Advertisement

Dopamine Synthesis Capacity and GABA and Glutamate Levels Separate Antipsychotic-Naïve Patients With First-Episode Psychosis From Healthy Control Subjects in a Multimodal Prediction Model

  • Anne K. Sigvard
    Correspondence
    Address correspondence to Anne K. Sigvard, M.D.
    Affiliations
    Center for Neuropsychiatric Schizophrenia Research & Center for Clinical Intervention and Neuropsychiatric Schizophrenia Research, Mental Health Center, Glostrup, Copenhagen University Hospital – Mental Health Services CPH, Copenhagen, Denmark

    Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for articles by this author
  • Kirsten Borup Bojesen
    Affiliations
    Center for Neuropsychiatric Schizophrenia Research & Center for Clinical Intervention and Neuropsychiatric Schizophrenia Research, Mental Health Center, Glostrup, Copenhagen University Hospital – Mental Health Services CPH, Copenhagen, Denmark
    Search for articles by this author
  • Karen S. Ambrosen
    Affiliations
    Center for Neuropsychiatric Schizophrenia Research & Center for Clinical Intervention and Neuropsychiatric Schizophrenia Research, Mental Health Center, Glostrup, Copenhagen University Hospital – Mental Health Services CPH, Copenhagen, Denmark
    Search for articles by this author
  • Mette Ødegaard Nielsen
    Affiliations
    Center for Neuropsychiatric Schizophrenia Research & Center for Clinical Intervention and Neuropsychiatric Schizophrenia Research, Mental Health Center, Glostrup, Copenhagen University Hospital – Mental Health Services CPH, Copenhagen, Denmark

    Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for articles by this author
  • Albert Gjedde
    Affiliations
    Department of Neuroscience, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark

    Translational Neuropsychiatry Unit, Aarhus University, Aarhus, Denmark

    Department of Clinical Research, University of Southern Denmark, Odense, Denmark
    Search for articles by this author
  • Karen Tangmose
    Affiliations
    Center for Neuropsychiatric Schizophrenia Research & Center for Clinical Intervention and Neuropsychiatric Schizophrenia Research, Mental Health Center, Glostrup, Copenhagen University Hospital – Mental Health Services CPH, Copenhagen, Denmark
    Search for articles by this author
  • Yoshitaka Kumakura
    Affiliations
    Department of Diagnostic Radiology and Nuclear Medicine, Saitama Medical Center, Saitama Medical University, Japan
    Search for articles by this author
  • Richard Edden
    Affiliations
    Russell H. Morgan Department of Radiology and Radiological Science, The Johns Hopkins University School of Medicine, Baltimore, Maryland

    FM. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, Maryland
    Search for articles by this author
  • Dan Fuglø
    Affiliations
    Department of Nuclear Medicine, Herlev Hospital, University of Copenhagen, Copenhagen, Denmark
    Search for articles by this author
  • Lars Thorbjørn Jensen
    Affiliations
    Department of Nuclear Medicine, Herlev Hospital, University of Copenhagen, Copenhagen, Denmark
    Search for articles by this author
  • Egill Rostrup
    Affiliations
    Center for Neuropsychiatric Schizophrenia Research & Center for Clinical Intervention and Neuropsychiatric Schizophrenia Research, Mental Health Center, Glostrup, Copenhagen University Hospital – Mental Health Services CPH, Copenhagen, Denmark
    Search for articles by this author
  • Bjørn H. Ebdrup
    Affiliations
    Center for Neuropsychiatric Schizophrenia Research & Center for Clinical Intervention and Neuropsychiatric Schizophrenia Research, Mental Health Center, Glostrup, Copenhagen University Hospital – Mental Health Services CPH, Copenhagen, Denmark

    Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for articles by this author
  • Birte Yding Glenthøj
    Affiliations
    Center for Neuropsychiatric Schizophrenia Research & Center for Clinical Intervention and Neuropsychiatric Schizophrenia Research, Mental Health Center, Glostrup, Copenhagen University Hospital – Mental Health Services CPH, Copenhagen, Denmark

    Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Search for articles by this author
Open AccessPublished:May 30, 2022DOI:https://doi.org/10.1016/j.bpsgos.2022.05.004

      Abstract

      Background

      Disturbances in presynaptic dopamine activity and levels of GABA (gamma-aminobutyric acid) and glutamate plus glutamine collectively may have a role in the pathophysiology of psychosis, although separately they are poor diagnostic markers. We tested whether these neurotransmitters in combination improve the distinction of antipsychotic-naïve patients with first-episode psychosis from healthy control subjects.

      Methods

      We included 23 patients (mean age 22.3 years, 9 male) and 20 control subjects (mean age 22.4 years, 8 male). We determined dopamine metabolism in the nucleus accumbens and striatum from 18F-fluorodopa (18F-FDOPA) positron emission tomography. We measured GABA levels in the anterior cingulate cortex (ACC) and glutamate plus glutamine levels in the ACC and left thalamus with 3T proton magnetic resonance spectroscopy. We used binominal logistic regression for unimodal prediction when we modeled neurotransmitters individually and for multimodal prediction when we combined the 3 neurotransmitters. We selected the best combination based on Akaike information criterion.

      Results

      Individual neurotransmitters failed to predict group. Three triple neurotransmitter combinations significantly predicted group after Benjamini-Hochberg correction. The best model (Akaike information criterion 48.5) carried 93.5% of the cumulative model weight. It reached a classification accuracy of 83.7% (p = .003) and included dopamine synthesis capacity (Ki4p) in the nucleus accumbens (p = .664), GABA levels in the ACC (p = .019), glutamate plus glutamine levels in the thalamus (p = .678), and the interaction term Ki4p × GABA (p = .016).

      Conclusions

      Our multimodal approach proved superior classification accuracy, implying that the pathophysiology of patients represents a combination of neurotransmitter disturbances rather than aberrations in a single neurotransmitter. Particularly aberrant interrelations between Ki4p in the nucleus accumbens and GABA values in the ACC appeared to contribute diagnostic information.

