Decreased N-acetylaspartate plus N-acetyl-aspartyl-glutamate levels in the caudate of schizophrenia patients with tardive dyskinesia

Abstract

Tardive dyskinesia (TD) may reflect an intrinsic neurobiological vulnerability associated with schizophrenia, yet metabolic alterations in the caudate nucleu, a key brain region regulating involuntary movements-remain poorly characterized by proton magnetic resonance spectroscopy (1 H-MRS). We investigated the relationship between caudate nucleus metabolite concentrations and TD symptoms using 1 H-MRS. We recruited 117 patients with schizophrenia, including 67 patients with TD and 50 patients without TD (NT). We also recruited 41 healthy controls (HCs). Absolute metabolite concentrations of N-acetylaspartate plus N-acetyl-aspartyl-glutamate (tNAA), creatine (Cr) and glutamine plus glutamate (Glx) in the caudate nucleus were quantified using a 3.0-T MRI scanner with water signal referencing. Clinical symptoms were assessed using the Positive and Negative Syndrome Scale (PANSS) and the Abnormal Involuntary Movement Scale (AIMS). Among-group differences were analyzed using analysis of covariance with sex, age and years of education as covariates, followed by Bonferroni correction for multiple comparisons. After correction, the adjusted p-value was 0.017. Regression analysis was performed on metabolic indexes in the TD group with disease duration as a covariate. No significant differences in age, sex, or education level were observed among the groups. Additionally, no significant differences in PANSS scores or antipsychotic medication dosage were observed between the TD and NT groups. The TD group exhibited a significantly longer disease duration than the NT group (p < 0.05). Absolute tNAA levels—a marker of neuronal integrity—were significantly lower in the TD group than in the NT group (p < 0.05), while Cr (involved in energy metabolism) and Glx (involved in excitatory neurotransmission) levels did not significantly differ. These findings suggest reduced neuronal viability in the caudate nucleus in TD, though the clinical implications warrant further investigation.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-37396-z.

Introduction

Tardive dyskinesia (TD) is an abnormal involuntary movement disorder that typically emerges after prolonged exposure to dopamine receptor–blocking agents, including antipsychotics. TD can substantially compromise quality of life, treatment adherence, and long-term outcomes¹. Increasing evidence, however, suggests that TD is not merely a medication-induced adverse effect but may also reflect an intrinsic vulnerability—or even a manifestation of the disease process—in schizophrenia spectrum disorders^2,3. In this framework, TD is conceptualized within a broader continuum of sensorimotor abnormalities related to the neurobiology of psychosis.

Mechanistically, the dopamine receptor supersensitivity hypothesis proposes that chronic dopamine blockade triggers compensatory receptor upregulation and subsequent hyperdopaminergic signaling^4,5. Yet this account does not fully explain why TD may persist after antipsychotic discontinuation. Accordingly, alternative and potentially complementary mechanisms—including oxidative stress–related neuronal injury, mitochondrial dysfunction, apoptosis, and excitotoxicity—have gained prominence^6,7. Converging neuroimaging evidence indicates altered brain physiology and functional organization in TD^8,9, although consistent, reproducible structural markers remain elusive¹⁰. Nonetheless, several studies have reported basal ganglia abnormalities in TD, including reduced basal ganglia volume¹¹, particularly smaller caudate nucleus volume¹² and shortened T2 signal¹³. Additional findings include increased choroid plexus volume¹⁴ and regional gray matter reductions in the cuneus and lingual gyrus¹⁰. Functional neuroimaging further demonstrates aberrant local activity and connectivity patterns in TD^8,9,15.

Among basal ganglia structures, the caudate nucleus is of particular interest because it integrates cortical motor input and modulates striatal output, serving as a key brain region within the cortico–striato–thalamo–cortical circuit implicated in involuntary movements. Moreover, the caudate may be especially vulnerable to oxidative stress and excitotoxicity, processes increasingly implicated in TD. Together, these observations support the caudate nucleus as a compelling region of interest for probing TD-related neurochemical alterations.

Proton magnetic resonance spectroscopy (¹H-MRS) provides a noninvasive approach to quantify brain metabolites, including N-acetylaspartate plus N-acetylaspartylglutamate (tNAA), a putative marker of neuronal integrity; creatine (Cr), related to cellular energy metabolism; glutamate–glutamine (Glx), reflecting excitatory neurotransmission; and myo-inositol (mI), often interpreted as an index of gliosis or neuroinflammation. Prior MRS studies in TD—typically targeting the basal ganglia and reporting metabolite ratios (e.g., metabolite/Cr or NAA/mI)—have yielded mixed results. For example, Katsuhisa et al. reported increased Cho/Cr in the left lenticular nucleus in TD versus controls¹⁶, whereas Yevgeniy et al. observed reduced Cho/Cr within the striatum among TD patients with spontaneous oral dyskinesia¹⁷.

Metabolite alterations have also been documented across other dyskinesias, including Parkinson’s disease (PD) with dyskinesia¹⁸, Huntington’s disease (HD)¹⁹, and spinocerebellar ataxias (SCAs)²⁰, but with disorder-specific regional patterns (e.g., thalamic and motor cortical changes in PD^18,21, predominant basal ganglia changes in HD²¹, and cerebellar alterations in SCAs²⁰. These findings underscore the importance of region-specific investigation when characterizing the neurochemical substrates of distinct dyskinesia phenotypes.

Importantly, to our knowledge, no prior ¹H-MRS studies have specifically examined the caudate nucleus in TD. In addition, contemporary expert recommendations²² increasingly favor acquiring water-referenced data to enable reporting of absolute metabolite concentrations, which may improve interpretability compared with ratio-based measures. Building on evidence of basal ganglia involvement, oxidative stress, and potential neuronal compromise in TD, we hypothesized that caudate tNAA would be reduced in schizophrenia patients with TD compared with schizophrenia patients without TD and healthy controls (HCs).

Materials and methods

Clinical characteristics

A total of 117 patients with schizophrenia, including those with TD (n = 67), those without TD (NT, n = 50), and 41 HCs, were recruited in 2021 and 2022. The patients were recruited at Beijing Huilongguan Hospital, and the HC group comprised healthy residents drawn from the same period. All patients with schizophrenia met the following criteria: (1) satisfying the diagnostic criteria for schizophrenia listed in the Structured Clinical Interview for DSM-IV (SCID); (2) Han Chinese and aged between 18 and 60 years; (3) having a minimum 5-year illness duration; and (4) exposure to steady antipsychotic dosages for at least 6 months prior to study admission. Subjects were excluded if they met one of the following criteria: (1) a DSM-IV Axis I diagnosis other than schizophrenia; (2) serious, acute, unstable and/or significant and untreated medical illness (e.g. infection, diabetes, and uncontrolled hypertension); (3) a history of brain trauma or neurological disease; (4) pregnant or breast-feeding; (5) taken neurotrophic or free radical metabolism drugs within the 12 weeks prior to participation. All patients were receiving second-generation antipsychotic treatment at the time of enrollment.

TD was diagnosed using Schooler and Kane’s Research Diagnostic Criteria (1982)²³, and its severity was assessed using the Abnormal Involuntary Movements Scale (AIMS). The TD clinical ratings were performed by two experienced psychiatrists who were clinically trained to consistently perform AIMS testing as part of the general hospital training procedures, although they were not specifically trained for this study. In addition, patients with TD were re-evaluated for AIMS score by the same psychiatrist at least one month later, and were diagnosed with TD only if both evaluations yielded positive results. The AIMS score in the NT group was 0.

The research protocol was approved by the Ethics Committee of Beijing Huilongguan Hospital, and all participants provided written informed consent according to the Declaration of Helsinki.

Clinical assessments

The Positive and Negative Syndrome Scale (PANSS) was used for psychopathological assessment. The AIMS was used to evaluate the severity of TD based on the diagnostic criteria of Schooler and Kane²³. Each item on the AIMS ranges from 0 to 4, with items 1–4 assessing orofacial movements and items 5–7 assessing limb-truncal dyskinesia²⁴. TD showed an AIMS score ≥ 3 (moderate degree) in any body part or ≥ 2 (mild degree) in two or more body parts.

Imaging data acquisition

Image data were collected using a 3.0-T MRI scanner (PRISMA; Siemens Healthineers, Erlangen, Germany) equipped with a 64-channel radiofrequency head coil. Participants’ heads were secured with a foam pad to minimize movement. The MRI protocol comprised 3D-T1 structural phase image and MRS sequence. For 3D-T1 structural image acquisition, the scan encompassed the entire brain using a 3D gradient echo imaging sequence (magnetization-prepared rapid gradient echo [MP-RAGE]). The T1-weighted phase scanning parameters were as follows: echo time (TE) = 2.98 ms; inversion time (TI) = 1100 ms; repetition time (TR) = 2530 ms; flip angle (FA) = 9°; field of view (FOV) = 240 × 256 mm²; matrix = 256 × 240; voxel size = 1 × 1 × 1 mm³; slice thickness = 1 mm; slice spacing = 0; and scanning = 208 slices. Single voxel Point RESolved spectroscopy (PRESS: TR = 2000 ms, TE = 30 ms, FA = 90^o, voxel size = 20 × 20 × 20 mm³, 192 averages) acquisitions were used. An additional PRESS acquisition in the same voxel without water suppression (16 averages) was collected. The spectroscopy voxel was placed in the caudate area since it is a region demonstrating degeneration in the course of TD. An example voxel placement is shown in Fig. 1.

Fig. 1.

Illustration of MRS caudate voxel placement and representative spectra. The left is T1-weighted MR images of TD; the right is ¹H MR spectrum acquired by PRESS sequence in the caudate of a patient with TD.

Metabolite concentration

Quantitative analysis of tNAA, Cr and Glx was performed using the LCModel software package (version 6.3), with the acquisition without water suppression as an internal reference. The absolute concentrations of tNAA, Cr and Glx were calculated, and corrections were made for white matter and cerebrospinal fluid content in the caudate nucleus. All analyses were conducted in accordance with the LCModel manual (http://lcmodel.com/lcm-manual.shtml).

Statistical analysis

Normality of continuous variables was assessed using the Shapiro-Wilk test. Demographic data (sex, age, and years of education) among the three groups were analyzed using analysis of variance (ANOVA) or chi-square tests. Disease course, initial age, CPZ and PANSS scores in the TD vs. NT group were assessed using a t-test. Differences in brain metabolism indicators among three groups were compared using analysis of covariance (ANCOVA) with sex, age and years of education as covariates, followed by Bonferroni correction for multiple comparisons. After correction, the adjusted p-value was 0.017. Metabolism indicators showing significant group differences in the three-group were further examined by ANCOVA to compare the TD and NT group, with disease duration included as a covariate. Regression analysis was performed on metabolic indices in the TD group with disease duration as a covariate. Statistical significance was set at P < 0.05.

Results

No significant differences were observed between the groups in terms of sex, age, or years of education, PANSS subscale score, total score, or antipsychotic drug to chlorpromazine equivalent. The disease course in the TD group is longer than that in the NT group (Table 1). Significant inter-group differences were observed in tNAA levels in the caudate (Table 2). After Bonferroni correction, tNAA level in the caudate was significantly lower in the TD group than in the NT group. The ANCOVA comparing tNAA level between the TD and NT group, with disease duration included as a covariate, showed a significant group effect (F = 4.528, P = 0.002). Regression analysis indicated that, after controlling for disease duration, there was no significant association between tNAA levels in the caudate and AIMS scores in the TD group (Table 3).

Table 1.

Demographics and clinical characteristics of participants.

Sex (male/female)	HC (n = 41)	NT (n = 50)	TD (n = 67)	F/T/χ2	P
	23/18	29/21	39/28	0.051	0.975
Age (years) (x̄± S)	51 ± 6	51 ± 7	53 ± 8	1.115	0.331
Years of education (years) (x̄± S)	11.05 ± 3.66	11.46 ± 3.05	11.61 ± 2.81	0.420	0.658
Disease course (years) (x̄± S)	NA	25.52 ± 9.58	29.61 ± 9.93	−2.238	0.027
Initial age (years) (x̄± S)	NA	26.24 ± 6.85	23.59 ± 7.87	1.897	0.060
CPZ (mg) (x̄± S)	NA	486.0 ± 286.79	482.99 ± 240.99	0.062	0.950
PANSS
Positive subscore (x̄± S)	NA	16.80 ± 7.3	15.74 ± 6.03	0.834	0.405
Negative subscore (x̄± S)	NA	24.77 ± 5.89	24.59 ± 6.09	0.148	0.882
General psychopathological subscore (x̄± S)	NA	24.45 ± 12.55	25.69 ± 6.02	−0.676	0.500
Total score (x̄± S)	NA	66.27 ± 21.41	66.03 ± 11.89	0.073	0.941
AIMS severity of abnormal movements (x̄± S)	NA	NA	2.49 ± 1.01	NA	NA
AIMS Total score (x̄± S)	NA	NA	9.15 ± 4.56	NA	NA

Demographic and clinical characteristics. p-value indicates significant differences (P < 0.05) in appropriate statistical tests. HC, healthy control; NT, schizophrenia without tardive dyskinesia; TD, schizophrenia with tardive dyskinesia; x̄, mean; S, standard deviation; CPZ, Antipsychotic to chlorpromazine equivalent; PANSS: Positive and Negative Syndrome Scale; AIMS: Abnormal Involuntary Movement Scale.

Table 2.

Comparison of metabolite levels between HC, NT, and TD groups.

	HC (n = 41)	NT (n = 50)	TD (n = 67)	F	P	P (HC-NT)	P (HC-TD)	P (NT-TD)
	50 (25–75%)	50 (25–75%)	50 (25–75%)
Cr	5.731 (5.528–6.041)	6.018 (5.51–6.655)	5.946 (5.155–6.417)	0.384	0.858	0.904	0.617	0.479
tNAA	5.555 (5.197–5.937)	5.803 (5.403–6.258)	5.219 (4.936–5.804)	3.513	0.007**	0.612	0.158	0.031*/0.002 #
Glx	1.149 (0.157–2.647)	0.966 (0.004–2.144)	0.308 (0.031–2.343)	1.454	0.216	0.649	0.839	0.778

Values are presented as median (25th–75th percentile). Differences in brain metabolism indicators among three groups were compared using analysis of covariance with sex, age and years of education, followed by Bonferroni correction for multiple comparisons. HC: Healthy control; NT: Schizophrenia patients without tardive dyskinesia; TD: Schizophrenia patients with tardive dyskinesia; tNAA: N-acetylaspartate plus N-acetyl-aspartyl-glutamate; Cr: creatine; Glx: glutamine plus glutamate.*: P < 0.05; **: P < 0.017 (0.017 = 0.05/3); #: The ANCOVA comparing tNAA level between the TD and NT group, with disease duration included as a covariate.

Table 3.

Correlations between caudate tNAA level and clinical characteristics in schizophrenia patients with tardive dyskinesia.

	Caudate tNAA level
	β value	P value
PANSS
Positive subscore	0.318	0.102
Negative subscore	−0.065	0.750
General psychopathological subscore	−0.032	0.873
Total score	0.111	0.608
AIMS severity of abnormal movements	0.033	0.864
AIMS Total score	0.010	0.963

Relationship between caudate tNAA level and clinical characteristics in schizophrenia patients with tardive dyskinesia. PANSS, Positive and Negative Syndromes Scale; Positive score AIMS, Abnormal Involuntary Movements Scale; tNAA: N-acetylaspartate plus N-acetyl-aspartyl-glutamate. P-value indicate significant differences (P < 0.05) in appropriate statistical tests.

Discussion

The primary aim of this study was to address a question that has been inconsistent in previous research: whether MRS can be used to detect biomarkers for TD. Therefore, we focused on tNAA. Because tNAA is a metabolite primarily present in neurons, its concentration reflects neuronal integrity and viability. We compared differences among healthy controls, patients with schizophrenia with and without TD. Additionally, we investigated the relationship between these metabolites and abnormal involuntary movements, as the primary clinical manifestation of TD is abnormal involuntary movement. A strength of this study is the normalization of metabolites to unsuppressed water signal, a technique previously reported in other dyskinesia spectroscopy studies, e.g., HD²⁵. By avoiding the standardization of metabolites to other metabolites (e.g., tCr), which are also affected in the TD brain, our measurements are more likely to reflect true biochemical changes and enable direct comparisons across studies. Furthermore, our study involved a larger sample size than previous studies. Unlike earlier high-definition MRS studies that employed a 1.5T scanner, the 3 T MRS used in this study provided a higher signal-to-noise ratio.

Reduced tNAA levels have been observed in various movement disorders. In early HD patients, the tNAA level in the putamen was found to be lower than that in controls^25,26. In SCA14, the tNAA level in the cerebellar vermis was reduced by 34% compared to healthy controls, while in SCA1, it was reduced by 24%²⁷. In addition to the cerebellar vermis, tNAA levels in the cerebellar hemispheres and pons were also reduced in SCA1 compared to controls²⁸. Moreover, some tNAA levels were associated with disease severity scores. Our study found differences in caudate tNAA concentrations among the three groups, suggesting the presence of neuronal alterations in the caudate nucleus that may be associated with vulnerability to TD. tNAA was lowest in the TD group, while it was moderately reduced in the NT group, indicating that schizophrenia itself may involve neuronal changes in the caudate nucleus, and prolonged antipsychotic exposure may further exacerbate these changes.

The underlying causes of these changes in patients with TD may involve oxidative stress, inflammation, or cerebral hypoperfusion. Previous research has suggested that oxidative stress plays a critical role in the pathophysiology of TD. Increased oxidative stress may lead to D2 receptor supersensitivity and neurodegenerative changes in neurons, resulting in secondary synaptic plasticity abnormalities in striatal medium spiny neurons. This, in turn, disrupts the balance of neurotransmitter metabolism between the direct and indirect pathways of the basal ganglia, leading to abnormal output to the sensorimotor cortex. Long-term use of dopamine receptor-blocking agents can induce maladaptive synaptic plasticity, causing abnormalities in both efferent projections from the basal ganglia to the sensorimotor cortex and afferent projections from the cortex, ultimately leading to the occurrence of abnormal involuntary movements²⁹. Maladaptive cortical synaptic plasticity, combined with abnormal basal ganglia output, may result in motor program encoding errors and the emergence of abnormal movements. These findings raise the possibility that oxidative stress may be involved in the reduction of caudate tNAA levels in patients with TD. Given that NAA is synthesized in neuronal mitochondria and is widely interpreted as a marker of neuronal integrity and mitochondrial energetic status, reduced caudate tNAA in TD may preliminarily index impaired neuronal bioenergetics or neuroaxonal dysfunction within dorsal striatal motor circuits. It is also possible that AIMS scores primarily capture the current expression of abnormal movements, whereas tNAA alterations may reflect long-term or cumulative neuronal changes that are not linearly related to cross-sectional symptom severity.Previous research by our group found that the association between immunosenescence-related T cell subpopulations and brain structure may underlie the pathophysiological processes of TD, indicating that immune dysfunction could also be a contributing factor in the present study³⁰. Additionally, our prior research demonstrated reduced functional connectivity in the right postcentral gyrus and the left triangular part of the inferior frontal gyrus in TD patients⁸. We also observed increased ReHo values in the left semilunar lobule and decreased ReHo values in the left lateral ventricle⁹. Although no functional imaging abnormalities were detected in the caudate nucleus, these findings still suggest abnormal cerebral blood flow signals in TD. This raises the possibility that the reduction in caudate tNAA levels in patients with TD might be associated with impaired cerebral perfusion. Based on the theory of neuronal degeneration and apoptosis, we speculate that it may be related to caudate hypoperfusion. In addition, the duration of TD in this study is longer than that of NT. The long disease duration may be associated with a decrease in tNAA concentration. This is because patients with a longer disease course may have experienced more neuronal damage or neurodegenerative changes, leading to neuronal injury or loss, which in turn reduces the tNAA levels³¹. Importantly, the performed regression analysis did not reveal a significant association between caudate tNAA concentrations and AIMS scores within the TD group. Therefore, the present findings should be interpreted cautiously, and the observed tNAA reduction is more likely to reflect a trait-like neurobiological vulnerability. This may be due to the limited sample size or the narrow range of AIMS scores, resulting in the absence of a detectable association.

Earlier research on TD-related metabolites found no differences in NAA/Cr ratios in the bilateral lentiform nucleus between schizophrenia patients without TD and healthy controls. This could be because the key site of metabolite alterations in patients with TD is not the lentiform nucleus but rather the caudate nucleus, consistent with prior structural imaging studies of TD. Alternatively, the resolution of the MRI scanner used in those studies may have been insufficient. Additionally, small sample sizes may have limited the detection of subtle NAA changes. Another possibility is that abnormalities in lentiform nucleus metabolites involved synchronous reductions in both NAA and Cr, resulting in an unchanged ratio. Our findings of reduced absolute tNAA concentrations in the caudate nucleus suggest that the reduction may be driven primarily by decreases in NAAG. NAAG regulates glutamatergic synaptic activity by activating presynaptic metabotropic glutamate receptors. Additionally, NAAG exerts neuroprotective effects by inhibiting excessive glutamate release, thereby mitigating glutamate-mediated neurotoxicity. NAAG reduction may increase glutamate-mediated neurotoxic effects, exacerbating neuronal degeneration and apoptosis. In both TD animal models and patients, cerebrospinal fluid analyses have revealed elevated Glu levels compared to controls^32,33. In TD mice, dopaminergic signaling was found to be increased, while glutamatergic signaling was blocked in thalamocortical neurons³⁴. Subcortical Glu uptake was reduced in a rat model treated with gabapentin and haloperidol, and this reduction was negatively correlated with the increased incidence of orofacial movement disorders. Hence, early changes in Glu transport may be involved in masticatory dysfunction³⁵. Therefore, we speculate that NAAG deficiency in the caudate nucleus may enhance glutamate-mediated neurotoxicity, playing a role in the pathogenesis and progression of TD.

Previous studies suggested changes in the ratio of metabolites to Cr in TD, and an increase in glutamate in TD was found in animal experiments³⁵. Therefore, we also detected Cr and Glx in caudate nucleus at the same time, but no differences between groups were found. Previous studies have found that the left putamen Cho/Cr ratio¹⁶ is elevated in patients with TD, suggesting that excessive damage to the neurons in the left putamen may contribute to the development of TD. Compared with HC, the striatal Cho/Cr ratio is significantly reduced in patients with drug-induced TD and spontaneous oral motor disorders¹⁷. This indicates a decrease in the metabolic activity of the striatum, particularly a reduction in metabolic activities related to cell membrane synthesis and neurotransmitter synthesis, leading to a decrease in choline concentration. This study did not find differences in the absolute concentrations of Cr and Cho among the three groups. This suggests that these changes may be relative, rather than reflecting overall metabolic or concentration changes. It is also possible that the changes in Cr and Cho occur in regions other than the caudate nucleus, such as the putamen and other microstructures of the striatum, or the degree of concentration change is small and beyond the scope of this study.

However, our study has certain limitations. For instance, we conducted measurements at a single time point and did not observe the longitudinal changes in these biomarkers. Additionally, there was heterogeneity in the disease duration among our participants, which may have influenced the results. Finally, we did not observe a clear medication dose effect across groups, which may limit the strength of our findings.

In conclusion, our findings indicate that reduced caudate tNAA is associated with the presence of TD. However, given the absence of a significant association with AIMS scores, caudate tNAA should be considered a preliminary and indirect biomarker reflecting neurobiological vulnerability rather than TD symptom severity. Longitudinal and interventional studies are required to determine its clinical utility.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Ting Yu and Yanli Li: Software, Methodology, Formal analysis, Writing-Original Draft, and VisualizationNa Li: Investigation and Data CurationMengzhuang Gou and Wenjin Chen: InvestigationZhiren Wang and Shuping Tan: Resources and ValidationFude Yang and Baopeng Tian: SupervisionXingguang Luo: Writing-Review and EditingYunlong Tan: Conceptualization, Funding acquisition, Writing-Review and Editing.

Funding

National Natural Science Foundation of China (Grant No. 81771452); Beijing Natural Science Foundation (Grant No. 7212054); Beijing Hospitals Authority Youth Program (Grant No. QML20232001); Capital’s Funds for Health Improvement and Research (Grant No. 2024-1-4111).

Data availability

The data associated with this study have not been deposited in any publicly available repository however, they are available from the corresponding author upon request.

Declarations

Competing interests

The authors declare no competing interests.