Introduction
Schizophrenia is a severe psychiatric disorder that typically begins in late adolescence or early adulthood and often follows a chronic, relapsing course. It is associated with reduced life expectancy and an increased risk of suicide.1 The condition is characterized by positive (eg, delusions, hallucinations) and negative symptoms (eg, social withdrawal, emotional blunting) which—if untreated—can lead to significant social isolation.2
A key feature of schizophrenia is impaired brain state transitions, particularly involving the anterior cingulate cortex (ACC). Dysfunction within the ACC disrupts cognitive control, emotional regulation, and behavior—core components of the disorder’s symptomatology. The posterior cingulate cortex (PCC), a metabolically active brain region, may also exhibit neuronal dysfunction in schizophrenia, paralleling abnormalities observed in the ACC and reflecting broader disruptions in cingulate network dynamics. Given its role in modulating energy efficiency and network control, alterations in neurometabolite concentrations within the PCC may provide critical insights into the neurobiological substrates underlying affective and cognitive disturbances in schizophrenia. This view aligns with evidence that symptom severity in first-episode schizophrenia correlates with excessive spontaneous switching between large-scale brain networks, implicating impaired neurometabolic regulation in both the ACC and PCC.3,4
Neurometabolic alterations, including reduced N-acetylaspartate (NAA) levels, have been observed in individuals with schizophrenia, particularly in the ACC and thalamus, and are associated with cognitive impairment and neuronal dysfunction.5,6 Abnormalities in glutamate (Glu) and myo-inositol (Ins) levels further suggest disrupted brain metabolism and neurotransmitter imbalances in the pathophysiology of the disorder.7,8
Magnetic resonance spectroscopy (MRS) is a valuable method for assessing these changes, enabling the measurement of key metabolites such as Glu, NAA, and Ins.9-13 Altered Glu/NAA and Glu/Ins ratios, particularly in the ACC, have been linked to emotional symptoms, including depression and anxiety.14-16 Proton MRS (¹H-MRS) allows for noninvasive evaluation of cortical metabolic activity, supporting both disease monitoring and treatment assessment.17-19
Glu, NAA, and Ins represent distinct components of neuronal and glial metabolism. Glu is the major excitatory neurotransmitter and a key intermediate in the tricarboxylic acid (TCA) cycle, reflecting both neuronal activity and astrocytic function through Glu–glutamine cycling.20,21 NAA, synthesized in neuronal mitochondria, serves as a marker of neuronal integrity and mitochondrial metabolism.22,23 Ins, predominantly localized in astrocytes, is involved in phosphoinositide signaling and osmoregulation, reflecting glial density and function.24,25 Thus, changes in Glu, NAA, and Ins concentrations provide valuable insights into neuron–glia interactions and broader alterations in brain energy metabolism.
This study investigates neurometabolic alterations in the ACC and PCC in relation to emotional symptoms, such as depression and anxiety, in schizophrenia. These symptoms often persist despite antipsychotic treatment and significantly impact quality of life. Identifying specific metabolic patterns may provide insights into the biological basis of emotional dysregulation and support the development of targeted therapies.
We hypothesized that altered Glu/NAA and Glu/Ins ratios, measurable via ¹H-MRS, are associated with the severity of emotional symptoms in schizophrenia, independent of medication status. While antipsychotics, such as olanzapine, may affect brain chemistry, this study focused on neurometabolic changes reflecting the core pathology of the disorder. By examining patients treated with olanzapine, we also assessed potential medication effects on these biomarkers. The findings may improve understanding of the neurobiological basis of emotional and cognitive symptoms in schizophrenia, as well as inform more targeted diagnostic and therapeutic approaches.
Methods
Study design
It is a cross-sectional, observational study aimed at examining neurometabolic alterations in the cingulate cortex, with a specific focus on the ACC and PCC, in individuals with schizophrenia. We examined whether Glu/NAA and Glu/Ins ratios were associated with the severity of emotional symptoms, as assessed by the Beck Depression Inventory II (BDI-II) and State-Trait Anxiety Inventory (STAI).
The participants were divided into 2 groups: the schizophrenia group (study group) and the control group. The schizophrenia group consisted of individuals diagnosed with schizophrenia, while the control group included age- and sex-matched healthy individuals. All participants underwent 1H-MRS to measure the levels of key neurometabolites (Glu, NAA, and Ins) in the ACC and PCC.
The study focused on evaluating potential associations between variations in Glu/NAA and Glu/Ins ratios and emotional symptoms (such as depression and anxiety) commonly observed in schizophrenia. Additionally, it explored how these neurometabolic changes related to other factors, such as echo time (TE; 30 vs 144 ms) and the specific cortical region examined (ACC vs PCC).
Characteristics of the sample
The study enrolled 96 participants at a median (interquartile range [IQR]) age of 30.5 (22.8–36) years, reflecting a population spanning from late adolescence to early adulthood—a period consistent with the typical onset of schizophrenia. The sex distribution was slightly male-predominant (52 men [54.2%] vs 44 women [45.8%]), aligning with epidemiological trends of a higher schizophrenia prevalence in men.
The control group consisted of 45 healthy volunteers at a median (IQR) age of 32 (27–36) years and a near-even sex distribution (22 men [48.9%] vs 23 women [51.1%]). These participants had no history of psychiatric disorders, neurological conditions, or psychotropic medication use, as inferred from their role as normative comparators.
The schizophrenia group included 51 patients with a male predominance (30 men [58.82%] vs 21 women [41.18%]) and a median (IQR) age of 30 (18–35.5) years, consistent with the study’s focus on early-onset schizophrenia and age-related neurometabolic changes.
The slightly older median age in the control group suggests a stable, mature neurodevelopmental baseline, ideal for contrasting with the pathological changes in schizophrenia. The lack of age (P >0.9) or sex (P = 0.33) differences between the groups ensures comparability for neurometabolic analyses.
Magnetic resonance techniques
Magnetic resonance imaging (MRI) and single-voxel spectroscopy (SVS) were performed at the University Hospital in Kraków using a 3T Siemens MAGNETOM Vida Fit scanner (Munich, Germany) with a 20-channel head / neck coil. Standard noncontrast MRI was used to rule out structural brain abnormalities, and the results were assessed by a neuroradiologist.