      Keywords

      We consider dysregulated dopamine, GABA (gamma-aminobutyric acid), and glutamate transmitter systems to be key elements in the development of psychosis. The dopamine hypothesis has a long history (
      • Howes O.D.
      • Kapur S.
      The dopamine hypothesis of schizophrenia: version III--the final common pathway.
      ), but alternative hypotheses of more complex disturbances of neurochemical circuits involving GABA and glutamate have been proposed (
      • Carlsson A.
      The neurochemical circuitry of schizophrenia.
      ). However, because of clinical and technical challenges, only limited empirical data have been available for testing whether disturbances of the neurotransmitter systems of dopamine, GABA, and glutamate interact in the early stages of psychosis, and joint measures of all 3 neurotransmitters have not previously been available for the purpose of testing this claim.
      The effect size of the evidence of elevated dopamine synthesis capacity (DSC) in psychosis appears to have lessened over time (
      • Reith J.
      • Benkelfat C.
      • Sherwin A.
      • Yasuhara Y.
      • Kuwabara H.
      • Andermann F.
      • et al.
      Elevated dopa decarboxylase activity in living brain of patients with psychosis.
      ,
      • Howes O.D.
      • Kambeitz J.
      • Kim E.
      • Stahl D.
      • Slifstein M.
      • Abi-Dargham A.
      • Kapur S.
      The nature of dopamine dysfunction in schizophrenia and what this means for treatment.
      ,
      • Lindström L.H.
      • Gefvert O.
      • Hagberg G.
      • Lundberg T.
      • Bergström M.
      • Hartvig P.
      • Långström B.
      Increased dopamine synthesis rate in medial prefrontal cortex and striatum in schizophrenia indicated by L-(beta-11C) DOPA and PET.
      ,
      • Hietala J.
      • Syvälahti E.
      • Vilkman H.
      • Vuorio K.
      • Räkköläinen V.
      • Bergman J.
      • et al.
      Depressive symptoms and presynaptic dopamine function in neuroleptic-naive schizophrenia.
      ,
      • Mcgowan S.
      • Lawrence A.D.
      • Sales T.
      • Quested D.
      • Grasby P.
      Presynaptic dopaminergic dysfunction in schizophrenia: A positron emission tomographic [18F]fluorodopa study.
      ,
      • Cheng P.W.C.
      • Chang W.C.
      • Lo G.G.
      • Chan K.W.S.
      • Lee H.M.E.
      • Hui L.M.C.
      • et al.
      The role of dopamine dysregulation and evidence for the transdiagnostic nature of elevated dopamine synthesis in psychosis: A positron emission tomography (PET) study comparing schizophrenia, delusional disorder, and other psychotic disorders.
      ,
      • Brugger S.P.
      • Angelescu I.
      • Abi-dargham A.
      • Mizrahi R.
      • Shahrezaei V.
      • Howes O.D.
      Heterogeneity of striatal dopamine function in schizophrenia: Meta-analysis of variance.
      ,
      • Hietala J.
      • Syvälahti E.
      • Vuorio K.
      • Räkköläinen V.
      • Bergman J.
      • Haaparanta M.
      • et al.
      Presynaptic dopamine function in striatum of neuroleptic-naive schizophrenic patients.
      ,
      • Dao-Castellana M.H.
      • Paillère-Martinot M.L.
      • Hantraye P.
      • Attar-Lévy D.
      • Rémy P.
      • Crouzel C.
      • et al.
      Presynaptic dopaminergic function in the striatum of schizophrenic patients.
      ,
      • Kumakura Y.
      • Cumming P.
      • Vernaleken I.
      • Buchholz H.G.
      • Siessmeier T.
      • Heinz A.
      • et al.
      Elevated [18F]fluorodopamine turnover in brain of patients with schizophrenia: An [18F]fluorodopa/positron emission tomography study.
      ,
      • Nozaki S.
      • Kato M.
      • Takano H.
      • Ito H.
      • Takahashi H.
      • Arakawa R.
      • et al.
      Regional dopamine synthesis in patients with schizophrenia using L-[β-11C]DOPA PET.
      ,
      • Katthagen T.
      • Kaminski J.
      • Heinz A.
      • Buchert R.
      • Schlagenhauf F.
      Striatal dopamine and reward prediction error signaling in unmedicated schizophrenia patients.
      ,
      • Egerton A.
      • Murphy A.
      • Donocik J.
      • Anton A.
      • Barker G.J.
      • Collier T.
      • et al.
      Dopamine and glutamate in antipsychotic-responsive compared with antipsychotic-nonresponsive psychosis: A multicenter positron emission tomography and magnetic resonance spectroscopy study (STRATA).
      ), and neither of the 2 largest first-episode studies replicated elevated DSC (
      • Jauhar S.
      • Mccutcheon R.
      • Borgan F.
      • Veronese M.
      • Nour M.
      • Pepper F.
      • et al.
      The relationship between cortical glutamate and striatal dopamine in first-episode psychosis: a cross-sectional multimodal PET and magnetic resonance spectroscopy imaging study.
      ,
      • Sigvard A.K.
      • Nielsen M.Ø.
      • Gjedde A.
      • Bojesen K.B.
      • Fuglø D.
      • Tangmose K.
      • et al.
      Dopaminergic activity in antipsychotic-naïve patients assessed with positron emission tomography before and after partial dopamine D 2 receptor agonist treatment: Association with psychotic symptoms and treatment response.
      ). Using refined 4-parameter (4P) modeling of DSC in antipsychotic-naïve patients with first-episode psychosis, we previously found the 4P model to be superior to the conventional model. Particularly, we found the striatal decarboxylation rate of 18F-fluorodopa (18F-FDOPA) to 18F-dopamine (k3) to be associated with psychotic symptoms and treatment outcome (
      • Sigvard A.K.
      • Nielsen M.Ø.
      • Gjedde A.
      • Bojesen K.B.
      • Fuglø D.
      • Tangmose K.
      • et al.
      Dopaminergic activity in antipsychotic-naïve patients assessed with positron emission tomography before and after partial dopamine D 2 receptor agonist treatment: Association with psychotic symptoms and treatment response.
      ). Yet, neither k3 nor DSC estimates from the 4P or conventional model (Ki4p and Kicer, respectively) discriminated patients from healthy control subjects (HCs), implying that disturbances of dopamine metabolism may be rather discrete at early stages of illness. Another explanation may be that patients are better characterized by combinations of neurotransmitter events in the cortico-striato-thalamo-cortical networks, such as by striatal dopaminergic activity combined with glutamatergic and GABAergic activities in the thalamus and prefrontal cortex (PFC).
      Knowledge of GABAergic involvement in psychosis largely is based on animal models and postmortem studies (
      • Nakazawaa K.
      • Zsirosa V.
      • Jianga Z.
      • Nakao K.
      • Kolata S.
      • Zhanga S.
      • Belforte J.E.
      GABAergic interneuron origin of schizophrenia pathophysiology.
      ,
      • Schmidt M.J.
      • Mirnics K.
      Neurodevelopment, GABA system dysfunction, and schizophrenia.
      ,
      • Fazzari P.
      • Paternain A.V.
      • Valiente M.
      • Pla R.
      • Luján R.
      • Lloyd K.
      • et al.
      Control of cortical GABA circuitry development by Nrg1 and ErbB4 signalling.
      ,
      • Gonzalez-Burgos G.
      • Fish K.N.
      • Lewis D.A.
      GABA neuron alterations, cortical circuit dysfunction and cognitive deficits in schizophrenia.
      ), and few clinical studies exist. Findings of GABA levels in frontal brain regions, typically the anterior cingulate cortex (ACC) and medial PFC, ranged from increased (
      • Öngür D.
      • Prescot A.P.
      • Mccarthy J.
      • Cohen B.M.
      • Renshaw P.F.
      Elevated gamma-aminobutyric acid levels in chronic schizophrenia.
      ,
      • Kegeles L.S.
      • Mao X.
      • Stanford A.D.
      • Girgis R.
      • Ojeil N.
      • Xu X.
      • et al.
      Elevated prefrontal cortex γ-aminobutyric acid and glutamate-glutamine levels in schizophrenia measured in vivo with proton magnetic resonance spectroscopy.
      ,
      • de la Fuente-Sandoval C.
      • Reyes-Madrigal F.
      • Mao X.
      • León-Ortiz P.
      • Rodríguez-Mayoral O.
      • Jung-Cook H.
      • et al.
      Prefrontal and striatal gamma-aminobutyric acid levels and the effect of antipsychotic treatment in first-episode psychosis patients.
      ,
      • Yang Z.
      • Zhu Y.
      • Song Z.
      • Mei L.
      • Zhang J.
      • Chen T.
      • et al.
      Comparison of the density of gamma-aminobutyric acid in the ventromedial prefrontal cortex of patients with first-episode psychosis and healthy controls.
      ,
      • De La Fuente-Sandoval C.
      • Reyes-Madrigal F.
      • Mao X.
      • León-Ortiz P.
      • Rodríguez-Mayoral O.
      • Solís-Vivanco R.
      • et al.
      Cortico-striatal GABAergic and glutamatergic dysregulations in subjects at ultra-high risk for psychosis investigated with proton magnetic resonance spectroscopy.
      ) to similar (
      • Goto N.
      • Yoshimura R.
      • Kakeda S.
      • Moriya J.
      • Hori H.
      • Hayashi K.
      • et al.
      No alterations of brain GABA after 6 months of treatment with atypical antipsychotic drugs in early-stage first-episode schizophrenia.
      ,
      • Tayoshi S.
      • Nakataki M.
      • Sumitani S.
      • Taniguchi K.
      • Shibuya-Tayoshi S.
      • Numata S.
      • et al.
      GABA concentration in schizophrenia patients and the effects of antipsychotic medication: A proton magnetic resonance spectroscopy study.
      ,
      • Sydnor V.J.
      • Roalf D.R.
      A meta-analysis of ultra-high fi eld glutamate, glutamine, GABA and glutathione 1HMRS in psychosis: Implications for studies of psychosis risk.
      ) or decreased (
      • Marsman A.
      • Mandl R.C.W.
      • Klomp D.W.J.
      • Bohlken M.M.
      • Boer V.O.
      • Andreychenko A.
      • et al.
      GABA and glutamate in schizophrenia: A 7 T 1H-MRS study.
      ,
      • Rowland L.M.
      • Kontson K.
      • West J.
      • Edden R.A.
      • Zhu H.
      • Wijtenburg S.A.
      • et al.
      In vivo measurements of glutamate, GABA, and NAAG in schizophrenia.
      ,
      • Rowland L.M.
      • Krause B.W.
      • Wijtenburg S.A.
      • Mcmahon R.P.
      • Chiappelli J.
      • Nugent K.L.
      • et al.
      Medial frontal GABA is lower in older schizophrenia: A MEGA-PRESS with macromolecule suppression study.
      ) GABA levels in patients with psychosis compared with HCs. A recent positron emission tomography (PET) study found decreased GABAA receptor levels in the hippocampus in antipsychotic-free patients but not in medicated patients (
      • Marques T.R.
      • Ashok A.H.
      • Angelescu I.
      • Borgan F.
      • Myers J.
      • Lingford-Hughes A.L.
      • et al.
      GABA-A receptor differences in schizophrenia: A positron emission tomography study using [11C]Ro154513.
      ).
      The literature of glutamate and glutamate plus glutamine (Glx) levels in patients with psychosis compared with HCs likewise is at variance with data indicating increased (
      • Kegeles L.S.
      • Mao X.
      • Stanford A.D.
      • Girgis R.
      • Ojeil N.
      • Xu X.
      • et al.
      Elevated prefrontal cortex γ-aminobutyric acid and glutamate-glutamine levels in schizophrenia measured in vivo with proton magnetic resonance spectroscopy.
      ,
      • Théberge J.
      • Bartha R.
      • Drost D.J.
      • Menon R.S.
      • Malla A.
      • Takhar J.
      • et al.
      Glutamate and glutamine measured with 4.0 T proton MRS in never-treated patients with schizophrenia and healthy volunteers.
      ,
      • Théberge J.
      • Williamson K.E.
      • Aoyama N.
      • Drost D.J.
      • Manchanda R.
      • Malla A.K.
      • et al.
      Longitudinal grey-matter and glutamatergic losses in first-episode schizophrenia.
      ,
      • Bustillo J.R.
      • Rowland L.M.
      • Mullins P.
      • Jung R.
      • Chen H.
      • Qualls C.
      • et al.
      1H-MRS at 4 Tesla in minimally treated early schizophrenia.
      ), decreased (
      • Sydnor V.J.
      • Roalf D.R.
      A meta-analysis of ultra-high fi eld glutamate, glutamine, GABA and glutathione 1HMRS in psychosis: Implications for studies of psychosis risk.
      ,
      • Wang J.
      • Tang Y.
      • Zhang T.
      • Cui H.
      • Xu L.
      • Zeng B.
      • et al.
      Reduced γ-aminobutyric acid and glutamate+glutamine levels in drug-naïve patients with first-episode schizophrenia but not in those at ultrahigh risk.
      ,
      • Jeon P.
      • Limongi R.
      • Ford S.D.
      • Mackinley M.
      • Dempster K.
      • Théberge J.
      • Palaniyappan L.
      Progressive changes in glutamate concentration in early stages of schizophrenia: A longitudinal 7-Tesla MRS study.
      ,
      • Merritt K.
      • McGuire P.K.
      • Egerton A.
      • Aleman A.
      • Block W.
      • et al.
      1H-MRS in Schizophrenia Investigators
      Association of age, antipsychotic medication, and symptom severity in schizophrenia with proton magnetic resonance spectroscopy brain glutamate level: A mega-analysis of individual participant-level data.
      ,
      • Godlewska B.R.
      • Minichino A.
      • Emir U.
      • Angelescu I.
      • Lennox B.
      • Micunovic M.
      • et al.
      Brain glutamate concentration in men with early psychosis: A magnetic resonance spectroscopy case-control study at 7 T.
      ), or similar (
      • Aoyama N.
      • Théberge J.
      • Drost D.J.
      • Manchanda R.
      • Northcott S.
      • Neufeld R.W.J.
      • et al.
      Grey matter and social functioning correlates of glutamatergic metabolite loss in schizophrenia.
      ,
      • Chen T.
      • Wang Y.
      • Zhang J.
      • Wang Z.
      • Xu J.
      • Li Y.
      • et al.
      Abnormal concentration of GABA and glutamate in the prefrontal cortex in schizophrenia.-An in vivo 1H-MRS study.
      ,
      • Merritt K.
      • Perez-Iglesias R.
      • Sendt K.V.
      • Goozee R.
      • Jauhar S.
      • Pepper F.
      • et al.
      Remission from antipsychotic treatment in first episode psychosis related to longitudinal changes in brain glutamate.
      ,
      • Maximo J.O.
      • Briend F.
      • Armstrong W.P.
      • Kraguljac N.V.
      • Lahti A.C.
      Salience network glutamate and brain connectivity in medication-naïve first episode patients – A multimodal magnetic resonance spectroscopy and resting state functional connectivity MRI study.
      ) values. A recent meta-analysis yielded decreased Glx levels in medial PFC in patients compared with HCs at group levels (
      • Merritt K.
      • McGuire P.K.
      • Egerton A.
      • Aleman A.
      • Block W.
      • et al.
      1H-MRS in Schizophrenia Investigators
      Association of age, antipsychotic medication, and symptom severity in schizophrenia with proton magnetic resonance spectroscopy brain glutamate level: A mega-analysis of individual participant-level data.
      ).
      A few studies combined measures of 2 of the 3 transmitters listed above. Treatment resistance has been suggested to be associated with elevated glutamate levels but normal DSC (
      • Demjaha A.
      • Egerton A.
      • Murray R.M.
      • Kapur S.
      • Howes O.D.
      • Stone J.M.
      • McGuire P.K.
      Antipsychotic treatment resistance in schizophrenia associated with elevated glutamate levels but normal dopamine function.
      ). An inverse relationship between striatal dopamine activity and glutamate levels in the ACC was reported for first-episode patients (
      • Jauhar S.
      • Mccutcheon R.
      • Borgan F.
      • Veronese M.
      • Nour M.
      • Pepper F.
      • et al.
      The relationship between cortical glutamate and striatal dopamine in first-episode psychosis: a cross-sectional multimodal PET and magnetic resonance spectroscopy imaging study.
      ), while another study revealed a negative relationship between prefrontal glutamate levels and striatal DSC in healthy individuals (
      • Gleich T.
      • Deserno L.
      • Lorenz R.C.
      • Boehme R.
      • Pankow A.
      • Buchert R.
      • et al.
      Prefrontal and striatal glutamate differently relate to striatal dopamine: Potential regulatory mechanisms of striatal presynaptic dopamine function?.
      ). An association between striatal DSC and prefrontal GABA levels has not been reported.
      In a large cohort of antipsychotic-naïve patients partly overlapping this study population, we previously reported reduced GABA levels in the dorsal ACC, but no alterations of glutamatergic metabolites (
      • Bojesen K.B.
      • Ebdrup B.H.
      • Jessen K.
      • Sigvard A.
      • Tangmose K.
      • Edden R.A.E.
      • et al.
      Treatment response after 6 and 26 weeks is related to baseline glutamate and GABA levels in antipsychotic-naïve patients with psychosis.
      ,
      • Bojesen K.B.
      • Broberg B.V.
      • Fagerlund B.
      • Jessen K.
      • Thomas M.B.
      • Sigvard A.
      • et al.
      Associations between cognitive function and levels of glutamatergic metabolites and gamma-aminobutyric acid in antipsychotic-naïve patients with schizophrenia or psychosis.
      ). We also found increased glutamate levels in the thalamus in patients with a diagnosis of schizophrenia, the increase determined by subsequent nonresponders after 6- or 26-week antipsychotic treatments (
      • Bojesen K.B.
      • Ebdrup B.H.
      • Jessen K.
      • Sigvard A.
      • Tangmose K.
      • Edden R.A.E.
      • et al.
      Treatment response after 6 and 26 weeks is related to baseline glutamate and GABA levels in antipsychotic-naïve patients with psychosis.
      ,
      • Bojesen K.B.
      • Broberg B.V.
      • Fagerlund B.
      • Jessen K.
      • Thomas M.B.
      • Sigvard A.
      • et al.
      Associations between cognitive function and levels of glutamatergic metabolites and gamma-aminobutyric acid in antipsychotic-naïve patients with schizophrenia or psychosis.
      ).
      So far, no single brain abnormality has allowed discrimination of patients with psychosis from HCs. In the past decades, some authors instead suggested more complex models of schizophrenia involving combined transmitter disturbances in linked brain networks (
      • Alexander G.E.
      • DeLong M.R.
      • Strick P.L.
      Parallel organization of functionally segregated circuits linking basal ganglia and cortex.
      ,
      • Haber S.N.
      The place of dopamine in the cortico-basal ganglia circuit.
      ). The models have support from the preclinical literature, but clinical data are missing. Examination of multiple key neurotransmitters believed to be involved in the development of psychosis must happen at the earliest stage of disease, and before patients are exposed to antipsychotic treatment to ensure undiluted assessments. Uncompromised data may provide novel insight into the identification of combined neurochemical profiles that characterize first-episode psychosis.
      For this purpose, we argue that a multimodal rather than a unimodal approach may increase the accuracy of prediction of psychotic illness within the schizophrenia spectrum (
      • Modinos G.
      • Pettersson-yeo W.
      • Allen P.
      • Mcguire P.K.
      • Aleman A.
      • Mechelli A.
      Multivariate pattern classification reveals differential brain activation during emotional processing in individuals with psychosis proneness.
      ,
      • Soh P.
      • Narayanan B.
      • Khadka S.
      • Calhoun V.D.
      • Keshavan M.S.
      • Tamminga C.A.
      • et al.
      Joint coupling of awake EEG frequency activity and MRI gray matter volumes in the psychosis dimension: A BSNIP study.
      ,
      • Valli I.
      • Marquand A.F.
      • Mechelli A.
      • Raffin M.
      • Allen P.
      • Seal M.L.
      • McGuire P.
      Identifying individuals at high risk of psychosis: Predictive utility of support vector machine using structural and functional MRI data.
      ,
      • Ramyead A.
      • Studerus E.
      • Kometer M.
      • Uttinger M.
      • Gschwandtner U.
      • Fuhr P.
      • Riecher-Rössler A.
      Prediction of psychosis using neural oscillations and machine learning in neuroleptic-naïve at-risk patients.
      ,
      • Lei D.
      • Pinaya W.H.L.
      • Young J.
      • Van Amelsvoort T.
      • Marcelis M.
      • Donohoe G.
      • et al.
      Integrating machining learning and multimodal neuroimaging to detect schizophrenia at the level of the individual.
      ). As a completely agnostic approach may obscure signals and complicate interpretation because of inclusion of multiple variables (
      • Ebdrup B.H.
      • Axelsen M.C.
      • Bak N.
      • Fagerlund B.
      • Oranje B.
      • Raghava J.M.
      • et al.
      Accuracy of diagnostic classification algorithms using cognitive-, electrophysiological-, and neuroanatomical data in antipsychotic-naïve schizophrenia patients.
      ,
      • Ambrosen K.S.
      • Skjerbæk M.W.
      • Foldager J.
      • Axelsen M.C.
      • Bak N.
      • Arvastson L.
      • et al.
      A machine-learning framework for robust and reliable prediction of short- and long-term treatment response in initially antipsychotic-naïve schizophrenia patients based on multimodal neuropsychiatric data.
      ), we applied domain knowledge and carefully selected the variables a priori, guided by other studies and findings in previous overlapping cohorts where we were not able to identify group differences in DSC (
      • Sigvard A.K.
      • Nielsen M.Ø.
      • Gjedde A.
      • Bojesen K.B.
      • Fuglø D.
      • Tangmose K.
      • et al.
      Dopaminergic activity in antipsychotic-naïve patients assessed with positron emission tomography before and after partial dopamine D 2 receptor agonist treatment: Association with psychotic symptoms and treatment response.
      ) or institutional unit–scaled proton magnetic resonance spectroscopy (1H-MRS) measures (
      • Bojesen K.B.
      • Ebdrup B.H.
      • Jessen K.
      • Sigvard A.
      • Tangmose K.
      • Edden R.A.E.
      • et al.
      Treatment response after 6 and 26 weeks is related to baseline glutamate and GABA levels in antipsychotic-naïve patients with psychosis.
      ).
      We tested the hypothesis that a multivariable logistic regression model of 3 neurochemical measures, i.e., striatal dopamine synthesis rate (k3) or DSC (Ki4p) clearance, levels of GABA in the ACC, and levels of Glx in the thalamus or ACC, predicts group affiliation (antipsychotic-naïve patients with first-episode psychosis or HCs) more accurately than models of single neurochemical measures. We also evaluated the contribution of each of the neurotransmitters to the overall discriminatory power of the combinations applied.

      Methods and Materials

      Participants

      Participants were part of a large multimodal cohort study (PECANS II) approved by the Danish National Committee on Biomedical Research Ethics (H-3-2013-149). Detailed description of the cohort is provided in previous publications (
      • Sigvard A.K.
      • Nielsen M.Ø.
      • Gjedde A.
      • Bojesen K.B.
      • Fuglø D.
      • Tangmose K.
      • et al.
      Dopaminergic activity in antipsychotic-naïve patients assessed with positron emission tomography before and after partial dopamine D 2 receptor agonist treatment: Association with psychotic symptoms and treatment response.
      ,
      • Bojesen K.B.
      • Ebdrup B.H.
      • Jessen K.
      • Sigvard A.
      • Tangmose K.
      • Edden R.A.E.
      • et al.
      Treatment response after 6 and 26 weeks is related to baseline glutamate and GABA levels in antipsychotic-naïve patients with psychosis.
      ,
      • Bojesen K.B.
      • Broberg B.V.
      • Fagerlund B.
      • Jessen K.
      • Thomas M.B.
      • Sigvard A.
      • et al.
      Associations between cognitive function and levels of glutamatergic metabolites and gamma-aminobutyric acid in antipsychotic-naïve patients with schizophrenia or psychosis.
      ). Patients were referred from Mental Health Centers in the Capital Region of Denmark and could be included if they were lifetime naïve to antipsychotic compounds and central nervous system stimulants; legally competent; 18 to 45 years of age; and fulfilled the diagnostic criteria for schizophrenia, schizoaffective disorder, or nonorganic psychosis according to the ICD-10. The diagnosis was evaluated by certified interviewers with the Schedules for Clinical Assessment in Neuropsychiatry (
      • Wing J.K.
      • Babor T.
      • Brugha T.
      • Burke J.
      • Cooper J.E.
      • Giel R.
      • et al.
      SCAN. Schedules for clinical assessment in neuropsychiatry.
      ).
      HCs matched on age, sex, and parental educational level were recruited via online advertisement. Exclusion criteria included lifetime psychiatric illness or ultrahigh risk of psychosis according to the Comprehensive Assessment of At-Risk Mental States (
      • Yung A.R.
      • Phillips L.J.
      • McGorry P.D.
      • McFarlane C.A.
      • Francey S.
      • Harrigan S.
      • et al.
      Prediction of psychosis. A step towards indicated prevention of schizophrenia.
      ) and having first-degree relatives with psychotic symptoms.
      General exclusion criteria included head injury with more than 5 minutes’ unconsciousness, metallic implants (incompatible with magnetic resonance imaging [MRI]), pregnancy, severe physical illness or previous substance abuse (ICD-10 criteria F1X.1), or use of antidepressant treatment within the past 30 days. Patients submitted to involuntary treatment or admission were excluded. Prescribed benzodiazepines were tolerated in patients prior to initiation of antipsychotic treatment. Two patients had a positive urine screening for benzodiazepines on the neuroimaging days and were excluded from this sample primarily because of the evidence of benzodiazepines’ effect on the GABAergic system (
      • Tan K.R.
      • Rudolph U.
      • Lüscher C.
      Hooked on benzodiazepines: GABA A receptor subtypes and addiction.
      ). Occasional substance use was accepted and self-reported and verified by a urine drug test (Rapid Response; Jepsen HealthCare) before neuroimaging.

      18F-FDOPA PET

      We determined dopamine synthesis (k3) and DSC (Ki4p) with 18F-FDOPA PET using integrated PET–computed tomography (Siemens Biograph m CT64, 2013). To minimize 18F-FDOPA metabolic degradation before passage through the blood-brain barrier, we administered carbidopa (150 mg) and entacapone (400 mg) orally 1 hour prior to PET. We acquired an initial low-dose computed tomography scan before all PET sessions to enable attenuation correction. We injected 18F-FDOPA as an intravenous bolus (mean dose: 330 MBq, SD: 33.0; range: 259–399 MBq) simultaneously with initiation of the first session.
      Detailed information about settings, collection of arterial blood samples, equipment, and data processing including kinetic modeling was described previously for the same cohort (
      • Sigvard A.K.
      • Nielsen M.Ø.
      • Gjedde A.
      • Bojesen K.B.
      • Fuglø D.
      • Tangmose K.
      • et al.
      Dopaminergic activity in antipsychotic-naïve patients assessed with positron emission tomography before and after partial dopamine D 2 receptor agonist treatment: Association with psychotic symptoms and treatment response.
      ) and in the associated Supplement.

      MRI Data Acquisition

      We acquired MRI scans at 3T (Achieva; Philips Healthcare) with a 32-channel head coil (Invivo). We instructed participants to not move during the scan and to keep the head in the same position. High-resolution, three-dimensional, structural T1-weighted images (response time 10 ms, echo time 4.6 ms, flip angle = 8°, voxel size = 0.79 × 0.79 × 0.80 mm) were obtained of the brain and used as reference for the tomography as well as for placement of the spectroscopic voxels and segmentation into cerebrospinal fluid and gray matter and white matter fractions.
      We used FreeSurfer version 5.3.0 (
      • Dale A.M.
      • Fischl B.
      • Sereno M.I.
      Cortical surface-based analysis. I. Segmentation and surface reconstruction.
      ,
      • Fischl B.
      • Sereno M.I.
      • Dale A.M.
      Cortical surface-based analysis. II: Inflation, flattening, and a surface-based coordinate system.
      ) for individual anatomical segmentation of participants’ brains and to obtain regions of interest for the nucleus accumbens (NAcc), putamen, and nucleus caudatus. We selected the whole striatum (NAcc, putamen, and nucleus caudatus) and the NAcc as regions of interest. We included the NAcc on the basis of results from preclinical research (
      • Flagstad P.
      • Mørk A.
      • Glenthøj B.Y.
      • van Beek J.
      • Michael-Titus A.T.
      • Didriksen M.
      Disruption of neurogenesis on gestational Day 17 in the rat causes behavioral changes relevant to positive and negative schizophrenia symptoms and alters amphetamine-induced dopamine release in nucleus accumbens.
      ). The NAcc has been tested as a potential therapeutic target of neurosurgical intervention for schizophrenia (
      • Mikell C.B.
      • Mckhann G.M.
      • Segal S.
      • Mcgovern R.A.
      • Wallenstein M.B.
      • Moore H.
      The hippocampus and nucleus accumbens as potential therapeutic targets for neurosurgical intervention in schizophrenia.
      ), and it was further found to be of interest to the understanding of psychosis and dopaminergic disturbances in clinical studies (
      • Mcgowan S.
      • Lawrence A.D.
      • Sales T.
      • Quested D.
      • Grasby P.
      Presynaptic dopaminergic dysfunction in schizophrenia: A positron emission tomographic [18F]fluorodopa study.
      ,
      • Nielsen M.O.
      • Rostrup E.
      • Wulff S.
      • Bak N.
      • Broberg B.V.
      • Lublin H.
      • et al.
      Improvement of brain reward abnormalities by antipsychotic monotherapy in schizophrenia.
      ,
      • Wulff S.
      • Nielsen M.Ø.
      • Rostrup E.
      • Svarer C.
      • Jensen L.T.
      • Pinborg L.
      • Glenthøj B.Y.
      The relation between dopamine D2 receptor blockade and the brain reward system: A longitudinal study of first-episode schizophrenia patients.
      ). We found this relationship also in an overlapping sample of subjects in whom we found that values of both k3 and Ki4p from the NAcc correlated with the severity of positive symptoms in the patients before the first treatment (
      • Sigvard A.K.
      • Nielsen M.Ø.
      • Gjedde A.
      • Bojesen K.B.
      • Fuglø D.
      • Tangmose K.
      • et al.
      Dopaminergic activity in antipsychotic-naïve patients assessed with positron emission tomography before and after partial dopamine D 2 receptor agonist treatment: Association with psychotic symptoms and treatment response.
      ).

      Proton MRS

      We completed 1H-MRS with the point-resolved spectroscopy (PRESS) sequence to determine Glx levels and with the Mescher-Garwood PRESS (MEGAPRESS) sequence to obtain GABA levels. We measured Glx using rectangular voxels in the dorsal ACC and left thalamus, while we measured GABA levels in the dorsal ACC only [see the Supplement; (
      • Bojesen K.B.
      • Broberg B.V.
      • Fagerlund B.
      • Jessen K.
      • Thomas M.B.
      • Sigvard A.
      • et al.
      Associations between cognitive function and levels of glutamatergic metabolites and gamma-aminobutyric acid in antipsychotic-naïve patients with schizophrenia or psychosis.
      )]. We used the LCModel version 6.3-1L (
      • Provencher S.
      LCModel & LCMgui User’s Manual, Version 6.3-1L.
      ) for analysis of PRESS data and Gannet version 3.1 (
      • Edden R.
      GABA Analysis toolkit.
      ) for MEGAPRESS data, as described in the Supplement. The in vivo water-scaled values of metabolites were corrected for partial volume contamination by cerebrospinal fluid to obtain concentrations in institutional units, as previously described (
      • Bojesen K.B.
      • Broberg B.V.
      • Fagerlund B.
      • Jessen K.
      • Thomas M.B.
      • Sigvard A.
      • et al.
      Associations between cognitive function and levels of glutamatergic metabolites and gamma-aminobutyric acid in antipsychotic-naïve patients with schizophrenia or psychosis.
      ). There was significantly higher full width at half maximum values of patients for both dorsal ACC PRESS and MEGAPRESS acquisitions (Tables S1 and S3), but inclusion of full width at half maximum as covariate in the statistical analyses did not affect the results, as described in the Supplement.

      Statistical Analysis

      We used binomial logistic regression to know whether striatal dopamine synthesis (k3) or DSC (Ki4p), GABA levels in the ACC, or Glx levels in the thalamus or ACC would predict group. First, we completed 7 simple logistic regressions with each of the 7 variables to predict group as independent variables. In the following, we denote these univariable models as individual neurotransmitter models. Second, we completed multiple logistic regressions combining either Ki4p or k3 (from the striatum/NAcc) with GABA levels (ACC) and Glx levels (thalamus/ACC) to predict group. The limited sample size required a priori selection of neurotransmitter combinations rather than combinations of all parameters in one model. With the following premises, all combinations represented all 3 neurotransmitters (dopamine, Glx, and GABA), and none of the included models could show multicollinearity. The latter premise was fulfilled when we constrained the neurotransmitter models to include either a Ki4p or k3 measure, resulting in a total of 8 different neurotransmitter models. In the following, we called these multivariable models neurotransmitter combinations, ordered 1–8. The a priori selection of regions of interest with regard to the dopamine measures was based on previous preclinical findings and our own results from an overlapping sample (
      • Sigvard A.K.
      • Nielsen M.Ø.
      • Gjedde A.
      • Bojesen K.B.
      • Fuglø D.
      • Tangmose K.
      • et al.
      Dopaminergic activity in antipsychotic-naïve patients assessed with positron emission tomography before and after partial dopamine D 2 receptor agonist treatment: Association with psychotic symptoms and treatment response.
      ). See the Supplement for a discussion on the striatal locus of dopaminergic disturbances.
      The interaction terms Ki4p × GABA or k3 × GABA were included in each of the neurotransmitter models because they contributed significantly to some of the models. Interaction terms combining the variables Ki4p or k3 with Glx levels, or with GABA and Glx levels combined, did not add significance and were excluded. We standardized all predictors (mean = 0, SD = 1) prior to fitting of the models to increase the comparability of predictors. HCs were used as reference population. Hence, negative β coefficients indicate a tendency toward lower values for patients than for HCs.
      We assessed the assumption of linearity of the continuous variables with respect to the logit of the dependent variable via the Box-Tidwell procedure (
      • Box G.E.P.
      • Tidwell P.W.
      Transformation of the independent variables.
      ). Exploration of outliers, high leverage points, and high influential points led us to exclude 2 HCs as outliers with respect to k3. These 2 outliers were excluded from all analyses.
      We used the Akaike information criterion (AIC) to find the relative quality of the statistical models for given sets of data. Lower AIC reflects a better model that balanced goodness of fit and model complexity. Here, we used the model selection corrected for small sample sizes (AICc) to distinguish among the 8 neurotransmitter combinations based on evidence ratios for AICc weights for the best-fitted and the next best-fitted models (
      • Wagenmakers E.J.
      • Farrell S.
      AIC model selection using Akaike weights.
      ).
      As primary outcomes, accuracies represented the ability of a given individual neurotransmitter model to predict group. We used 10,000 permutations to test for significant differences from baseline accuracy. Baseline accuracy defines the accuracy of predicting the majority class of each observation (
      • Phipson B.
      • Smyth G.K.
      Permutation P-values should never be zero: calculating exact P-values when permutations are randomly drawn.
      ), i.e., the more unequal the sample sizes of patients and HCs, the higher the baseline accuracy. We also provided the area under the curve (AUC) of the receiver operating characteristic curves. The receiver operating characteristic curve indicates the overall discriminatory ability of the neurotransmitter combination. For the interpretation of AUC, we used the following classifications: 0.7 ≤ AUC < 0.8 indicates acceptable discrimination, and 0.8 ≤ AUC < 0.9 indicates excellent discrimination (
      • Hosmer Jr., D.W.
      • Lemeshow S.
      • Sturdivant R.X.
      Applied Logistic Regression.
      ). We applied the Benjamini-Hochberg procedure (BH) (
      • Benjamini Y.
      • Hochberg Y.
      Controlling the false discovery rate: A practical and powerful approach to multiple testing.
      ) to correct for multiple comparisons covering 7 univariable and 8 multivariable analyses. In the Results section, we denoted BH-corrected p values as pBH.
      We further ran cross-validation with 100 random splits stratified on group (70% training, 30% test) to estimate the generalizability of the models to an independent sample, and we tested significance with a test of 10,000 permutations.
      We used Python 3.7 incorporating the statsmodels (
      • Seabold S.
      • Perktold J.
      Statsmodels: Econometric and statistical modeling with Python.
      ) and scikit-learn (
      • Pedregosa F.
      • Varoquaux G.
      • Gramfort A.
      • Michel V.
      • Thirion B.
      • Grisel O.
      • et al.
      Scikit-learn: Machine learning in Python.
      ) modules for binominal logistic regression including cross-validation, and we completed demographic exploration and assumption testing using SPSS Statistics version 25 (IBM Corp.).

      Results

      We obtained combined measures of the variables Ki4p-estimates or k3-variables, GABA levels in the ACC, and Glx levels in the thalamus of 23 patients and 20 HCs (Table 1) and included a few more members of the group of HCs in the samples of dopamine measures combined with Glx levels in the ACC (Table S4). The individual neurotransmitter models had larger sample sizes (see the Supplement and Table S4). The patients were moderately ill, with a mean Positive and Negative Syndrome Scale total score of 80. Fourteen patients (61%) were diagnosed with schizophrenia (Table 1).
      Table 1Demographics and Clinical Characteristics of Participants With 18F-FDOPA PET and Spectroscopy Including GABA Spectra in the ACC and Glx in the Thalamus, n = 43
      CharacteristicsPatients, n = 23HCs, n = 20Statistics
      Age, Years, Mean ± SD22.30 ± 3.922.35 ± 4.3t38.535 = −0.036, p = .971
      Sex, Female/Male, n14/912/8Fisher’s p = 1.000
      Educational Level, Years, Mean/Median12.513
      Normal distribution violated; median reported instead of mean. The significance level is set to .05.
      Mann-Whitney p = .233
      Parental Educational Level, High/Moderate/Low, n6/12/58/11/1χ2 = 0.003, p = .954
      Race, Caucasian/Other, n19/419/1Fisher’s p = .351
      Current Substance Use
       Smoking, yes/no, n8/152/18Fisher’s p = .076
       Cannabis, positive urine screening, yes/no, n0/231/19Fisher’s p = .465
      Diagnosis ICD-10, n (%)
       Schizophrenia14 (61%)
       Nonorganic psychosis9 (39%)
      PANSS Score, Baseline, Mean
       Total79.87
       Positive19.65
       Negative21.83
       General38.39
      18F-FDOPA, 18F-fluorodopa; ACC, anterior cingulate cortex; GABA, gamma-aminobutyric acid; Glx, glutamate plus glutamine; HCs, healthy control subjects; PANSS, Positive and Negative Syndrome Scale; PET, positron emission tomography.
      a Normal distribution violated; median reported instead of mean. The significance level is set to .05.
      None of the 7 individual neurotransmitter models had significant predictive performance (Table S5). Of the 8 combined neurotransmitter models, the best-fitted model included Ki4p in the NAcc, GABA in the ACC, Glx in the thalamus, and the interaction term Ki4p × GABA (combination 1). This model carried 93.5% of the cumulative model weight (AIC 48.475, AIC weight 0.935). In comparison, the next best combined model, combination 3, only carried 2.2% of the cumulative model weight. Hence, the evidence ratio shows that combination 1 is 42.5 times (0.935/0.022) more likely to be superior to the next best model. Table 2 displays the variables included in each of the 8 neurotransmitter models and their accuracies, AUC, AICc, and AICc weight.
      Table 2Predictive Values for the 8 Neurotransmitter Combinations
      Neurotransmitter CombinationAccuracyAccuracy p ValueAccuracy p Value BHAUCAUC p ValueAUC p Value BHAICcAICc Weight
      Ki4p NAcc Combined With
       1GABAACC and GlxThalamus0.837.000
      p < .05.
      .003
      p < .05.
      0.838.000
      p < .05.
      .003
      p < .05.
      48.4750.935
       2GABAACC and GlxACC0.778.002
      p < .05.
      .012
      p < .05.
      0.778.003
      p < .05.
      .015
      p < .05.
      55.9680.022
      Ki4p whole str Combined With
       3GABAACC and GlxThalamus0.791.001
      p < .05.
      .010
      p < .05.
      0.791.001
      p < .05.
      .010
      p < .05.
      55.9470.022
       4GABAACC and GlxACC0.711.038
      p < .05.
      .1150.709.068.20361.8440.001
      k3 NAcc Combined With
       5GABAACC and GlxThalamus0.721.031
      p < .05.
      .1150.713.054.20356.5650.016
       6GABAACC and GlxACC0.667.140.3000.666.174.34359.9660.003
      k3 whole str Combined With
       7GABAACC and GlxThalamus0.651.222.3450.648.259.38964.280<0.001
       8GABAACC and GlxACC0.644.230.3450.644.260.38967.634<0.001
      k3 is the enzymatic activity of DOPA-decarboxylase, converting DOPA to dopamine. Ki4p is the dopamine synthesis capacity estimate from the 4-parameter model. The whole striatum is composed of putamen, caudate, and nucleus accumbens. AICc weight is the proportion of the total amount of predictive power provided by the full set of models contained in the model being assessed. The BH procedure was used to correct for 15 comparisons (7 univariable and 8 multivariable).
      ACC, anterior cingulate cortex; AICc, Akaike information criterion controlled for small sample sizes; AUC, area under the receiver operating characteristic curve; BH, Benjamini-Hochberg; GABA, gamma-aminobutyric acid; Glx, glutamate plus glutamine; NAcc, nucleus accumbens; str, striatum.
      a p < .05.
      Combination 1 predicted group with an accuracy of 83.7% (pBH = .003) and an AUC of 83.8% (pBH = .003) (Table 2 and Figure 1). In combination 1, the values of GABA in the ACC (p = .019) and of the interaction term Ki4p × GABA (p = .016) were predictive, while Ki4p in the NAcc (p = .664) and Glx in the thalamus (p = .678) were not. GABA levels tended to be lower in patients than in HCs (negative β coefficients). β Coefficients and contribution of each of the independent variables to the neurotransmitter models (combination 1–8) are provided in Table 3.
      Figure thumbnail gr1
      Figure 1This graph shows the accuracies for prediction of group in univariable and multivariable models, illustrating the predictive accuracy of the tested models. None of the individual neurotransmitter models (red bars) could predict group. Among the multivariable models (blue bars), the best fit model, combination 1, carried out 93.5% of the cumulative model weight (Akaike information criterion controlled for small sample sizes [AICc] = 48.5 and AICc weight = 0.935) and differed from the rest of the models. The variables included in each of the 8 models are provided in . The dotted green lines indicate balanced accuracy. ∗ indicates significant predictions (pBH < .05). k3 is the decarboxylation rate (conversion of 18F-FDOPA to 18F-dopamine). Ki is the estimate of the dopamine synthesis capacity from the 4-parameter model. ACC, anterior cingulate cortex; 18F-FDOPA, 18F-fluorodopa; GABA, gamma-aminobutyric acid; Glx, glutamate plus glutamine; NAcc, nucleus accumbens; Str, striatum; Thal, thalamus.
      Table 3The Contribution of the Independent Variables in Each of the 8 Neurotransmitter Combinations
      Neurotransmitter CombinationsβSEp > |z|CI 0.025, 0.975
      1Ki4p NAcc−0.3010.693.664−1.658, 1.057
      GABAACC−1.9190.816.019
      p < .05.
      −3.519, −0.320
      GlxThalamus−0.1940.466.678−1.108, 0.720
      Interaction, Ki × GABA3.5701.479.016
      p < .05.
      0.672, 6.468
      2Ki4p NAcc0.1910.542.725−0.871, 1.252
      GABAACC−1.3430.636.035
      p < .05.
      −2.589, −0.097
      GlxACC−0.1310.342.701−0.801, 0.538
      Interaction, Ki × GABA2.4431.078.024
      p < .05.
      0.330, 4.556
      3Ki4p whole striatum−0.0990.457.828−0.994, 0.796
      GABAACC−1.1970.546.028
      p < .05.
      −2.267, −0.127
      GlxThalamus−0.2600.426.542−1.096, 0.575
      Interaction, Ki × GABA1.6190.717.024
      p < .05.
      0.214, 3.023
      4Ki4p whole striatum0.1130.402.779−0.674, 0.900
      GABAACC−0.9860.482.041
      p < .05.
      −1.931,−0.041
      GlxACC−0.1340.330.685−0.781, 0.513
      Interaction, Ki × GABA1.2590.607.038
      p < .05.
      0.069, 2.449
      5k3 NAcc0.1420.404.726−0.650, 0.933
      GABAACC−1.3430.558.016
      p < .05.
      −2.436, −0.251
      GlxThalamus0.0080.419.985−0.814, 0.829
      Interaction, k3 × GABA1.1020.476.021
      p < .05.
      0.168, 2.035
      6k3 NAcc0.0890.373.8110.642, 0.820
      GABAACC−1.2030.531.024
      p < .05.
      −2.244, −0.162
      GlxACC−0.3400.359.344−1.044, 0.364
      Interaction, k3 × GABA1.0470.448.019
      p < .05.
      0.169, 1.924
      7k3 whole striatum0.0290.357.934−0.670, 0.728
      GABAACC−0.8640.398.030
      p < .05.
      −1.643, −0.084
      GlxThalamus−0.0860.357.809−0.785, 0.613
      Interaction, k3 × GABA0.2800.368.446−0.440, 1.001
      8k3 whole striatum−0.0590.346.866−0.737, 0.620
      GABAACC−0.7540.383.049
      p < .05.
      −1.504, −0.004
      GlxACC−0.1750.323.588−0.809, 0.459
      Interaction, k3 × GABA0.3170.345.357−0.358, 0.993
      The interaction terms Ki4p × GABA or k3 × GABA were included in models because the term Ki4p × GABA added significantly to some models. Other interaction terms were insignificant and hence left out. k3 is the enzymatic activity of DOPA-decarboxylase, converting DOPA to dopamine. Ki4p is the dopamine synthesis capacity estimate from the 4-parameter model.
      ACC, anterior cingulate cortex; GABA, gamma-aminobutyric acid; Glx, glutamate plus glutamine; NAcc, nucleus accumbens.
      a p < .05.
      We list the estimated predictive performances of an independent sample of the 8 models after cross-validation in Table S6. Combination 1 significantly predicted group (accuracy 74.1%, pBH = .021, AUC 82.8%, pBH = .007). Combinations 2 and 3 were significant at trend level (p = .064 and .043, respectively).
      We completed post hoc analyses, excluding the consistently insignificant independent variable Glx (in the thalamus/ACC) to reduce model complexity and to explore whether combined dopamine measures and GABA levels in the ACC (double neurotransmitter models) would be comparable to triple neurotransmitter models. However, the double transmitter models were inferior to triple neurotransmitter models (Supplement and Table S7).

      Discussion

      In this study, we found that a combination of striatal dopaminergic metabolism and levels of GABA and glutamatergic metabolites discriminated antipsychotic-naïve patients from HCs, unlike the individual neurotransmitter measures that failed to do so. Specifically, we discovered that 3 neurotransmitter models (1, 2, and 3) resulted in satisfactory prediction of group. The combination of DSC in the NAcc, GABA levels in the ACC, and Glx levels in the thalamus was markedly superior to other combinations with respect to AIC weight and predicted the status of having a first-episode psychosis with an accuracy of 83.7%. In this model, GABA levels and the DSC-GABA interaction made significant contributions, unlike DSC and the Glx levels. However, all triple transmitter models, including Glx levels in the thalamus or ACC, had higher accuracy than the corresponding double transmitter models, excluding the Glx levels. Triple transmitter models additionally were a better fit in terms of AIC, although AIC penalize more complex models. Hence, overall, we found that all 3 neurotransmitters contributed to variable extents to the discrimination between patients and HCs.
      DSC was not in itself a predictive factor. A large body of evidence suggests that disturbances in DSC are involved in the development of psychotic symptoms (
      • Reith J.
      • Benkelfat C.
      • Sherwin A.
      • Yasuhara Y.
      • Kuwabara H.
      • Andermann F.
      • et al.
      Elevated dopa decarboxylase activity in living brain of patients with psychosis.
      ,
      • Lindström L.H.
      • Gefvert O.
      • Hagberg G.
      • Lundberg T.
      • Bergström M.
      • Hartvig P.
      • Långström B.
      Increased dopamine synthesis rate in medial prefrontal cortex and striatum in schizophrenia indicated by L-(beta-11C) DOPA and PET.
      ,
      • Hietala J.
      • Syvälahti E.
      • Vilkman H.
      • Vuorio K.
      • Räkköläinen V.
      • Bergman J.
      • et al.
      Depressive symptoms and presynaptic dopamine function in neuroleptic-naive schizophrenia.
      ,
      • Mcgowan S.
      • Lawrence A.D.
      • Sales T.
      • Quested D.
      • Grasby P.
      Presynaptic dopaminergic dysfunction in schizophrenia: A positron emission tomographic [18F]fluorodopa study.
      ,
      • Hietala J.
      • Syvälahti E.
      • Vuorio K.
      • Räkköläinen V.
      • Bergman J.
      • Haaparanta M.
      • et al.
      Presynaptic dopamine function in striatum of neuroleptic-naive schizophrenic patients.
      ,
      • Kumakura Y.
      • Cumming P.
      • Vernaleken I.
      • Buchholz H.G.
      • Siessmeier T.
      • Heinz A.
      • et al.
      Elevated [18F]fluorodopamine turnover in brain of patients with schizophrenia: An [18F]fluorodopa/positron emission tomography study.
      ,
      • Nielsen M.O.
      • Rostrup E.
      • Wulff S.
      • Bak N.
      • Broberg B.V.
      • Lublin H.
      • et al.
      Improvement of brain reward abnormalities by antipsychotic monotherapy in schizophrenia.
      ,
      • Wulff S.
      • Nielsen M.Ø.
      • Rostrup E.
      • Svarer C.
      • Jensen L.T.
      • Pinborg L.
      • Glenthøj B.Y.
      The relation between dopamine D2 receptor blockade and the brain reward system: A longitudinal study of first-episode schizophrenia patients.
      ,
      • Fusar-Poli P.
      • Meyer-Lindenberg A.
      Striatal presynaptic dopamine in schizophrenia, part II: Meta-analysis of [(18)F/(11)C]-DOPA PET studies.
      ,
      • Veronese M.
      • Santangelo B.
      • Jauhar S.
      • D’Ambrosio E.
      • