Abstract
In groups of patients suffering from psychosis, redox dysregulation was reported in both peripheral fluids and brain. It has been hypothesized that such dysregulation, including alterations of the glutathione (GSH) cycle could participate in the brain white matter (WM) abnormalities in schizophrenia (SZ) due to the oligodendrocytes’ susceptibility to oxidative stress (OxS).
In this study we aim to assess the differences between 99 psychosis patients (PT) and 86 healthy controls (HC) in GSH-redox peripheral blood markers: GSH peroxidase (GPx), reductase (GR) enzymatic activities and their ratio (GPx/GR-ratio), evaluating the hypotheses that alterations in the homeostasis of the systemic GSH cycle may be associated with pathological mechanisms in the brain WM in PT. We also consider the hypothesis that different homeostatic states of the system (oxidative vs reductive) may reflect differently into the WM microstructure. To do so, we employ the advanced diffusion MRI methods: Diffusion Kurtosis Imaging (DKI) and White Matter Tract Integrity-Watson (WMTI-W), which provide excellent sensitivity to demyelination and neuroinflammation.
We show that GPx levels are higher (p=0.00050) in female participants and modulate the effect of aging on the WM. We found differences between PT and HC in the association of GR and mean kurtosis (MK, p=0.028). Namely, lower MK was associated with higher blood GR activity in HC, but not in PT, suggesting that high GR activity (a hallmark of reductive stress) in HC was linked to changes in myelin integrity. Remarkably, GSH-redox peripheral blood markers did not explain the WM anomalies detected in PT.
Introduction
Psychosis and schizophrenia (SZ) present significant challenges for patients, their families, and society1. Only partially understood, its etiology is attributed to a multifaceted interplay of genetic, environmental and developmental factors that affect a variety of biological processes2–7 among which the redox balance8. In patients suffering from psychosis, a redox dysregulation is reported in peripheral fluids (i.e., diminished antioxidant capacity, oxidative damage of lipids, DNA and RNA9–11) and brain (i.e., lower glutathione (GSH), lower NAD+/NADH ratio12,13). The causes of such redox dysregulation and the resulting oxidative stress (OxS) are multifactorial, including compromised energy metabolism, endocrine dysregulation, chronic subclinical inflammation and weakened antioxidant capacities. Anomalies in the GSH redox system, a key antioxidant system for cell protection and detoxification, have been widely observed in SZ9. While polymorphisms in genes of the GSH system have been associated with SZ14–19, meta-analyses also show lower mean levels of GSH in blood9, and brain13,20–22 in patients vs. controls. The redox regulation and antioxidant function of GSH acts largely via the GSH redox cycle that involves the GSH peroxidase (GPx) and GSH reductase (GR). GPx neutralizes peroxides by oxidizing GSH into GSH-disulfide (GSSG). In turn, GR reduces GSSG back to GSH. Thus, GPx and GR activities and the GPx/GR ratio inform about the homeostasis of this system. Two out of three postmortem studies have found lower brain GPx in patients vs. controls20–22. Meta-analyses suggest that SZ patients tend to display also lower blood GPx than controls9. The decrease in peripheral GPx activity seems, however, mostly found in chronic patients23, with variable levels among early psychosis (EP) patients. Thus, EP patients with worse psychosocial outcomes after a 3-year clinical follow-up displayed significantly higher GPx activity at baseline than those with better outcome24. By contrast, there is no evidence of consistent alterations of GR in brain20 and blood9.
It has been hypothesized that a GSH deficit could participate to the brain white matter (WM) abnormalities in SZ25,26, consequential to the oligodendrocytes susceptibility to OxS27, due to their elevated metabolic activity and high iron content for the production and maintenance of the myelin sheaths28,29. The GSH redox cycle is involved in the integrity of WM, from the survival, proliferation and differentiation of the oligodendrocyte precursor cells26,30 (OPC), to the protection of oligodendrocytes against OxS thanks to high GPx expression31,32. A dysregulation of the GSH system will also alter axons’ energy metabolism33 and mitochondrial activity34. Furthermore, this dysregulation can lead to subclinical activation of microglia35, which may in turn affect WM, due to its role in the WM myelin remodeling via OPC recruitment and phagocytosis of myelin debris, de-myelination and remyelination36. Post-mortem studies on SZ patients have described diminished oligodendrocyte densities in cortical layer III and subcortical areas, presence of swollen or dystrophic oligodendroglia37–39, adjacent microglia40, decompacting and splitting of the myelin sheath with inclusions of vacuoles between myelin layers37–39.