SVS was performed using the point resolved spectroscopy sequence (TE, 30 and 144 ms; repetition time, 2000 ms; 40 averages). Spectra were acquired from the ACC and PCC, with consistent voxel placement (3.4 cm³) guided by anatomical landmarks and a brain atlas (Figure 1). Water suppression was achieved using chemical shift selective saturation, and standard shimming protocols were applied.

Figure 1. Magnetic resonance (MR) imaging; A – location of the volume of interest (VOI; rectangle) in the ACC in the sagittal plane; B – location of the VOI (rectangle) in the PCC in the sagittal plane; C – MR spectrum in the ACC at the TE of 30 ms; D – MR spectrum in the ACC at the TE of 144 ms; E – MR spectrum in the PCC at the TE of 30 ms; F – MR spectrum in the PCC at the TE of 144 ms.
Abbreviations: ACC, anterior cingulate cortex; Glu, glutamate; Ins, myo-inositol; GABA, γ-aminobutyric acid, NAA, N-acetylaspartate; PCC, posterior cingulate cortex; TE, echo time
Spectral quality was evaluated by a signal-to-noise ratio (SNR) and full-width at half maximum (FWHM). Data with FWHM above 20 Hz or SNR below 90% were excluded or repeated. Metabolites were quantified using LCModel version 6.3-1R (LCModel Inc., Oakville, Ontario, Canada), including only those with a Cramér–Rao lower bound below 20%.26 Both absolute concentrations and metabolite ratios (eg, NAA / creatine-phosphocreatine) were analyzed. All procedures followed international ¹H-MRS guidelines.27 Acquisition and analysis details are summarized in Table 1.

Reporting item | Comments |
|---|---|
Scanner manufacturer and model | Siemens MAGNETOM Vida Fit, syngo MR XA20 |
Magnetic field strength | 3 T, 123.260 703 MHz |
Voxel location and size | ACC and PCC; 3.4 cm³ |
Acquisition sequence (eg, PRESS, STEAM) | PRESS sequence used |
RT and TE | RT, 2000 ms; TE, 30 ms and 144 ms |
Number of averages (excitations) | 40 averages per spectrum |
Water suppression method | Standard method |
Shimming method | Optimized according to standard protocol |
Spectral quality (eg, linewidth, SNR) | FWHM >20 Hz and SNR <90% were exclusion criteria |
Basis set description | Standard fitting model based on reference metabolite resonance patterns |
Quantification software (eg, LCModel) | LCModel Version 6.3–1R |
Tissue correction applied | Not applied in this study |
Metabolite ratios reported | Glu/NAA and Glu/Ins ratios reported |
Abbreviations: FWHM, full-width at half maximum; PRESS, point resolved spectroscopy; RT, repetition time; SNR, signal-to-noise ratio; STEAM, stimulated echo acquisition mode; others, see Figure 1 | |
Statistical analysis
Robust multivariable regression models were employed to examine the Glu/NAA and Glu/Ins ratios, accounting for potential outliers and non-normality in the data. Multivariable modeling was utilized to simultaneously adjust for multiple covariates, thereby isolating the independent effects of key predictors while exploring potential interactions. TE was treated as a factor variable (levels: 30 ms as reference, 144 ms). The effects of numeric predictors were initially centered by the sample medians to make the intercept value clinically interpretable. The models were initially fitted for the combined sample of patients with schizophrenia and healthy controls (n = 335 for Glu/NAA; n = 334 for Glu/Ins), followed by separate models restricted to the schizophrenia group (n = 200 for both ratios). Multicollinearity was assessed via variance inflation factors. In this design, group and olanzapine therapy were between-subjects factors, while cortical region (cortex) and TE were within-subjects factors.
For the combined sample, the model for the Glu/NAA ratio included a 3-way interaction term (group × cortex × TE) as the primary predictor structure, with group (schizophrenia as reference), cortex (ACC as reference), and TE (30 ms as reference) specified as categorical factors. Additional covariates comprised sex (female as reference), age, BDI-II score, STAI score, Global Impression Scale score, white blood cell count (WBC), reactive lymphocyte (Re-Lymph) count, red blood cell count, complement C3c, and complement C4. Estimated marginal means (EMMs) were computed for the Glu/NAA ratio by group, cortex, and time, with pairwise between-group contrasts (schizophrenia vs control) at each level of cortex and TE, adjusted using the Šidák method.28
An analogous model was fitted for the Glu/Ins ratio, employing the same predictor structure and covariates. EMMs were derived similarly, and pairwise between-group contrasts were calculated with the Šidák adjustment. For each contrast, the Cohen d was estimated as the ratio of the contrast estimate to its standard error.
Subsequent models were fitted exclusively to the schizophrenia group. For the Glu/NAA ratio, the model incorporated a 3-way interaction (olanzapine × cortex × TE), with olanzapine therapy (no as reference), cortex (ACC as reference), and TE (30 ms as reference) as categorical factors, alongside covariates of sex (female as reference) and age. EMMs were obtained for the Glu/NAA ratio by olanzapine therapy, cortex, and echo time, with pairwise between-subjects contrasts (without olanzapine therapy vs with olanzapine therapy) at each level of cortex and TE, adjusted using the Šidák method.
The model for the Glu/Ins ratio in the schizophrenia group mirrored this structure, including the same interaction and covariates. EMMs and pairwise between-subjects contrasts were computed equivalently, with the Cohen d derived for each contrast as the estimate divided by its standard error.
Characteristics of the statistical tool
Analyses were conducted using the R Statistical language, version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria)29 on Windows 10 pro 64 bit (build 19045; One Microsoft Way, Redmond, Washington, United States), using the packages rio (version 1.0.1),30 emmeans (version 1.8.9),31 sjPlot (version 2.8.15),32 parameters (version 0.21.3)33 performance (version 0.10.8),34 report (version 0.5.7),35 MASS (version 7.3.60),36 readxl (version 1.4.3),37 and dplyr (version 1.1.3).38
Ethics
The study was conducted in accordance with the Declaration of Helsinki and approved by the Jagiellonian University Bioethics Committee (1072.6120.252.2021). Informed written consent was obtained from all participants.
