12	14	−2	3.88
Ventral Striatum (L/R)		−10	14	4	2.75
Insula (L)		−36	22	−4	2.71

Z-statistic maps were thresholded using cluster-corrected statistics with a height-threshold of Z > 2.3 and cluster-forming threshold of p < 0.05.

^ax, y, z reflect coordinates for peak voxel or for other local maxima in MNI space.

^bClusters are numbered and presented in order of decreasing size.

^cL or R refers to left or right hemisphere.

Figure 1. Relationship between IL6 and parameter estimates of DLPFC and ventral striatal resting-state functional connectivity.

A. RSFC with ventral striatal seed. Group x IL6 interaction, where MA users show greater positive relationship between IL6 and RSFC between ventral striatum, amygdala, hippocampus, insula, temporal and occipital cortices (p < 0.05, whole-brain corrected). B. RSFC with DLPFC seed. Group x IL6 interaction, where MA users show greater negative relationship between IL6 and RSFC between DLPFC, orbital frontal cortex, dorsal and ventral striatum and insula (p < 0.05, whole-brain corrected). Scatter plots show parameter estimates from significant functional clusters from the whole-brain regression of DLPFC and ventral striatum seed RSFC.

For comparison purposes, and as negative controls, arbitrary anatomically-defined regions of interest (precentral gyrus and visual cortex) as defined by the Harvard Oxford Cortical Atlas were tested. Group x IL-6 interactions on striatal RSFC with precentral gyrus and visual cortex were p = 0.449 and 0.420, respectively. Also after removing two MA participants that appeared to be outliers, the results remained significant and maintained significance after controlling for sex and cigarettes per day. Post-hoc exploratory analysis with IL-1β and IL-10 levels showed no significant group interactions for IL-1β on RSFC using the ventral striatal or DLPFC seed. IL-10 levels did not interact with group on DLPFC RSFC; however, groups differed in the relationship with ventral striatal RSFC. The control group showed greater positive relationships between IL-10 and RSFC between the ventral striatum and right middle and superior frontal gyrus (p < 0.05, whole-brain corrected).

4. Discussion

This study provides support that MA-induced inflammation contributes to abnormalities in dopaminergic brain regions (32) and shows that IL-6 levels are associated with the functional connectivity of mesolimbic and corticostriatal pathways. Microglia and astrocytes are primary sources of IL-6 in the central nervous system and, when induced, IL-6 can contribute to further inflammatory signaling in brain (reviewed in (33)). MA-induced activation of microglia is dose-dependent and results in the production of pro-inflammatory cytokines including IL-6, IL-1β, and TNF-α (10). Our results are in line with studies showing that MA exposure increases IL-6 RNA levels in an astrocytic cell line (34) and that acute MA administration increases IL-6 mRNA levels in striatum (2, 35), hippocampus and prefrontal cortex (2). Neuroimaging studies provide further evidence for MA-induced inflammation, where individuals with MA dependence exhibit greater microglia activation indexed by [¹¹C] PK11195 positron emission tomography (PET) in midbrain, striatum, orbitofrontal and insular cortex (36), and lower levels of microglia activation are associated with greater duration of abstinence. Similarly, individuals with MA use disorder show an increase in inflammatory intermediates and the ratio of creatine plus phosphocreatine (Cr+PCr) and in prefrontal N-acetylaspartate (NAA) is correlated with MA use (37, 38). Furthermore, in individuals with MA use disorder, midbrain D2/3 DA receptor binding is positively correlated with gray-matter volume in striatum, amygdala, insula and orbitofrontal cortex, suggesting that the capacity for regulating DA signaling through midbrain DA autoreceptors may mediate the neurotoxic effects of MA and that DA system abnormalities can affect diffuse brain networks (39).

