TSPO Expression and Brain Structure in the Psychosis Spectrum

Abstract

Psychosis is associated with abnormal structural changes in the brain including decreased regional brain volumes and abnormal brain morphology. However, the underlying causes of these structural abnormalities are less understood. The immune system, including microglial activation, has been implicated in the pathophysiology of psychosis. Although previous studies have suggested a connection between peripheral proinflammatory cytokines and structural brain abnormalities in schizophrenia, no in-vivo studies have investigated whether microglial activation is also linked to brain structure alterations previously observed in schizophrenia and its putative prodrome. In this study, we investigated the link between mitochondrial 18kDa translocator protein (TSPO) and structural brain characteristics (i.e. regional brain volume, cortical thickness, and hippocampal shape) in key brain regions such as dorsolateral prefrontal cortex and hippocampus of a large group of participants (N = 90) including individuals at clinical high risk (CHR) for psychosis, first-episode psychosis (mostly antipsychotic naïve) patients, and healthy volunteers. The participants underwent structural brain MRI scan and [¹⁸F]FEPPA positron emission tomography (PET) targeting TSPO. A significant [¹⁸F]FEPPA binding-by-group interaction was observed in morphological measures across the left hippocampus. In first-episode psychosis, we observed associations between [¹⁸F]FEPPA V_T (total volume of distribution) and outward and inward morphological alterations, respectively, in the dorsal and ventro-medial portions of the left hippocampus. These associations were not significant in CHR or healthy volunteers. There was no association between [¹⁸F]FEPPA V_T and other structural brain characteristics. Our findings suggest a link between TSPO expression and alterations in hippocampal morphology in first-episode psychosis.

1. INTRODUCTION

Neuroinflammation is proposed to play a role in the pathogenesis of schizophrenia ^1–3. This idea is supported by several preclinical ⁴, epidemiological ⁵, and genome wide association studies ⁶. Moreover, clinical trials show a potential role for anti-inflammatory medication in the management of psychosis as an adjuvant therapy ⁷. Postmortem studies on brains of schizophrenia patients are, however, inconclusive due to the wide variability of their findings ⁸.

Neuroinflammation is characterised by activation of microglia, the resident macrophages of the brain. Upon activation, microglia increase the expression of 18kDa translocator protein (TSPO), in their mitochondria, making TSPO a target for positron emission tomography (PET) imaging studies. The majority of PET studies that investigated TSPO expression in psychosis thus far reported no significant group effect ^9–15; one found reduced TSPO expression in first-episode psychosis patients ¹⁶, and four reported elevated TSPO expression in schizophrenia patients or individuals at high risk for psychosis ^17–20. Three of the latter studies, however, used the first-generation radioligand for TSPO, [¹¹C]PK11195, which is known to have several limitations including low signal-to-noise ratio and brain penetration ^17–19; while the Bloomfield and colleagues study used distribution volume ratio (instead of total volume of distribution) as outcome measure which presents important limitations for data interpretation ²¹. Also, in the study by Holmes and colleagues, increase in TSPO expression was only observed in medicated schizophrenia, not in drug-naïve patients¹⁸. It is noteworthy that a recent meta-analysis of individual data of previous PET studies with second-generation radioligands in psychosis suggests an overall reduction of TSPO expression in psychosis²².

The relationship between schizophrenia, inflammation and brain anatomy is still unclear. Previous studies of brain morphometry and volumetry have reported a wide range of brain abnormalities in schizophrenia such as reduced grey matter volume, ventricular enlargement and decreased brain volume ²³, although inconsistent results have also been reported ²⁴. Although these brain deficits are suggested to be linked to factors such as neuronal loss, changes in neuroglia number, and reduced synaptic density, the mechanisms underlying these brain volume deficits in schizophrenia are not yet understood.