Results
The study group involved 51 individuals with schizophrenia, aged 13–32 years. The control group consisted of 45 healthy volunteers, aged 13–30 years, with a men-to-women ratio of 1.43. Of the 51 participants, 25 (49%) were not taking olanzapine, while 26 (51%) were receiving olanzapine as part of their treatment for schizophrenia. Among the drugs studied, olanzapine was the only one with a sufficiently large sample size to enable statistical analysis of its effects on brain metabolic activity and clinical outcomes. The significance and magnitude of the effects associated with olanzapine use were determined using the ordinary least squares estimation. The results of this analysis were presented in our previous publications,39,40 where descriptive statistics for the study variables in this cohort were also provided.
Glutamate / N-acetylaspartate ratio analysis vs age- and emotion-related neurochemical changes
The results of the robust linear model (RLM) for the Glu/NAA ratio as outcome, based on 335 observations (representing the full dataset, including repeated measures across 4 combinations of cortical regions and time variables: ACC and PCC at 30 ms and 144 ms), are detailed in Table 2. The mean Glu/NAA ratio in the group of women with schizophrenia for ACC at 30 ms, considering the medians of clinical parameters and psychological test scores, was B0 = 1.34 (P <0.001).

Predictors | Glu/NAA | Glu/Ins | ||||
|---|---|---|---|---|---|---|
B | 95% CI | P value | Β | 95% CI | P value | |
Intercept | 1.34 | 1.29–1.39 | <0.001 | 1.71 | 1.61–1.8 | <0.001 |
Group (schizophrenia) | Reference category | Reference category | ||||
Group (control) | –0.01 | –0.07 to 0.06 | 0.86 | 0.1 | –0.02 to 0.22 | 0.11 |
Cortex (ACC) | Reference category | Reference category | ||||
Cortex (PCC) | –0.22 | –0.27 to –0.16 | <0.001 | 0.06 | –0.04 to 0.17 | 0.23 |
Echo time, 30 ms | Reference category | Reference category | ||||
Echo time, 144 ms |
| –0.78 to –0.67 | <0.001 | –0.8 | –0.9 to –0.69 | <0.001 |
Sex (women) | Reference category | Reference category | ||||
Sex (men) | 0.03 | –0.01to 0.07 | 0.18 | 0.07 | –0.01 to 0.14 | 0.08 |
Age, y (median, 31) | –3.36 × 10–3 | –5.3 × 10–3 to –1.4 × 10–3 | 0.001 | –0.01 | –0.01 to –0.01 | <0.001 |
GAF score, points (median, 71) | Not retained in the final model | Not retained in the final model | ||||
BDI II score, points (median, 11) | 3.52 × 10–3 | 1.3 × 10–3–5.7 × 10–3 | 0.002 | 4.7 × 10–3 | 0.5 × 10–3–8.9 × 10–3 | 0.03 |
STAI score, points (median, 87) | –1.62 × 10–3 | –2.9 × 10–3 to –0.3 × 10–3 | 0.02 | –2.94 × 10–3 | –0.01 to –0.5 × 10–3 | 0.02 |
GiS score, points (median, 22) | 9.99 × 10–5 | –1 × 10–3 to 1.2 × 10–3 | 0.85 | –3.92 × 10–4 | –2.4 × 10–3 to 1.6 × 10–3 | 0.71 |
WBC, × 103/µl (median, 6.54) | 4.48 × 10–3 | –4.5 × 10–3 to 0.01 | 0.33 | 0.01 | –0.01 to 0.02 | 0.37 |
Re-Lymph, × 103/µl (median, 0.05) | –0.4 | –0.78 to –0.01 | 0.04 | 0.02 | –0.72 to 0.75 | 0.97 |
RBC, × 106/µl (median, 4.82) | –0.02 | –0.06 to 0.03 | 0.45 | –0.08 | –0.16 to –3 × 10–3 | 0.06 |
Complement C3c, g/l (median, 1.15) | 0.07 | –0.04 to 0.17 | 0.21 | 0.08 | –0.12 to 0.28 | 0.41 |
Complement C4, g/l (median, 0.21) | 0.11 | –0.26 to 0.49 | 0.56 | 0.13 | –0.58 to 0.85 | 0.71 |
a The effects of numeric predictors were initially centered by the sample medians to make the intercept value clinically interpretable. Abbreviations: BDI-II, Beck Depression Inventory II; GAF, Global Assessment of Functioning; GiS, global impression of severity; STAI, State-Trait Anxiety Inventory; RBC, red blood cells; Re-Lymph, reactive lymphocytes; WBC, white blood cell; others, see Figure 1 | ||||||
However, the significant negative coefficients for TE at 144 ms in the PCC in relation to the ACC (B = –0.72; P <0.001) contained Table 2 indicate a substantial decrease in the Glu/NAA ratio, which may reflect changes in neurotransmitter balance related to neuronal activity. Moreover, the Glu/NAA ratio in the PCC with the shorter TE(30 ms) was by 0.22 units lower than in the ACC in the patients with schizophrenia. Additionally, age was associated with a slight decrease in the Glu/NAA ratio (P = 0.001; Table 2) in the group of patients with schizophrenia, suggesting age-related changes in neuronal health or metabolic activity, highlighting the need for age-adjusted clinical interpretations.
Psychologic states also affected the outcome, as illustrated in Table 2. Higher BDI II scores (more severe depressive symptoms) were linked to an increase in the Glu/NAA ratio (B = 3.52 × 10–3; P = 0.002), while higher STAI scores (greater anxiety) were associated with its decrease (B = –0.002; P = 0.02). This provided differing neurometabolic responses in depression and anxiety in the group of patients with schizophrenia.
Furthermore, Re-Lymph was negatively associated with the Glu/NAA ratio (B = –0.4; P = 0.04), indicating potential interactions between immune status and brain metabolism, which could be relevant for understanding the neuroimmunological aspects of schizophrenia (Table 2).
Interaction effects between group, cortex localization, and TEs, as presented in Tables 3 and 4, show a consistent Glu/NAA ratio across the groups in the ACC at the short TE (30 ms), indicating similar neurochemical environments. At the long TE (144 ms), a reduction in the Glu/NAA ratio was observed in both groups; however, there was no difference between them (P = 0.13). In the PCC, a comparable pattern was noted, with no difference between the groups at the long TE (P = 0.64). This demonstrated that while glutamatergic activity changed for 2 TEs (30 ms and 144 ms), the differences between the schizophrenia and control groups were minimal and may have not been detectable solely through the Glu/NAA ratio (Table 5).