Consistent with this notion, RSFC studies show that individuals with MA use disorder have stronger midbrain connectivity with DA terminal regions, including striatum and limbic structures (15, 18) and stronger midbrain RSFC is associated with less DA D2 receptor availability and with greater levels of self-reported impulsivity (15). Corticostriatal RSFC is also associated with less DLPFC activation during risky decision-making (18) in individuals with MA use disorder. We extend these results to show that a peripheral marker of inflammation interacts with the functional connectivity of corticostriatal and mesolimbic systems. As the mesocorticolimbic system modulates a balance between goal-directed and reward-based behavior, the positive relationship between IL-6 and RSFC within the mesolimbic system, along with the negative relationship with corticostriatal RSFC suggests that inflammation dysregulates the connectivity between regions activated by rewarding stimuli and regions involved in executive functioning. Individuals with MA use disorder consistently show enhanced sensitivity for potential reward and diminished cortical inhibition of reward-driven responses, which may result from the altered connectivity of frontostriatal and mesolimbic pathways. This notion is consistent with a recent finding that successful treatment outcomes for alcohol use disorder is associated with RSFC of executive control and striatolimbic networks (40). The negative relationship between IL-6 and corticostriatal RSFC is also in line with findings from a study of clinically depressed individuals, where C-reactive protein and IL-6 levels were negatively associated with RSFC between ventral striatum and prefrontal cortex (14). Inflammatory cytokines increase glucose metabolism (41) and extrasynaptic levels of glutamate by decreasing glutamate transporters (42) and increasing astrocytic glutamate release (43). One potential mechanism for the effect of IL-6 on corticostriatal RSFC is altered glutamate transmission. In the context of MA, activation of mGLUR5 receptors by excess prefrontal glutamate promotes the release and translocation of nuclear transcription factors, which facilitates the transcription of inflammatory cytokines (34, 44) and may contribute to alterations in corticostriatal synaptic activity following repeated stimulant administration (45, 46). As peripheral markers of immune activation are associated with impaired cognitive functioning (4) and with abnormalities in prefrontal and striatal function (4, 14), our findings suggest that MA-mediated inflammation may promote the mesocorticolimbic and cognitive deficits that often accompany MA dependence (21).

Although this study extends our understanding of the effect of inflammation on the connectivity of neural networks implicated in addiction, it is not without limitations. Smoking could also contribute to differences in connectivity (47–49). The majority of individuals with MA dependence in this study smoked cigarettes, while only two of the controls were cigarette smokers. Results were significant with and without cigarettes per day as a covariate, but it is not completely possible to dissociate the effects of smoking. Previous studies that have controlled for smoking status, however, have found differences between individuals with MA use disorder and controls in RSFC (15, 18) and in one study amplified the group effect (50). Although beyond the scope of this study, inflammation can affect brain and behavior through various signaling pathways and through the expression of a number of cytokines and factors. This study examined IL-6 and IL-1β (and the anti-inflammatory cytokine, IL-10) but focused on measures of IL-6, as groups significantly differed in IL-6. More work is needed to interpret the effects of IL-1β and IL-10 on brain function and to evaluate the effects of other immune factors and their interactions. In addition, although results remained significant after controlling for the time between blood draws and scanning, the time between neuroimaging and inflammatory assessments need to be systematically controlled for in future studies. We used a DLPFC ROI that has shown significant group differences in brain function and connectivity between controls and individuals with MA use disorder, however, the DLPFC is a difficult region to define (51) and these results may not generalize towards RSFC of the entire middle frontal gyrus. Lastly, there were only four women with a history of MA dependence in this study, thereby limiting our power to examine the interactive effects of sex. A growing literature, however, highlights sex differences in drug use behavior (52), and we recently reported that in alcohol use disorder, there are sex differences in the effects of alcohol use history on peripheral immune-related factors [i.e., tissue inhibitor of metalloproteinase 1 (TIMP-1) and brain derived neurotrophic factor (BDNF)] (53). Future studies with more women with MA use disorder are warranted to determine the mechanisms by which inflammation may interact with sex. Despite these limitations, this study provides preliminary evidence that IL-6 contributes to RSFC abnormalities often seen in addiction.

HIGHLIGHTS.

• Interleukin 6 (IL-6) levels are higher in methamphetamine (MA) users than in controls

• Relationships between functional connectivity (FC) and IL-6 differ by group

• Striatolimbic FC and IL-6: positively related in MA users but negatively in controls

• Corticostriatal FC and IL-6: unrelated in controls and negatively related in MA users