Microglial activation is suggested to affect brain volume. Preclinical studies and studies using cell cultures reported a link between excessive microglial activation and brain tissue destruction ^{25, 26}. Supporting this evidence, studies on other neuropsychiatric disorders such as Alzheimer’s disease, Parkinson disease, acquired immunodeficiency syndrome, multiple sclerosis and prion related disease have all shown co-localization of TSPO density and neurodegenerative alterations in the brain ²⁷. More recently some studies suggested a link between inflammation (in the periphery) and gray matter deficits in psychosis risk ²⁸. In individuals at clinical high risk (CHR) for psychosis who converted to psychosis, the gray matter loss in the frontal lobe was associated with the baseline levels of proinflammatory cytokines in plasma ²⁹. Moreover, in patients with schizophrenia a polymorphism in interleukin-1β gene was associated with gray matter volume reduction ³⁰. Similarly, a recent postmortem study showed associations between grey matter volume reduction in prefrontal cortex and gene expression of proinflammatory cytokines in schizophrenia ³¹.

In the current study, for the first time, we investigated the link between brain TSPO expression and brain volumetry/morphometry in-vivo in a large group (N=90) of first-episode psychosis patients, individuals at CHR for psychosis and healthy volunteers using the gold standard quantification method for second-generation TSPO radioligands. Based on the previous post-mortem evidence on TSPO expression and on structural alterations in dorsolateral prefrontal cortex (DLPFC) and hippocampus observed in psychosis ^{32, 33}, we hypothesized a positive association between TSPO expression and structural brain deficits (i.e. cortical thinning, decreased cortical volume, and morphometric deficits) in dorsolateral prefrontal cortex and hippocampus ²⁸.

2. MATERIALS AND METHODS

2.1 Participants

Twenty-seven first-episode psychosis patients (n = 22 antipsychotic-naïve), 35 CHR (n = 31 antipsychotic-naïve), and 28 healthy volunteers were enrolled in this study. The majority of the participants (healthy volunteers, n= 25; CHR, n = 24; and first-episode psychosis, n = 19) were also included in our previous cohorts^{10, 14}. To be eligible, first-episode psychosis patients had to have a diagnosis of one of the following psychotic disorders: schizophrenia, schizophreniform, delusional disorder, and psychosis not otherwise specified (NOS), as determined with the Structured Clinical Interview for DSM-IV (SCID) with no other concurrent axis I disorders ³⁴. CHR individuals had to meet the following criteria: fulfillment of diagnostic criteria for prodromal syndrome as per the Criteria of Prodromal Syndromes (COPS) ³⁵ with no current axis I disorders, as determined with the SCID. Healthy volunteers did not have any history of psychiatric illness, psychoactive drug use, and/or first-degree relative with a major mental illness. Participants were excluded for any of the following: clinically significant medical illness, current diagnosis of substance abuse (other than tobacco and cannabis), pregnancy or current breastfeeding, and the presence of metal implants precluding an MRI scan. In CHR, clinical status and severity of symptoms (e.g. psychosis-risk symptoms) were assessed with the structured interview for psychosis-risk syndromes (SIPS) and scale of psychosis-risk symptoms (SOPS) ³⁵. Positive and negative syndrome scale (PANSS) was used to assess severity of symptoms in the first-episode psychosis group ³⁶.

2.2 PET and MRI data acquisition and analysis

PET and MRI data acquisition have been described in detail elsewhere and are summarized below and in the methods section of the supplementary material ^{10, 11}. T1 weighted and Proton density-weighted (PD) brain MRI scans were obtained for each subject using a 1.5T General Electric Sigma scanner (General Electric Medical Systems, Milwaukee, WI, USA) for 21 participants. For the remaining participants, MRI images were acquired using a 3T MR-750 scanner (General Electric Medical Systems). All [¹⁸F]FEPPA scans were performed using a high-resolution neuro-PET camera system (HRRT, Siemens Molecular Imaging, Knoxville, TN, USA) for 125 minutes following an intravenous bolus injection of 186.21±11.39 MBq of [¹⁸F]FEPPA. Arterial blood samples were collected automatically using an automatic blood sampling system (Model PBS-101, Veenstra Instrument, Joure, Netherland) for the first 22.5 minutes after radioligand injection at a rate of 2.5 mL/min and manually at -5, 2.5, 7, 12, 15, 20, 30, 45, 60, 90, and 120 min to measure radioactivity in blood and determine the relative proportion of radiolabelled metabolites. Dispersion-and metabolite-corrected plasma input function was generated as previously described ¹¹.