Cortex | Echo time | Group | EMM | SE |
|---|---|---|---|---|
Glu/NAA | ||||
ACC | 30 ms | Schizophrenia | 1.37 | 0.02 |
30 ms | Control | 1.36 | 0.02 | |
144 ms | Schizophrenia | 0.65 | 0.02 | |
144 ms | Control | 0.7 | 0.02 | |
PCC | 30 ms | Schizophrenia | 1.15 | 0.02 |
30 ms | Control | 1.13 | 0.02 | |
144 ms | Schizophrenia | 0.47 | 0.02 | |
144 ms | Control | 0.49 | 0.02 | |
Glu/Ins | ||||
ACC | 30 ms | Schizophrenia | 1.77 | 0.04 |
30 ms | Control | 1.87 | 0.04 | |
144 ms | Schizophrenia | 0.97 | 0.04 | |
144 ms | Control | 1.14 | 0.04 | |
PCC | 30 ms | Schizophrenia | 1.83 | 0.04 |
30 ms | Control | 1.85 | 0.04 | |
144 ms | Schizophrenia | 0.72 | 0.04 | |
144 ms | Control | 0.77 | 0.04 | |
Abbreviations: EMM, estimated marginal mean; others, see Figure 1 | ||||

Cortex | Echo time | Contrast (control) | Estimate | SE | P value | d |
|---|---|---|---|---|---|---|
Glu/NAA | ||||||
ACC | 30 ms | Schizophrenia | 0.01 | 0.03 | 0.86 | –0.17 |
144 ms | Schizophrenia | –0.05 | 0.03 | 0.03 | 1.5 | |
PCC | 30 ms | Schizophrenia | 0.02 | 0.03 | 0.52 | –0.64 |
144 ms | Schizophrenia | –0.02 | 0.02 | 0.64 | –0.64 | |
Glu/Ins | ||||||
ACC | 30 ms | Schizophrenia | –0.1 | 0.06 | 0.11 | 1.62 |
144 ms | Schizophrenia | –0.16 | 0.06 | 0.01 | 2.66 | |
PCC | 30 ms | Schizophrenia | –0.01 | 0.06 | 0.86 | 0.18 |
144 ms | Schizophrenia | –0.05 | 0.06 | 0.4 | 0.84 | |
Abbreviations: see Figure 1 | ||||||

Cortex | Echo time | Olanzapine therapy | EMM | SE |
|---|---|---|---|---|
ACC | 30 ms | No | 1.35 | 0.03 |
30 ms | Yes | 1.38 | 0.02 | |
144 ms | No | 0.63 | 0.03 | |
144 ms | Yes | 0.66 | 0.02 | |
PCC | 30 ms | No | 1.16 | 0.03 |
30 ms | Yes | 1.14 | 0.02 | |
144 ms | No | 0.42 | 0.03 | |
144 ms | Yes | 0.52 | 0.02 | |
Abbreviations: see Figure 1 and Table 3 | ||||
Glutamate / myo-inositol ratio analysis vs age- and anxiety-related neurochemical shifts
The RLM results (Table 2) highlight key factors influencing the Glu/Ins ratio across 334 observations. The mean Glu/Ins ratio in women with schizophrenia in the ACC at 30 ms, adjusted for median clinical and psychological scores, was B0 = 1.71 (P <0.001).
The group analysis indicates that the control group, on average, exhibited a Glu/Ins ratio that was higher by a margin of 0.1, as compared with the schizophrenia group, although this difference was insignificant (P = 0.11). This shows some tendency toward an altered glutamatergic / inositol metabolism in schizophrenia, but further investigation is necessary to substantiate this finding comprehensively.
Regarding cortical areas, the slight increase (by 0.06) in the Glu/Ins ratio in the PCC, as compared with the ACC, was also insignificant (P = 0.23), indicating a potential homogeneity in Glu/Ins ratios across these cortical regions in the group of patients with schizophrenia under the conditions studied.
A noteworthy finding is the decrease in the Glu/Ins ratio at the long TE of 144 ms vs TE of 30 ms (B = –0.8; P <0.001; Table 2). This stark decrease suggests dynamic changes in the metabolic balance or neurotransmitter flux for different TEs in the group of patients with schizophrenia. It also underscores the importance of selecting appropriate MRS parameters to detect subtle neurochemical changes.
Age was negatively associated with the Glu/Ins ratio (B = –0.01/y; P <0.001), while depressive symptoms (BDI-II) showed a weak positive correlation (B = 0.005; P = 0.03). Anxiety (as indicated by the STAI scores) was negatively correlated (B = –0.00 294; P = 0.02), suggesting involvement of Glu and inositol signaling in affective symptoms. Other biological markers, including WBC, lymphocytes, and complements, showed no effects (Table 2).
The EMMs presented in Table 3 demonstrate that both cortex location and TE significantly influenced the Glu/Ins ratio, with marked decreases observed at the long TE of 144 ms across all categories, as compared with the short TE of 30 ms.
Additionally, the differences between the schizophrenia patients and controls showed subtle but significant variations at specific TEs. For instance, at the long TE of 144 ms in the ACC, there was a notable decrease in the Glu/Ins ratio for the schizophrenia patients, as compared with the controls (B = –0.16; P = 0.008; Table 4), indicating a distinct neurochemical profile associated with schizophrenia in this particular cortical area and time frame. This finding aligns with the hypothesis that schizophrenia may involve disrupted glutamatergic and Ins-related processes, which could be associated with the pathophysiology of the disorder.
However, similar comparisons in the PCC at the long TE of 144 ms and the ACC at the short TE of 30 ms show insignificant differences between the groups, suggesting that the effects might be more localized or specific to certain neural circuits within the brain.
The lack of difference observed in the PCC at the short TE of 30 ms between the schizophrenia patients and controls (B = –0.01; P = 0.86; Table 4) indicates that, in some contexts, the Glu/Ins ratio remains stable irrespective of schizophrenia status, highlighting potential regional variations in pathology or compensatory mechanisms at play.
The analysis demonstrated distinct profiles of similarities and differences in neurochemical interactions between the schizophrenic patients and healthy individuals. The schizophrenic patients exhibited disrupted correlations at shorter TEs, particularly in the PCC, and stronger correlations at prolonged TEs, potentially reflecting disease-specific adaptation.
Discussion
Results of the assessment of adjusted group differences in glutamate / N-acetylaspartate ratios
The regression model results for the Glu/NAA ratio indicate it was influenced by age of the participants, with each additional year of age associated with a slight decrease in the Glu/NAA ratio. This may reflect age-related changes in neuronal health or metabolic activity.41 Although subtle, these changes are significant and highlight the importance of age-adjusted interpretation in clinical settings to account for normal cognitive, social, and behavioral changes, as well as the high prevalence of schizophrenia during brain maturation.
Higher severity of depressive symptoms (as indicated by the BDI-II scores) is associated with an increase in the Glu/NAA ratio, which may reflect a combination of increased glutamatergic activity and a decreased number of healthy neurons with reduced metabolic function. Our findings are consistent with the most recent meta-analysis by Saccaro et al,42 which demonstrated markedly lower NAA levels in individuals with depressive disorders than in healthy controls. Decreased NAA levels may be associated with cerebral hypometabolism, which plays a key role in the pathophysiology of depression and may underlie its symptoms.43
Conversely, higher STAI scores, indicating greater levels of anxiety, were associated with a decrease in the Glu/NAA ratio. This may reflect adaptive homeostatic mechanisms aimed at reducing Glu levels, which could lead to decreased neuronal excitability. Our results align with those obtained by Bonnekoh et al,44 who found a relationship between NAA metabolism in the ACC and the activity of the hypothalamic-pituitary-adrenal axis, as represented by long-term cortisol levels.
The results concerning the level of Re-Lymph indicate its significant negative relationship with the Glu/NAA ratio. This confirms the associations and interactions between the immune system and brain metabolism observed in our previous studies.40
The analysis of interactions between the groups, cortical location, and TE is presented in Tables 3 and 4. For the short TE of 30 ms in the ACC, the Glu/NAA ratio was similar between the schizophrenia and control groups. At the 144 ms TE, both groups showed a decrease in the Glu/NAA ratio, with a more pronounced decrease in the schizophrenia group, though this difference is insignificant. In the PCC, the differences in the short TE are also in significant, but at the longer TE of 144 ms, the decrease in Glu/NAA was slightly greater in the schizophrenia group.
Assessment of adjusted differences between the groups in glutamate / myo-inositol ratios
Group analysis indicated that the Glu/Ins ratio was, on average, by 0.1 higher in the control group than the schizophrenia group, although this difference was insignificant (P = 0.11). The elevated Glu/Ins ratio in the control group might suggest defective neuron–glial coupling in schizophrenia. These findings are consistent with results of the Stanley Neuropathology Consortium, which demonstrated a reduction in glial cell numbers in the frontal and cingulate cortices of individuals with schizophrenia,45,46 as compared with nonaffected controls. Similarly, the difference in the present study did not reach significance. Our results are consistent with those of Jeon et al,46 who observed a significant reduction in Ins concentration in the ACC of patients with early, untreated acute schizophrenia, which became nonsignificant with treatment.
The Gln/Ins ratio in the brain is crucial for maintaining proper neuronal function. Changes in this ratio might reflect alterations in neurotransmission and energy metabolism, potentially linked to the underlying mechanisms of schizophrenia. For instance, an elevated Gln/Ins ratio might indicate an increased glutaminergic activity which could be associated with certain psychotic symptoms or an overactive neurotransmitter cycle.47 A significant decrease in the Glu/Ins ratio at the TE of 144 ms vs TE of 30 ms aligns with previous findings of TE-dependent Ins variability. Conversely, a decreased ratio might suggest deficiencies in neurotransmitter synthesis or regeneration, possibly correlating with the negative or cognitive symptoms seen in schizophrenia.48
A slight positive correlation was found between the Glu/Ins ratios and BDI-II scores, linking depressive symptoms to glial or glutamatergic alterations, while the STAI scores showed a negative correlation, suggesting a link between lower Ins levels with higher anxiety.48
Other biological markers, including WBC, lymphocyte count, and complement components, did not show associations with the Glu/Ins ratio. This suggests that these systemic inflammatory or immunological factors may not have a direct impact on this specific neurochemical balance under the study conditions.
Both cortical region and TE influenced the Glu/Ins levels, with consistent reductions at the TE of 144 ms. Importantly, at the TE of 144 ms in the ACC, the schizophrenia patients showed a markedly lower Glu/Ins ratio than the controls (−0.16; P = 0.01; Table 4), indicating a region- and TE-specific neurometabolic alteration related to glutamatergic and Ins dysregulation.
We acknowledge that the minimal group differences and modest effect sizes reported in some comparisons, such as this one, might initially appear to limit their immediate clinical utility. However, we contend that these findings, even with small effect sizes, hold meaningful implications for understanding schizophrenia pathophysiology, particularly when contextualized within its complex and heterogeneous nature. Below, we elaborate on this justification, incorporating the estimated effect sizes and aligning with the study’s broader findings.
Firstly, the modest effect size of d = –0.64 for the Glu/NAA ratio in the PCC at 144 ms, though insignificant, falls within the moderate range according to Cohen conventions (0.5–0.8), suggesting a potentially meaningful difference that may be obscured by sample size constraints (n = 335) or interindividual variability inherent in schizophrenia. Schizophrenia is characterized by multifactorial neurochemical alterations, where subtle deviations in neurotransmitter ratios, such as the Glu/NAA ratio, may reflect early or compensatory changes in excitatory neurotransmission and neuronal integrity—key components of its pathophysiology.49 The slight reduction of the Glu/NAA ratio in the schizophrenia patients, as compared with the controls (0.47 vs 0.49) could indicate an insidious decline in neuronal health or glutamatergic function in the PCC, a region implicated in self-referential processing and emotional regulation. While this difference was insignificant (P = 0.64), its moderate effect size suggests it may contribute to the cumulative burden of disease, aligning with evidence that schizophrenia involves a continuum of dysregulation rather than stark anomalies.