Abnormal WM integrity in EP is also widely reported using in-vivo diffusion magnetic resonance imaging (dMRI). DMRI leverages the random motion of water molecules to infer the microstructural properties of the underlying biological tissue41. Most studies investigating the psychosis spectrum show widespread increased mean diffusivity (MD) and/or reduced fractional anisotropy (FA)42–47 with medium effect sizes46 (Cohen’s d=0.42), consistent with anomalies described in the post-mortem studies. However, the evidence of a link between WM anomalies and the GSH system in subjects with psychosis remains sparse. GSH levels in frontal brain regions have been found to correlate positively with generalized FA (gFA) in the cingulum of both EP patients and controls26, while it was shown in earlier work that the precursor of GSH and antioxidant, N-acetylcysteine (NAC), improves the integrity of the fornix (i.e., increases gFA) in EP patients48 and increases brain GSH49.
Here, we aim to assess in patients with psychosis whether the activity of GPx, GR, and the GPx/GR ratio measured in blood cells (which inform about the systemic homeostasis of the GSH redox cycle) are associated with the WM microstructure estimated using dMRI. For this purpose, we utilized advanced dMRI methods: Diffusion Kurtosis Imaging50,51 (DKI), an extension of the diffusion tensor imaging (DTI), and the biophysical model White Matter Tract Integrity-Watson52,53 (WMTI-W). In addition to classical DTI measures such as MD and FA, DKI enables the estimation of the mean kurtosis (MK), a measure of diffusion variance within each voxel arising from heterogeneous sub-domains, such as cellular compartments, which provides complementary information about tissue heterogeneity. The WMTI-W model allows the estimation of specific microstructural cellular features such as the axonal density (f) in each voxel, with less ad hoc simplifying assumptions, less fit constraints54,55 and increased validity55 as compared to other biophysical methods such as free-water imaging56 (FWI) and Neurite Orientation Dispersion and Density Imaging54 (NODDI). We focused our analysis on these four metrics (MD, FA, MK, f) because of their ability at detecting abnormal myelination or demyelination57–60 (↑MD, ↓FA, ↓MK, ↓f), but also neuroinflammation which has the effect of reducing diffusivities (↓MD, ↓FA) and increasing kurtosis and apparent axonal water fraction (↑MK, ↑f) due to higher cellular crowding associated with microgliosis and astrocytosis57,58,61,62. In a recent study, we have shown an increase in MD and a decrease in FA, MK and f in the WM of patients with psychosis when compared to age range-matched controls, with these alterations being surprisingly more pronounced during the first years of psychosis63.
Here we hypothesize that alterations in the homeostasis of the systemic GSH redox cycle may be associated with pathological mechanisms in the brain (i.e., redox dysregulation and microglia dysfunction) that affect WM. Therefore, we expected that peripheral markers related to the GSH redox cycle may correlate with some of the WM microstructure metrics, sensitive to neuroinflammation and myelination, as previous treatment studies targeting the GSH redox system via NAC have shown an increase in brain GSH49 and improvement of gFA in the fornix48. We also anticipate that relationships between the GSH redox cycle and WM differ in healthy controls (HC) and patients, but also between subgroups of participants classified as having low GPx/GR-ratio versus high GPx/GR-ratio.
Methods
Participants
Data were collected from 185 individuals (Table 1) divided into two groups: 86 healthy controls (HC), 99 patients (PT) with a diagnosis from the psychosis spectrum (Supplementary Table S1). Patients that suffered at least one psychotic episode (as defined by Comprehensive Assessment of At-Risk Mental States scale64, CAARMS) or had DSM-IV diagnosis of SZ and schizoaffective disorder were recruited from the Lausanne University Hospital. PT with psychosis related to intoxication or organic brain disease, IQ<70, reporting alcoholism, drug abuse, major somatic disease, documented anamnestic or current organic brain damage were excluded. HC were recruited from the same sociodemographic area as the patients. HC were excluded if they had a first-degree family member who suffered from psychosis or prodromal symptoms, or if they reported current or past antipsychotic treatments. Symptoms in PT were assessed via the Positive and Negative Syndrome Scale65 (PANSS), and global functioning was estimated for the whole sample via the Global Assessment of Function66 scale (GAF). The study was approved by the local Ethics Committee of the Canton of Vaud (Switzerland) under authorization numbers CER-VD 382/11 and 2018-01731.
P-values refer to Wilcox’s tests between patients and healthy control. χ2 test is computed for the scanner, sex, and reader contingency tables. GPx and GR activities are expressed in nmoles NADPH used/min/g hemoglobin. HC: healthy controls, PT: patients, GPx: glutathione peroxidase, GR: glutathione reductase, GAF: Global Assessment of Functioning, PANSS: Positive and Negative Syndrome Scale. †: the delay MRI-biomarker statistics are reported excluding one PT participant with an extreme delay of 271 days.
MRI acquisition
MRI scanning sessions were performed on two different 3-Tesla systems (Magnetom TrioTim and PRISMA, Siemens Healthineers, Erlangen, Germany), each equipped with a 32-channel head coil. A 1-mm isotropic T1-weighted image was acquired for anatomical reference. Whole-brain diffusion-weighted images (DWI) were acquired using diffusion spectrum imaging (DSI) scheme across 15 b-values, ranging from 0 to 8000 s/mm2, spatial resolution of 2.2 x 2.2 x 3 mm3 - See Supplementary Material for further acquisition details.
Image preprocessing
The MPRAGE image was bias field corrected67 and skull-stripped via nonlinear registration to the MNI-152 template68 using Advanced Normalization Tools69 (ANTs). The diffusion preprocessing pipeline included MP-PCA denoising, Gibbs ringing-, EPI-, eddy current and motion corrections, following most recent guidelines70 - see Supplementary Material for preprocessing details.
Microstructure estimation
For DKI and WMTI-W estimation, the diffusion dataset was truncated50 at b≤2500 s/mm2. DKI was fit voxel-wise in the entire brain71 using a weighted linear-least squares algorithm in Matlab71, from which seven scalar maps were derived. Four from DTI: RD, MD, AD and FA, and three from DKI: radial, mean, axial kurtosis (RK, MK, AK). However, only three parameter maps were included in the statistical analysis: MD, FA, and MK. WMTI-W parameters were estimated voxel-wise from the seven DKI scalars, using an in-house Python script, yielding five parameter maps: axonal density f, intra-axonal diffusivity Da, extra-axonal parallel and perpendicular diffusivities De,//, De,⊥ and axon orientation alignment c2. However, only axonal density f was included in the statistical analyses, to limit the number of comparisons.
Blood analyses
GR and GPx activity were assessed in hemolyzed blood cells incubated in phosphate buffer solution (100mM, pH 7.5) containing EDTA (0.6mM), and non-limiting levels of either oxidized glutathione (GSSG, 2.5mM) and NADPH (0.25mM) (for GR), or GSH (2.5mM), NADPH (0.25mM), GR (0.84U/ml; Fluka) and tert-butyl hydroperoxide (TBHP, 0.8mM, Fluka) (for GPx). We expressed GPx and GR activities in nmoles NADPH used/min/g hemoglobin. These enzymatic activities were measured with two different plate readers. Batch effects between plate readers were corrected via Combat for biological data72 (see additional details on the blood analyses and harmonization procedure in the supplementary material).
Statistical analysis
Blood markers analysis
Statistical analyses were run in R73, group comparisons were computed via Welch’s t-test and ANOVA. Robust linear regression74 was used for all the regression and contrasts analyses, always including a correction by (quadratic) age, sex. Post-hoc pairwise contrasts were multiple comparisons corrected via False-Discovery Rate (FDR).
Evidence about the influence of aging on the redox system has been reported in the literature75,76, suggesting the system is maintained until ∼45 years of age76. We evaluated this hypothesis first by considering age as a continuous variable and then by comparing subjects under and above 40 years old (N<40y.o.=165, N≥40y.o.=21).
Association between blood markers and WM microstructures
Before any statistical analysis, all the microstructure parameter estimates were harmonized for scanner type via ComBat harmonization77, which was proven efficient at correcting scanner effects in the same cohort78. Furthermore, participants with a delay between MRI scan and blood samples larger than 60 days were excluded whenever an association between blood biomarkers and WM microstructure metrics was tested (84% of the original cohort, NHC=80, NPT=76, delayHC=10.0±11.7, delayPT=19.0±16.0, p<0.0001). All analyses included a correction for (quadratic) age, sex, and delay. Associations were initially tested at a whole WM level and then at voxel level using FSL’s Tract-Based Spatial Statistics79 (TBSS). First, individual FA maps were used to build a study-specific FA template using ANTs69 and the estimated warps were used to spatially normalize each individual FA, MD, MK and f map. From the average FA, a WM skeleton was estimated and warped-back to native space where individual FA, MD, MK, and f averages were computed for the whole WM skeleton analysis. While, at voxel level, the skeleton voxels were harmonized in template space with ComBat77 before permutation testing (FSL randomise80, 5000 permutations). The resulting statistical maps were FDR and Threshold-Free Cluster Enhancement (TFCE) corrected.
Low and High GPx/GR ratio analysis
To study if there could be differences between participants showing different levels of GPx and GR activity, we repeated the analysis described above after re-classifying the participants into two groups according to their GPx/GR-ratio. Low-ratio and high-ratio were defined as having a GPx/GR-ratio respectively lower and higher than the median GPx/GR ratio of the HC group (Fig.1A, MedianHC=6.64).
Correlations between GPx and GR activity in the whole cohort, colored by the quartiles of the ratio between GPx and GR (A), clinical status (B), and sex (C). GPx and GR activities are expressed in nmoles NADPH used/min/g hemoglobin. GPx: GSH peroxidase; GR: GSH reductase; HC: healthy controls; PT: patients; M: males; F: females.
Results
Demographics
Summary demographics of the cohort can be found in Table 1. PT (29±9 y/o) and HC (28±8 y/o) did not differ in age (p=0.39). Differences were found between PT and HC in functional levels as measured by the GAF scale (p<0.0001) and in the delay between MRI acquisition and blood collection (p<0.0001), leading to the inclusion of this variable as covariate.
GSH-redox cycle system
Overall, GPx (p=0.60) and GR (p=0.19) activities or the GPx/GR-ratio (p=0.39) did not differ between PT and HC groups. When pooling together HC and PT, female participants showed higher GPx activity than males (p=0.00050). GPx and GR activities correlated significantly for the whole cohort (r=0.38, p<0.0001, Fig. 1A), but also for PT (r=0.35, p<0.0001) and HC (r=0.40, p=0.00013) when considered independently. The slope contrast between PT and HC did not show any significant difference (p=0.72, Fig. 1B). GPx and GR activities correlated also in both males and females (respectively, r=0.47, p=0.00024 and r=0.37, p=0.00037) with no significant sex difference for the slopes either (p=0.15, Fig. 1C). Age was only found significantly associated with a decrease in GPx activity (p=0.026). This decrease was particularly significant in participants over 40 years old. This group displayed lower GPx levels (p=0.00083) and ratio (p<0.0001, but not GR, p=0.89) when compared to under 40 years old. Finally, GPx activity and GPx/GR-ratio were found negatively associated with the duration of illness in PT (p<0.0001) but not with age at psychosis onset or medication.
Associations between GSH-redox cycle and WM microstructure
Within the whole cohort, irrespective of the clinical status or sex, neither GPx, GR, or their ratio correlated with average WM skeleton values of MD, FA, MK, or f. When considering the clinical status, GR regression models with MK (p=0.028, Fig. 2A) and f (p=0.043, Fig. 2B) showed significant differences between HC and PT slopes. The GPx/GR-ratio association with f also showed a difference between HC and PT slopes (p=0.027, Fig. 2C), while a trend could be observed for MK (p=0.054). No significant differences between HC and PT were found for GPx association with either MD, FA, MK or f.
Scatter plots A and B show the associations between GR activity and MK (A) and f (B) mean WM skeleton of HC (blue) vs patients (orange). Plot C shows the association of GPx/GR-ratio and f WM skeleton. GR activity is expressed in nmoles NADPH used/min/g hemoglobin. Brain plots D and E show the results for the TBSS analysis for the regressions of plot A (pfdr≤0.05). GR was found significantly associated with MK in HC (E) and differed from the PT slope (D). Clusters were identified in the bilateral superior corona radiata, genu and body of the corpus callosum. Results are corrected for age2, sex, and delay. GR: glutathione reductase; MK: mean kurtosis; f: axonal water fraction; βHC: slope of HC; βPT: slope of the PT.
TBSS analysis of the same regressions further substantiates the relationships of GR with MK but not the associations of f with GR or the GPx/GR-ratio. In HC, MK was significantly and negatively associated with GR in multiple WM clusters (Fig. 2D). Furthermore, the slope (GR–MK) in HC differed significantly from the one in PT (Fig. 2E). The significant contrast was localized bilaterally in the superior corona radiata (CR), genu and body of the corpus callosum (CC).
Associations between GPx, age, sex and WM microstructure
In light of the negative association between GPx levels and age, we also evaluated whether their interaction could be reflected in the WM microstructure. Irrespectively of the clinical status, the interaction model of GPx and quadratic age, corrected for sex and delay, resulted in a significant interaction for MD (p=0.0017, Fig. 3). Namely, with aging, MD was overall higher in individuals with low GPx activity than in individuals with elevated GPx activity, irrespective of the clinical status. We also considered the 3-way interaction of GPx, aging, and clinical status on the association with MD, but we did not find any significant interaction (p=0.18). Complementary TBSS analysis did not show any significant cluster for the same regressions.
The regression lines represent the robust fitting of the association between MD and (quadratic) age-GPx interaction, when GPx is fixed at its 2nd (blue curve) and 4th quartiles (red curve). PT and HC are plotted together. At lower GPx (blue curve), MD increases faster after 40 years of age than in subjects with high GPx, possibly indicating worse aging. At a younger age the difference is not detectable. GPx and GR activities are expressed in nmoles NADPH used/min/g hemoglobin.
Low GPx/GR-ratio versus high GPx/GR-ratio
We expected individuals with high GPx/GR-ratio to have a more oxidized redox state of the GSH/GSSG couple than those with a low GPx/GR-ratio. The highest GPx/GR-ratios may indicate systemic OxS, while the lowest GPx/GR-ratios may reflect the presence of reductive stress.
PT were distributed equally between the low-ratio and a high-ratio group (Table 1), indicating no overall differences between PT and HC. First, we established that the low- vs high-ratio is driven by both GPx and GR levels. Considering both PT and HC together, participants with low ratio (p<0.0001) had significantly lower GPx (μGPx=13.09) and higher GR (μGR=2.61) as compared to subjects with high ratio (μGPx=19.76, μGR=2.16). The same pattern was observed within the clinical groups and sex (p<0.0001). GPx and GR significantly correlated within both low- (r=0.65, p<0.0001) and high-ratio (r=0.74, p<0.0001) groups, but their slopes significantly differed (p=0.036) suggesting a tighter coordination of GPx and GR in the high-ratio group. Pairwise contrasts between the slopes of the interaction models of low- and high-ratio with the clinical status (6 tests) did not yield significant results.
Associations between GSH-redox cycle and dMRI WM microstructure in the low and high GPx/GR-ratio groups
We then assessed how low and high GPx/GR-ratio reflected at the WM microstructure level. Overall, the WM metrics did not significantly differ between individuals with low and high ratio after correcting for quadratic age, sex and delay. Two-way interaction models of low- and high-ratio with either sex or clinical status did not yield significant differences. However, the 3-way interaction model of MK (low- and high-ratio, sex and clinical status) showed a significant interaction in GPx (p=0.012) and a trend for GR (p=0.080). Post-hoc comparisons (4 tests: low- vs high-ratio, and PT vs HC) of the two MK models found significant differences in the slopes of GPx (pfdr=0.038, Fig. 4A) and GR (pfdr=0.016, Fig. 4B) in low-ratio HC vs low-ratio PT.
The scatterplots show the GPx (A) and GR (B) association with the MK mean-WM skeletons of individuals with low GPx/GR-ratio. GR activity is expressed in nmoles NADPH used/min/g hemoglobin. Plot (C) shows the TBSS analysis results of the same associations. In red/yellow are plotted the significant voxels (pfdr≤0.05) of the contrast between the slopes of HC and PT. Both analyses included a correction for quadratic age, sex and delay. GPx and GR activities are expressed in nmoles NADPH used/min/g hemoglobin. Green: WM skeleton; HC: healthy controls; PT: patients; GR: glutathione reductase; GPx: GSH peroxidase; MK: mean kurtosis; βHC: slope of HC; βPT: slope of the PT; Low: participants with GPx/GR-ratio < median ratio of the HC group.
TBSS voxel-wise analysis of the same 3-way interaction model showed widespread significant differences in the slopes of MK vs GR between low ratio HC and PT (Fig. 4C) but not for any other comparison, indicating the low-ratio group drives the difference between HC and PT. Identified areas for the GR-MK association were the bilateral cerebellar peduncle (CP), internal capsule, the splenium and body of the CC, the whole CR, the right superior longitudinal fasciculum (SLF), right tapetum, genu of the CC, thalamic radiation, fronto-occipital fasciculum (FOF), and the fornix. Notably, many voxels of the peripheral WM were also significant.
Discussion
In this work, we quantified peripheral biomarkers of the GSH redox cycle (GPx, GR activities and their ratio) in patients with psychosis and control subjects, and we explored the relationships of these redox markers with the WM microstructure estimated with advanced dMRI methods. We did not find significant differences in GPx, GR or GPx/GR-ratio between groups. However, female participants showed higher GPx activity than males, independently of their clinical status. GPx was also the only biomarker whose activity decreased with age. Regarding the WM metrics (MD, FA, MK, f), only MD changed significantly with aging, with the older subjects displaying in higher MD at lower GPx. Finally, GR activity was negatively associated with MK in HC, but not in PT, mostly driven by the subgroup of individuals with a low GPx/GR-ratio where the differences between HC and PT was considerably enhanced and more widespread across the WM.
Peripheral GSH redox cycle
Many studies have investigated blood GPx and GR activities in cohorts of psychotic patients9,22,23,81, sometimes with mixed results. However, meta-analyses9,23 show overall significantly lower GPx, but not GR, in PT as compared to HC. To our knowledge, the GPx/GR-ratio has not yet been examined. The analysis of Flatow et al.23 indicates that a decrease in GPx activity was prevalent in patients with chronic SZ or in PT with acute relapse but not with first psychosis episode. In our cohort, which includes PT with early psychosis or chronic SZ, we did not find any difference between PT and HC regarding GPx, GR and their ratio, even after correction for age and sex. However, we observed higher GPx activity in females than males and decreased GPx activity with aging (particularly above 40 years old), independently of the clinical status. Thus, some of the discrepancies among studies on PT with psychosis could be explained, not only by the stage of the disease, but also by overlooked effects on the peripheral markers of aging76,82,83, sex84, medication85 and circadian rhythm (see 8 supplementary), methodological differences in sample collection, processing, storage, and quantification. However, our findings of sex and age effects on blood GPx activity, regardless of the clinical status, corroborate results from numerous studies76,82–84,86. Higher levels of GPx, but not GR, have been also widely reported in the brain of females as compared to males in both animals and humans87–89. A higher GPx activity in females is consistent with more efficient antioxidant systems and less basal inflammation in females than in males87. Our observation of a diminished GPx activity with age is also in line with the literature75 and with the notion that the antioxidant defenses weaken over the years, contributing to the accumulation of OxS and its harmful effects83. In our work, we found that GPx and GR activities correlated positively within both PT and HC and within both males and females, suggesting a generally quite well-coordinated GSH redox cycle. Of note, the activity of both enzymes correlated in female, adolescents only from the general population84, but not in EP patients who had been exposed to childhood trauma90. This suggests that the relationship between GPx and GR activities may differ in a sex-dependent manner during adolescence but also among subgroups of individuals according to environmental and other yet unidentified factors. Elevated GPx/GR-ratio and/or GPx activity in blood cells may reflect a condition of systemic OxS, while low GPx/GR-ratio and/or high GR activity may indicate systemic reductive stress. Of note, PT who benefited from a NAC treatment were mostly those with high blood GPx activity48. On the other hand, reductive stress is the result of a shift of the redox balance in biological redox couples (i.e., GSSG/GSH, or NAD+/NADPH) to a more reducing state. This reductive stress can disturb redox-sensitive biological processes, but also lead paradoxically to increase mitochondrial reactive oxygen species (ROS) production, thus overwhelming the ROS neutralization capacity and causing mitochondrial dysfunction and oxidative damage91. Of note, PT with SZ younger than 40 years-old were reported to display signatures of reductive stress as compared to age-matched HC12.
Associations between peripheral GSH redox cycle and WM microstructure
Due to the diversity in the GSH redox cycle status across individuals and the complex effects of sex and age, it was therefore essential to tackle the relationships between the peripheral redox markers and WM microstructure by carefully considering age, sex and their interactions, and by investigating the links between the GSH redox cycle and WM separately in subgroups of individuals characterized by different peripheral redox profiles.
Thus, we first found that MD (the average diffusion coefficient in the tissue) was predicted by the interaction between GPx and aging. MD has been shown to increase in WM with normal aging92, due to the reduced numbers of myelinated axons, inefficient remyelination processes, increased number oligodendrocytes and excessive myelination around some axons93.
Our results suggest that GPx may modulate the effect of aging on the WM microstructure as reflected by MD. We found that individuals with low GPx activity tended to display higher MD than subjects with high GPx activity with aging, independently from the clinical status (Fig. 3). This is consistent with the antioxidant effect of GPx and with the idea that OxS may contribute to accelerated aging in SZ94. We believe the reason for the lack of differences between PT and HC in this association is the sparsity of subjects above 40 years old (N>40y.o.=21) which makes the modelling at older ages challenging.
The strongest differences between PT and HC were found in the association of GR with the WM microstructure (Fig. 2, 4), particularly with MK. MK is considered a measure of tissue complexity50,51, which has demonstrated sensitivity at detecting abnormal myelination57–60, neurodegeneration and neuroinflammation61. Our previous work on the same data focusing on JHU region-of-interest (ROI) analysis demonstrated that MK was significantly lower in EP and chronic SZ as compared to age-range matched controls63. Our current analyses on the average WM skeleton and voxel-wise on the skeleton revealed that lower MK was associated with higher blood GR activity in HC, but not in PT. This suggests that high GR activity was linked with changes in myelin and tissue heterogeneity in HC. As high GR activity may be sign of reductive stress, we looked at the relationship between GR and MK in individuals with low GPx/GR-ratio. This revealed that the difference in the GR-MK association between PT and HC were indeed mostly driven by individuals displaying a low GPx/GR-ratio. This suggests that reductive stress in blood is linked to a low MK in WM of HC, while this does not appear to be the sole modulator of MK in PT. Here, the lack of a direct association between MK and GR in PT could suggest there are multiple mechanisms affecting WM integrity in PT, and thus a less direct relationship than in HC.
Prior research has shown that NAC treatment can improve brain GSH levels after six months49 and WM integrity as measured by gFA in PT48. Notably, improvements in cognitive abilities and in symptoms49, particularly in individuals with high GPx levels, have also been reported, highlighting the interplay between blood markers, WM integrity and cognitive function. However, we found no associations between GPx or high-ratio GPx with WM microstructure in PT, contrary to our expectation, further substantiating the possibility that multiple mechanisms are influencing the WM integrity in PT.
As a side note, while we found group differences in WM microstructure metrics at the JHU ROI level and their ensemble in our prior research63, here we did not find group differences in the WM skeleton as defined by TBSS analysis. This suggests that considering larger or specific ROIs may be more important for detecting group differences than just the WM skeleton. However, the previously described MK-GR relationship remains unaffected by the mask used, showing consistent results for both the JHU ensemble and WM skeleton (not shown). Interestingly, the spatial localization of the strongest GR-MK significant contrasts between HC and PT (Fig. 2D) coincides with the FA WM ROIs (CR, genu and body of the CC) showing the strongest effect size between HC and PT found in the large meta-analysis of the ENIGMA study group46 and similarly reported in our previous study63.
Limitations
Our study has a few limitations that need to be considered. First, we used harmonization procedures to remove both the effects of MRI scanner and of plate readers used to quantify the redox markers. In addition, the delay between MRI acquisition and blood collection could be another source of bias8 despite being capped and controlled during the statistical analyses. The small sample size and heterogeneity among the PT cohort (that includes early psychosis PT and subjects with chronic SZ) may significantly reduce the statistical power due to potential conflicting trends that may arise at WM microstructure or GSH-system levels across the different stages and types of psychosis. Finally, GPx and GR activities and their ratio in blood cells inform us about the systemic redox homeostasis which does not necessarily reflect the conditions found in the brain. However, it is likely that systemic redox dysregulation impact brain structure and function. Indeed, several studies have found associations of peripheral markers of the GSH-redox cycle with brain GSH levels95, brain atrophy and enlargement of the lateral ventricles96, or EEG sensory evoked responses97.
Conclusions
Altogether, our extensive association analyses indicate that while the peripheral GSH-redox cycle may predict WM microstructural properties such as MK in HC and age-dependent increase of MD in WM, it does not alone explain the WM anomalies detected in PT. Our results do not however exclude the involvement of the GSH redox cycle during the development/maturation of WM. The dysregulation of this system during brain development could still contribute to WM anomalies in psychosis.
Acknowledgments
This work was supported by the Swiss National Science Foundation (PCEFP2_194260, to I.J.; 320030_197787 to PH and YA), the National Center of Competence in Research (NCCR) “SYNAPSY - The Synaptic Bases of Mental Diseases” from the Swiss National Science Foundation (n° 51NF40 – 185897 to KQD & PC) and the Foundation Alamaya. Dr. Alameda is supported by Carigest fellowship and by Frutiger Adrian et Simone fellowship. Dr. Dwir and P. Klauser are supported by Frutiger Adrian & Simone fellowship. The authors have nothing to disclose and there are no conflicts of interest. We are grateful to Gloria Reuteler, Adeline Cottier and Morgane Baumgartner for expert technical assistance. Dr Luis Alameda thanks the Foundation Adrian and Simone Frutiger and Carigest SA Foundation for their support
References
- 1.↵
- 2.↵
- 3.
- 4.
- 5.
- 6.
- 7.↵
- 8.↵
- 9.↵
- 10.
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.
- 16.
- 17.
- 18.
- 19.↵
- 20.↵
- 21.
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.
- 44.
- 45.
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
Subject Area
Reviews and Context
0
Comment
0
TRIP Peer Reviews
0
Community Reviews
0
Automated Services
0
Blogs/Media
Author Videos