2.2.1 PET image processing and calculation of total distribution volumes (V_T)

Time-activity curves were extracted for the dorsolateral prefrontal cortex and hippocampus using validated in-house imaging pipeline ROMI ³⁷. The region of interest was delineated using individual PD MRI. Kinetic parameters of [¹⁸F]FEPPA were derived from the time-activity curves using two- tissue compartment model (2TCM) and plasma input function to obtain the total distribution volume (V_T) for each region of interest, which has been validated for [¹⁸F]FEPPA quantification and described elsewhere ³⁸.

2.2.2 MRI image pre-processing and total brain volume extraction

T1-weighted MRIs were converted to the MINC file format (https://www.mcgill.ca/bic/software/minc) and preprocessed using the minc-bpipe-library pre-processing pipeline from the CoBrA Laboratory tools (http://github.com/CobraLab/minc-bpipe-library). First, a two-step whole-scan bias field correction was applied using N4ITK ³⁹. Excess data was then removed around the head to improve subsequent image processing steps (through cropping of the image volume). Finally, a brain mask was computed using the BEaST patch based segmentation technique⁴⁰. This mask was used to calculate total brain volume (TBV) and for subsequent processing steps.

2.2.3 Cortical thickness and cortical volume analysis

Cortical thickness was estimated using the CIVET pipeline (version 1.1.12; Montreal Neurological Institute, McGill University, Montreal, QC, Canada). Pre-processed images were registered to MNI- space using the MNI ICBM 152 model ⁴¹. Next, tissue was classified into gray matter (GM), white matter (WM), or cerebral spinal fluid (CSF) at the voxel level ^{42, 43}. Thickness was estimated by the distance between GM and WM surfaces using the t-link metric and blurred using a 20-mm surface based diffusion kernel ⁴⁴. Cortical surface area was calculated at the mid-cortical surface (the geometric half-way point between the GM and WM surfaces. Cortical surface area was then multiplied by cortical thickness to calculate cortical volume at every vertex. The Automated Anatomical Labeling (AAL) atlas was used to label the resampled surfaces generated from the CIVET pipeline to compute average thickness, and total volume in 90 cortical areas (including the dorsolateral prefrontal cortex). Quality control was performed by visual inspection to determine whether cortical surfaces respect anatomy.

2.2.4 Automatic segmentation of the hippocampus for volume, surface area, and displacement

2.2.4.1 Volume

Bilateral hippocampus volumes were extracted from pre-processed T1 images using MAGeT Brain, a multi-atlas segmentation algorithm ^{45, 46}. Briefly, five high-resolution atlases onto which structures have been expertly segmented ⁴⁷ were used to label a set of 21 template scans, stratified across the data. Standard model-based segmentation procedure was performed using the ANTs algorithm for atlas-to-template nonlinear registration ³⁹. All subjects were then registered to the 21 templates, yielding 105 possible candidate segmentations; final segmentations were decided by a voxel-voting procedure ⁴⁸. Following segmentation a quality control was performed by visual inspection to ensure that labels respected anatomical borders.

2.2.4.2 Vertex-wise surface displacement

Hippocampal surface displacement was used as a metric for measuring shape, and was assessed as previously described ^{49, 50}. Briefly, this is estimated by calculating the dot product between the average nonlinear deformation vector derived from the average atlas-to-subject transformation, and the surface normal. This provides a local measure of inward (concave) or outward (convex) displacement along the normal ⁵¹.