Secondly, the clinical relevance of small effect sizes is supported by their consistency with broader trends in the study and the neuroimaging literature. For instance, the Glu/NAA ratio in the ACC at 144 ms showed a larger effect size than in the PCC (d = 1.5; P = 0.13; Table 4), indicating regional variability in schizophrenia-related changes. Similarly, the Glu/Ins ratio in the ACC at 144 ms exhibited a large, significant effect (d = 2.66; P = 0.01), underscoring that subtle findings in one region (eg, PCC) coexist with more pronounced differences elsewhere. This heterogeneity suggests that small effect sizes in the PCC may represent a part of a distributed neurometabolic profile, where incremental alterations across multiple regions collectively underpin the clinical phenotype. The lack of significance in the PCC finding may also reflect measurement sensitivity or the cross-sectional design’s inability to capture progressive changes, rather than a lack of biological importance.
Moreover, small effect sizes in neuroimaging studies are common due to the complexity of brain metabolism and the influence of confounding factors, such as age and psychological states, which were adjusted for in our robust regression models. The observed effect size of d = –0.64, while modest, is comparable to those reported in meta-analyses of MRS studies in schizophrenia (eg, Merritt et al50), where subtle neurometabolite differences often correlated with symptom severity or cognitive deficits over time. In our study, the association of the Glu/NAA ratio with depressive (B = 3.52 × 10–3; P = 0.002) and anxiety symptoms (B = –1.62 × 10–3; P = 0.02) further suggests that even small group differences may reflect clinically relevant processes, such as altered excitatory-inhibitory balance, which could inform therapeutic targeting.
Finally, the potential clinical relevance is enhanced when considering within-group treatment effects. Olanzapine therapy in the PCC at 144 ms yielded a large effect size (d = –2.83; P = 0.01; Supplementary material, Table S1), significantly increasing the Glu/NAA ratio (EMM = 0.52 vs 0.42 without treatment; Table 5), indicating that subtle baseline differences may amplify under intervention. This suggests that small effect sizes in group comparisons could serve as precursors or markers of vulnerability, detectable only in specific contexts (eg, treatment response or longitudinal progression). Thus, while not immediately diagnostic, these findings contribute to understanding schizophrenia pathophysiology by highlighting nuanced, region-specific alterations that may guide future research into early detection or personalized interventions.
Supplementary material provides comprehensive details on the methodology, participant characteristics, inclusion and exclusion criteria, clinical and laboratory data, and multivariable analyses related to the effects of olanzapine therapy on neurometabolite ratios.
Limitations
This observational study has limitations, including the inability to establish causality and a limited sample size, especially for treatment-related analyses. Disease duration was not controlled due to group heterogeneity, and metabolite levels were not adjusted for partial volume effects or tissue composition, potentially affecting accuracy. Other antipsychotic classes were excluded due to small subsamples, and clinical factors, such as symptom triggers or episode frequency, were not examined. Anatomical variability may have influenced voxel placement.
Despite these limitations, the study’s strength lies in its stratification by olanzapine treatment and detailed analysis of Glu/NAA and Glu/Ins ratios in the ACC and PCC at 2 TEs (as presented in Supplementary material, Tables S2 and S3), offering valuable insights for translational schizophrenia research.
Conclusions
This study highlights the significant neurometabolic imbalances in the ACC and PCC associated with schizophrenia, particularly in the ratios of Glu/NAA and Glu/Ins. The effect of TE on neurometabolite ratios emphasizes the importance of optimizing MRS parameters for accurate assessment.
Age-related decreases in the Glu/NAA and Glu/Ins ratios suggest that schizophrenia may involve progressive changes in neuronal and glial metabolism over time. Additionally, the study found that higher depression severity (as indicated by the BDI II scores) correlates with increased Glu/Ins ratios, while higher anxiety severity (as reflected in the STAI scores) is associated with decreased Glu/Ins ratios, pointing to the interplay between emotional states and neurometabolic processes.
Our findings suggest that olanzapine does not have a significant effect on the Glu/Ins ratio, as it did not show an impact on the Glu/NAA ratio, regardless of the cortical region or TE. This indicates that the neurochemical effects of olanzapine on Glu, Ins, and NAA levels may be minimal in patients with schizophrenia. Alternatively, these effects could vary among individuals or be influenced by other factors not accounted for in this analysis, such as the duration of therapy or dosage. Overall, our results underscore a lack of a clear neurochemical impact of olanzapine treatment on these key brain metabolites in the studied population.
In summary, our research is crucial for advancing the diagnosis, treatment, and understanding of schizophrenia, potentially leading to an improved quality of life for patients with this challenging condition. This study establishes a precedent for future research by pioneering the integration of multifaceted factors into predictive models for schizophrenia. These findings provide valuable insights into the neurochemical underpinnings of the disorder and may inform the development of personalized therapeutic strategies for managing this complex condition.
Wirginia Krzyściak, PhD, Department of Medical Diagnostic, Faculty of Pharmacy, Jagiellonian University Medical College, ul. Medyczna 9, 30-688 Kraków, Poland, phone: +48 12 620 59 85, email: wirginia.krzysciak@uj.edu.pl
July 11, 2025.
September 2, 2025.
September 2, 2025.
We would like to thank Anna Skalniak, PhD, for her insightful comments on the manuscript. We also thank Translmed Publishing Group (Bedford, New Hampshire, United States) for assistance with proofreading and copy-editing this manuscript.
This work was supported by the Priority Research Area BioS under the program “Excellence Initiative – Research University” at the Jagiellonian University in Kraków. The project was also funded by the National Science Centre, Poland, under grant number 2024/08/X/NZ7/00479, awarded to WK as the principal investigator.
WK conceptualized the study, developed the methodology, ensured quality control of brain imaging data, conducted the experiments, constructed the statistical model, performed data analysis, and created all visual and video materials. She also managed the overall project, secured funding, and served as the primary supervisor of the study. Additionally, WK was responsible for revising the manuscript in response to reviewers’ comments, improving the text, supplementing and interpreting statistical data, and approving the final version of the manuscript. WK and MS jointly interpreted the data. WK, MS, and PM drafted the manuscript. PK performed magnetic resonance spectroscopy. AB performed the MRI radiological examination, prepared the radiological description for the patients, and contributed to the conceptualization of the MRS protocol. RCH verified the accuracy of the performed MRI and MRS examinations and conducted quality control of the imaging data. WK, AT, MP, NŚ, KF, and MS were responsible for patient recruitment and the collection of clinical data and biological samples. WK critically revised the manuscript for important intellectual content. WK supervised the research process. TK critically reviewed the initial version of the manuscript, contributed to its writing, and participated in the assessment of the clinical relevance of the study findings. TP was responsible for reviewing and editing the initial version of the manuscript prior to submission for peer review, and also led the team of radiologists and the medical physicist involved in the study. WK made the most substantial contribution to all stages of the study and manuscript preparation. In recognition of her leading role, she is designated as the first and corresponding author. All authors have read and approved the final version of the manuscript.
Artificial intelligence was not used in the preparation of this manuscript.
none declared.
Krzyściak W, Szwajca M, Karcz P, et al. Neurometabolic alterations and emotional states in schizophrenia: an MRS study of the cingulate cortex. Prz Lek Jagiellonian Med Rev. 2025; 77: 20004. doi:10.20452/jmr.2025.20004
- 1.
- Hjorthøj C, Stürup AE, McGrath JJ, Nordentoft M. Years of potential life lost and life expectancy in schizophrenia: a systematic review and meta-analysis. Lancet Psychiatry. 2017; 4: 295-301.Crossref
- 2.
- Kaul I, Sawchak S, Correll CU, et al. Efficacy and safety of the muscarinic receptor agonist KarXT (xanomeline-trospium) in schizophrenia (EMERGENT-2) in the USA: results from a randomised, double-blind, placebo-controlled, flexible-dose phase 3 trial. Lancet. 2024; 403: 160-170.Crossref
- 3.
- Li Q, Yao L, You W, et al. Controllability of functional brain networks and its clinical significance in first-episode schizophrenia. Schizophr Bull. 2023; 49: 659-668.Crossref
- 4.
- Smucny J, Wylie KP, Lesh TA, et al. Whole-brain intrinsic functional connectivity predicts symptoms and functioning in early psychosis. J Psychiatr Res. 2024; 175: 411-417.Crossref
- 5.
- Mihaljevic M, Chang YH, Witmer AM, et al. Reduction of N-acetyl aspartate (NAA) in association with relapse in early-stage psychosis: a 7-Tesla MRS study. Schizophrenia. 2024; 10: 29.Crossref
- 6.
- Bachmann D, Buchmann A, Studer S, et al. Age-, sex-, and pathology-related variability in brain structure and cognition. Translational Psychiatry. 2023; 13: 1-9.Crossref
- 7.
- Maddock RJ, Buonocore MH. MR spectroscopic studies of the brain in psychiatric disorders. Curr Top Behav Neurosci. 2012; 11: 199-251.Crossref
- 8.
- Kraguljac NV, Reid M, White D, et al. Neurometabolites in schizophrenia and bipolar disorder - a systematic review and meta-analysis. Psychiatry Res. 2012; 203: 111-125.Crossref
- 9.
- Luo L, You W, DelBello MP, et al. Recent advances in psychoradiology. Phys Med Biol. 2022; 67: 23TR01.Crossref
- 10.
- Cui L, Li S, Wang S, et al. Major depressive disorder: hypothesis, mechanism, prevention and treatment. Signal Transduct and Target Ther. 2024; 9: 30.Crossref
- 11.
- Moffett JR, Arun P, Ariyannur PS, Namboodiri AMA. N-Acetylaspartate reductions in brain injury: impact on post-injury neuroenergetics, lipid synthesis, and protein acetylation. Front Neuroenergetics. 2013; 5: 11.Crossref
- 12.
- Roberts D, Rösler L, Wijnen JP, Thakkar KN. Associations between N-Acetylaspartate and white matter integrity in individuals with schizophrenia and unaffected relatives. Psychiatry Res Neuroimaging. 2023; 330: 111612.Crossref
- 13.
- Friedman JI, Tang C, Carpenter D, et al. Diffusion tensor imaging findings in first-episode and chronic schizophrenia patients. Am J Psychiatry. 2008; 165: 1024-1032.Crossref
- 14.
- Fisher SK, Novak JE, Agranoff BW. Inositol and higher inositol phosphates in neural tissues: homeostasis, metabolism and functional significance. J Neurochem. 2002; 82: 736-754.Crossref
- 15.
- Das TK, Dey A, Sabesan P, et al. Putative astroglial dysfunction in schizophrenia: a meta-analysis of 1H-MRS studies of medial prefrontal myo-inositol. Front Psychiatry. 2018; 9: 438.Crossref
- 16.
- Stanley JA, Daugherty AM, Richter Gorey C, et al. Basal glutamate in the hippocampus and the dorsolateral prefrontal cortex in schizophrenia: relationships to cognitive proficiency investigated with structural equation modelling. World J Biol Psychiatry. 2023; 24: 730-740.Crossref
- 17.
- Kubota M, Moriguchi S, Takahata K, et al. Treatment effects on neurometabolite levels in schizophrenia: a systematic review and meta-analysis of proton magnetic resonance spectroscopy studies. Schizophr Res. 2020; 222: 122-132.Crossref
- 18.
- Seraji M, Ellis CA, Sendi MSE, et al. Uncovering effects of schizophrenia upon a maximally significant, minimally complex subset of default mode network connectivity features. bioRxiv. 2024 Apr 25. [Epub ahead of print].Crossref
- 19.
- Yang H, Zhang H, Di X, et al. Reproducible coactivation patterns of functional brain networks reveal the aberrant dynamic state transition in schizophrenia. Neuroimage. 2021; 237: 118193.Crossref
- 20.
- Schousboe A, Hertz L. Role of astroglial cells in glutamate homeostasis. Adv Biochem Psychopharmacol. 1981; 27: 103-113.Crossref
- 21.
- Hertz L. Astrocytic energy metabolism and glutamate formation—relevance for 13C-NMR spectroscopy and importance of cytosolic/mitochondrial trafficking. Magn Reson Imaging. 2011; 29: 1319-1329.Crossref
- 22.
- Moffett JR, Ross B, Arun P, et al. N-Acetylaspartate in the CNS: from neurodiagnostics to neurobiology. Prog Neurobiol. 2007; 81: 89-131.Crossref
- 23.
- Baslow MH. N-acetylaspartate in the vertebrate brain: metabolism and function. Neurochem Res. 2003; 28: 941-953.Crossref
- 24.
- Brand A, Richter-Landsberg C, Leibfritz D. Multinuclear NMR studies on the energy metabolism of glial and neuronal cells. Dev Neurosci. 1993; 15: 289-298.Crossref
- 25.
- Choi JK, Dedeoglu A, Jenkins BG. Application of MRS to mouse models of neurodegenerative illness. NMR Biomed. 2007; 20: 216-237.Crossref
- 26.
- Provencher SW. Automatic quantitation of localized in vivo 1H spectra with LCModel. NMR Biomed. 2001; 14: 260-264.Crossref
- 27.
- Lin A, Andronesi O, Bogner W, et al. Minimum reporting standards for in vivo magnetic resonance spectroscopy (MRSinMRS): experts’ consensus recommendations. NMR Biomed. 2021; 34: e4484.Crossref
- 28.
- Šidák, Z. Rectangular Confidence Regions for the Means of Multivariate Normal Distributions. J Am Stat Assoc. 1967; 62: 626-633. Crossref
- 29.
- R Core Team. R: a language and environment for statistical computing. R foundation for statistical computing. https://www.R-project.org/.Crossref
- 30.
- Chan C, Leeper T, Becker J, Schoch D. rio: a Swiss-army knife for data file I/O. https://cran.r-project.org/web/packages/rio/citation.html. Accessed June 14, 2024.Crossref
- 31.
- Lenth R. emmeans: Estimated Marginal Means, aka Least-Squares Means. R package version 1.8.9. 2023.Crossref
- 32.
- Lüdecke D. sjPlot: data visualization for statistics in social science. R package version 2.8.14. 2023. https://CRAN.R-project.org/package=sjPlot.Accessed June 14, 2024.Crossref
- 33.
- Lüdecke D, Ben-Shachar M, Patil I, Makowski D. Extracting, computing and exploring the parameters of statistical models using R. J Open Source Softw. 2020; 5: 2445.Crossref
- 34.
- Lüdecke D, Ben-Shachar MS, Patil I, et al. performance: an R package for assessment, comparison and testing of statistical models. J Open Source Softw. 2021; 6: 3139.Crossref
- 35.
- Makowski D, Lüdecke D, Patil I, et al. Automated results reporting as a practical tool to improve reproducibility and methodological best practices adoption. 2023. https://easystats.github.io/report/. Accessed June 14, 2024.Crossref
- 36.
- Venables WN, Ripley BD. Modern applied statistics with S, Fourth Edition. Springer; 2002. https://doi.org/10.1007/978-0-387-21706-2, Hardcover ISBN 978-0-387-95 457-8Crossref
- 37.
- Wickham H, Bryan J. readxl: Read excel files. R package version 1.3.1. 2019. https://CRAN.R-project.org/package=readxl. Accessed June 14, 2024.Crossref
- 38.
- Wickham H, François R, Henry L, Müller K, Vaughan D. dplyr: A grammar of data manipulation. R package version 1.1.2. 2023.Crossref
- 39.
- Krzyściak W, Bystrowska B, Karcz P, et al. Association of blood metabolomics biomarkers with brain metabolites and patient-reported outcomes as a new approach in individualized diagnosis of schizophrenia. Int J Mol Sci. 2024; 25: 2294.Crossref
- 40.
- Krzyściak W, Szwajca M, Śmierciak N, et al. From periphery immunity to central domain through clinical interview as a new insight on schizophrenia. Sci Rep. 2024; 14: 5755.Crossref
- 41.
- Whitaker KJ, Vértes PE, Romero-Garciaa R, et al. Adolescence is associated with genomically patterned consolidation of the hubs of the human brain connectome. Proc Natl Acad Sci USA. 2016; 113: 9105-9110.Crossref
- 42.
- Saccaro LF, Tassone M, Tozzi F, Rutigliano G. Proton magnetic resonance spectroscopy of N-acetyl aspartate in first depressive episode and chronic major depressive disorder: a systematic review and meta-analysis. J Affect Disord. 2024; 355: 265-282.Crossref
- 43.
- Su L, Cai Y, Xu Y, et al. Cerebral metbolism in major depressive disorder: a voxel-based meta-analysis of positron emission tomography studies. BMC Psychiatry. 2014; 14: 321.Crossref
- 44.
- Bonnekoh LM, Seidenbecher S, Knigge K, et al. Long-term cortisol stress response in depression and comorbid anxiety is linked with reduced N-acetylaspartate in the anterior cingulate cortex. World J Biol Psychiatry. 2023; 24: 34-45.Crossref
- 45.
- Webster MJ, O’Grady J, Kleinman JE, Weickert CS. Glial fibrillary acidic protein mRNA levels in the cingulate cortex of individuals with depression, bipolar disorder and schizophrenia. Neuroscience. 2005; 133: 453-461.Crossref
- 46.
- Jeon P, Mackinley M, Théberge J, Palaniyappan L. The trajectory of putative astroglial dysfunction in first episode schizophrenia: a longitudinal 7-Tesla MRS study. Sci Rep. 2021; 11: 22333.Crossref
- 47.
- Chiappelli J, Rowland LM, Wijtenburg SA, et al. Evaluation of myo-inositol as a potential biomarker for depression in schizophrenia. Neuropsychopharmacology. 2015; 40: 2157-2164.Crossref
- 48.
- Brunst KJ, Ryan PH, Altaye M, et al. Myo-inositol mediates the effects of traffic-related air pollution on generalized anxiety symptoms at age 12 years. Environ Res. 2019; 175: 71-78.Crossref
- 49.
- Howes O, McCutcheon R, Stone J. Glutamate and dopamine in schizophrenia: an update for the 21st century. J Psychopharmacol. 2015; 29: 97-115.Crossref
- 50.
- Merritt K, Egerton A, Kempton MJ, et al. Nature of glutamate alterations in schizophrenia: a meta-analysis of proton magnetic resonance spectroscopy studies. JAMA Psychiatry. 2016; 73: 665-674.Crossref