2.3 TSPO rs6971 polymorphism genotyping

The participants were genotyped for TSPO rs6971 as described elsewhere ⁵² and based on that they were categorized as high-, mixed-, or low-affinity binders.

2.4 Statistical analysis

2.4.1 Demographics

Chi-square test was used to evaluate group differences in the categorical variables (e.g. gender and TSPO polymorphism). Differences in continuous variables (e.g. age) were assessed using analysis of variance (ANOVA).

2.4.2 Vertex-wise analyses

RMINC1.4.3.2. software package (https://github.com/Mouse-Imaging-Center/RMINC) was used for analysis of group differences on vertex-wise cortical thickness measures and their relationship with [¹⁸F]FEPPA V_T. First, a general linear model was used to test whether interaction between [¹⁸F]FEPPA V_T in the DLPFC and diagnostic group was related to cortical thickness (each hemisphere was run independently). The model was constructed to test for a group by [¹⁸F]FEPPA V_T interaction, covarying for age, sex, magnet strength (1.5 vs. 3 Tesla), and TSPO rs6971 polymorphism. The group variable was releveled so that comparisons were run with the healthy volunteer group as a reference; i.e. healthy volunteers vs. first-episode psychosis and healthy volunteers vs. high risk. A False Discovery Rate (FDR) correction was applied for multiple comparisons correction (reported as q value), as the test was run at each voxel ⁵³. Next, a general linear model was used to test for group differences in cortical thickness at the vertex level (with the same covariates as above). Again, the group variable was releveled so that comparisons were run with healthy volunteers as the reference group.

Next, the relationship between hippocampal morphometry and [¹⁸F]FEPPA V_T was tested. The same general linear model with FDR correction as described above was applied to test whether the interaction between [¹⁸F]FEPPA V_T in the hippocampus and group (healthy volunteer set as the reference, CHR, and first-episode psychosis) was related to hippocampal displacement at the vertex- level. Finally, a general linear model was used to test for group differences on vertex-wise hippocampal displacement using the same covariates as above.

2.4.3 Region of interest based analysis

Multivariate analysis of variance (MANOVA) was used to examine differences in [¹⁸F]FEPPA V_T, cortical thickness and cortical volume in DLPFC, or hippocampal volume between diagnostic groups. Pearson’s partial correlations were employed to explore the association between structural measures (i.e. cortical thickness/volume in DLPFC and hippocampal volume) and [¹⁸F]FEPPA V_T in each diagnostic group while controlling for TSPO genotype. All the correlations with cortical/hippocampal volume and regional mean cortical thickness were controlled for gender and age, respectively. Bonferroni correction was utilized to account for multiple comparisons arising from the number of regions (i.e. left and right dorsolateral prefrontal cortex and left and right hippocampus). Statistical analyses were performed using SPSS (SPSS, Chicago, IL, USA). The corrected significance threshold was set at 0.0125.

3. RESULTS

Table 1 presents the characteristics of the participants. There were no significant group differences in gender, TSPO rs6971 genotype, specific activity, and mass injected. There was a significant group effect on age and post-hoc analysis revealed that both CHR and healthy volunteer groups were significantly younger than the first-episode psychosis group (both p values <0.024). A significant group effect was also found on amount injected and post-hoc analysis showed a trend toward significance between amount injected in healthy volunteer group as compared to CHR and first- episode psychosis groups. Hippocampal [¹⁸F]FEPPA V_T values of two CHR participants were excluded from the final analysis due to technical issues with the quantification of [¹⁸F]FEPPA binding.

Table 1.

Demographic characteristics of the participants and radioligand injection parameters.

Demographics		Healthy volunteer (n =28)	CHR (n =35)	First-episode psychosis (n = 27)
Age (years)		23.57±1.10	20.86±2.90	26.89±6.30	F = 13.60, P = <0.001
Gender	Male/Female	11/17	18/17	18/9	χ2= 4.15, P = 0.13
Genotype	HAB/MAB	20/8	23/12	21/6	χ2= 1.08, P = 0.58
PET measures	Amount Injected (mCi)	4.91±0.37	5.08±0.26	5.09±0.27	F = 3.31, P = 0.04
	Specific activity (mCi/μmol)	3272.64±3603.65	1691.80±1322.67	2479.66±2440.7 9	F = 3.00, P = 0.06
	Mass injected (μg)	1.38±1.24	1.79±1.38	1.17±0.61	F = 2.29, P = 0.11
Drug use (current)	Nicotine	0	5	8
Lifetime recreational history of drug use (>10 times lifetime)	Cannabis	0	15	8
	MDMA	0	1	0
	Cocaine	0	1	0
	LSD	0	1	0
	Barbiturate	0	1	0
SOPS	Total	–	35.91±10.02	–
PANSS	Total	–	–	66.33±13.56

Abbreviations: HAB, high-affinity binder; MAB, mixed-affinity binder; PANSS, positive and negative symptoms scale; PET, positron emission tomography; SOPS, scale of prodromal symptoms

3.1 Differences in [¹⁸F]FEPPA V_T, cortical thicknesses and cortical volumes across groups

We did not find any significant group effect on hippocampal [¹⁸F]FEPPA V_T (F_{(4, 166)} = 1.62, P = 0.17; Right hippocampus: F_{(2, 84)} = 1.3, p = 0.28; Left hippocampus: F_{(2, 84)} = 2.99, p = 0.06) or DLPFC [¹⁸F]FEPPA V_T (F_{(4, 170)} = 0.86, P = 0.49; Right DLPFC: F_(2,86) = 0.17, P = 0.85; Left DLPFC: F_(2,86) =0.1, P =0.91). There were no significant group differences for hippocampal displacement at the vertex-level following FDR correction for multiple comparisons.

There was no significant effect of group on cortical thickness in DLPFC while controlling for age (F_{(4, 170)} = 0.34, P = 0.85; Right DLPFC: F_{(2, 86)} = 0.21, P = 0.81; Left DLPFC: F_{(2, 86)} = 0.55, P = 0.58). There were no vertex-wise group differences in cortical thickness following FDR correction for multiple comparisons.

There was no significant group effect on cortical volume in DLPFC while controlling for gender (F_(4,170) = 0.89, P = 0.47; Right DLPFC: F_{(2, 86)} =1.37, p = 0.26; Left DLPFC: F_{(2, 86)} =0.69, p =0.51) or hippocampal volume (F_{(4, 166}) = 0.7, P = 0.6; Right hippocampus: F_{(2, 84)} = 0.5, P = 0.61; Left hippocampus: F_{(2, 84)}= 0.81, P = 0.45). These results remained after controlling for age.

3.2 Interaction between [¹⁸F]FEPPA V_T, diagnosis, and morphological measures of hippocampus and DLPFC

A significant [¹⁸F]FEPPA V_T-by-group interaction was observed in displacement measures across the left hippocampus while controlling for age, gender, magnet strength, and TSPO rs6971 polymorphism specifically in the anterior portion of the structure corresponding to CA1. In first-episode psychosis group (compared to the healthy volunteers), on the dorsal surface we observed that an outward displacement was significantly associated with [¹⁸F]FEPPA V_T (t=2.54, q=0.05) following 5%FDR correction (Figure 1). Similarly, on the ventro-medial portion of the anterior hippocampus (CA1), an inward displacement was associated with higher [¹⁸F]FEPPA V_T, suggesting that the hippocampus in first-episode psychotic patients with higher TSPO expression is shifted dorsally/laterally (t=2.54, q=0.05). [¹⁸F]FEPPA V_T was not associated with alterations in surface displacement in the healthy volunteer or in the CHR groups.

Figure 1. Significant Interaction between [¹⁸F]FEPPA V_T, diagnosis for left hippocampal displacement.

A T-statistics for outward displacement ovelaid on group average hippocampal surface (<5%FDR). The hippocampal surface is displayed both within the left himisphere for anatomical context, and on its own for better visualization of significant outward displacement of the dorsal hippocampal head. B. Linear regression showing positive relationship between increased outward displacement (mm) and increasing [¹⁸F]FEPPA V_T values only for individuals with first-episode psychosis (SCZ) for the peak displacement vertex. C. T-statistics for inward displacement of the ventro-medial hippocampal head ovelaid on group average hippocampal surface (<5%FDR). Surface display is the same is in A. D. Linear regression showing negative relationship between increased inward displacement (mm) and decreasing [¹⁸F]FEPPA V_T values only for first-episode psychosis (SCZ) for the peak displacement vertex.

There was no significant association between hippocampal volume and [¹⁸F]FEPPA V_T in any of the diagnostic groups while controlling for TSPO polymorphism and gender (all p >0.18).

[¹⁸F]FEPPA V_T-by-group interaction was not significant in predicting morphological measures of DLPFC while controlling for age, gender, magnet strength, and TSPO rs6971 polymorphism. Association between [¹⁸F]FEPPA V_T and cortical thickness and cortical volume in DLPFC are reported in the supplementary results. These results remained after removing those with previous antipsychotic use (n= 9/90), concurrent tobacco (n = 13/90), or cannabis use (n= 3/90), or controlling for age or drug use (n=3/90).

4. DISCUSSION

In this largest TSPO PET study to date, for the first time, we investigated the association between TSPO expression and morphometric brain changes in the psychosis spectrum. Our results show a significant group interaction such that there is a significant association between hippocampal displacement and TSPO expression in first-episode psychosis group, suggesting a morphometric shift in those with higher TSPO density. This association, however, was not significant in healthy volunteer and CHR groups.

TSPO expression and morphometric brain changes in dorsolateral prefrontal cortex and hippocampus

We found no significant group effect on [¹⁸F]FEPPA V_T which is in line with five previous studies that examined TSPO expression in psychosis using second generation TSPO radioligands and V_T as outcome measure ^{9–12, 14, 20} and also two other studies that used [¹¹C]PK11195^{13, 15}. A recent study, however, reported a significantly lower TSPO expression in first-episode psychosis as compared to healthy volunteer using [¹¹C]PBR28¹⁶. Supporting this, a recent meta-analysis of PET studies with second-generation radioligands in psychosis suggests an overall reduction of TSPO expression in psychosis²².

We also did not find any significant group differences in cortical thickness and cortical volume of dorsolateral prefrontal cortex. Overall the results of studies on structural brain changes in early psychosis are not conclusive which is mostly attributed to heterogeneity and the complex underlying pathophysiology of the disease ⁵⁴. Our findings are in line with recent studies on high risk populations and first episode psychosis showing no significant alterations in cortical thickness/volume ⁵⁵. While our findings can be interpreted to be in contrast with structural brain studies showing progressive reduction in cortical thickness as CHR transition to psychosis ²⁹, our study did not follow individuals in the CHR group longitudinally, so it is unclear at this stage if the changes in brain would appear as the disease progresses. We also did not find any group effect on hippocampal volume, which is in line with a recent meta-analysis of structural brain studies on CHR⁵⁶. Although, hippocampal volume loss has been repeatedly reported in patients with chronic schizophrenia ⁵⁷, the results of studies on first-episode psychosis patients are not consistent ^58–60.

Association between [¹⁸F]FEPPA V_T and morphometric brain changes in dorsolateral prefrontal cortex and hippocampus

We did not find any association between TSPO density and cortical thickness/cortical volume in DLPFC or hippocampal volume. While TSPO density and structural brain changes in psychosis have been studied separately, there is only one in vivo study investigating the association between TSPO density and morphological brain changes in psychosis ⁶¹. Selvaraj and colleagues reported a negative association between cortical gray matter volume and [¹¹C]PBR28 distribution volume ratio (DVR) in schizophrenia patients⁶¹; however, DVR as an index for [¹¹C]PBR28 binding is controversial ^{16, 21}, and the sample size per group was too small (n = 14) to provide meaningful information. Our findings seem inconsistent with a recent study that observed a negative association between the peripheral mRNA levels of IL-1β and volume of the left pars opercularis in patients with schizophrenia or schizoaffective disorder ⁶². In the same way, our results seem in contrast to a recent postmortem study which reported an association between reduction of cortical volume in prefrontal cortex and expression of inflammatory cytokines in the brain of schizophrenia patients ³¹. However, we measured brain TSPO density which was shown to be independent from inflammatory markers such as cytokines in the periphery or cerebrospinal fluid ^{9, 63}. Moreover, recent studies suggest that alteration of TSPO expression may also reflect other abnormalities including change in metabolism at the cellular level and oxidative stress ^64–66.

In the first-episode psychosis group, we observed associations between [¹⁸F]FEPPA V_T and outward and inward displacements, respectively, in the dorsal and ventro-medial portions of the left hippocampus. These associations were not significant in healthy volunteer and CHR groups. This suggests that alterations due to TSPO expression may be subtle, especially at the initial stages of illness, which may explain why we did not detect them at the volume level. Hippocampal shape deformities were previously reported in both first-episode psychosis and CHR groups ⁶⁷. Brambilla and colleagues reported an association between shape deflation of hippocampus and severity of symptoms in schizophrenia ⁶⁰. In a recent study in CHR, Dean and colleagues observed a positive association between inversion of the left ventral posterior hippocampus and severity of prodromal symptoms ⁶⁸. Supporting our findings, Schobel and colleagues reported a significant volume loss localized to the subiculum and CA1 subregions of hippocampus in high risk individuals who converted to psychosis as compared to non-converters ⁶⁹.

Overall, our findings support the developmental model of psychosis such as the neonatal ventral hippocampal lesion and methylazoxymethanol acetate (MAM) models which consider a central role for hippocampus dysfunction in the development and progression of the disease ^70–72. Based on our findings, TSPO expression could be one of the contributing factors to hippocampal abnormalities in psychosis.

This study has some limitations that should be taken into consideration when interpreting findings. First, our first episode psychosis group was significantly older than our CHR and healthy volunteer groups. However, the mean age of first episode psychosis group was only 3 and 6 years more than healthy volunteer and CHR groups, respectively. Nevertheless, there is no evidence for a significant age effect on TSPO expression ⁷³. Furthermore, age was included as a covariate in our statistical analyses. Second, amount injected was significantly different across the groups; however, there was no association between amount injected and [¹⁸F]FEPPA V_T in our sample. Third, although the PET and structural MRI scans were done on two separate days (mean interval ~15 days), both [¹⁸F]FEPPA PET scan and structural MRI scans were shown to have satisfactory test-retest reliability. Fourth, while antipsychotic medications are known to have a significant effect on structural brain changes in psychosis⁷⁴, and it is considered an important limitation of similar studies on structural changes in psychosis; most of our participants in CHR (n = 31) and first-episode psychosis (n = 22) were antipsychotic naïve. Fifth, TSPO is not specific for microglia and it is also present on astrocytes and endothelial cells ⁷⁵. However, both of these cell types are involved in the immune response and it would not alter the general conclusion of the current study. Lastly, some of the participants in the CHR (n = 1) and first-episode psychosis (n = 2) groups had a positive urine drug test for cannabis. Our results, however, remained after excluding these participants from the analysis.

The findings of this study suggest that TSPO expression in first-episode psychosis patients is associated with morphological brain changes in hippocampus. This association is not observed in the clinical high risk that precedes psychosis.

Highlights.

• Neuroinflammation and abnormal brain structure alteration are implicated in the pathophysiology of psychosis

• In first-episode psychosis, we observed associations between TSPO expression and outward and inward morphological alterations of hippocampus

• There was no association between TSPO expression and cortical thickness or cortical volume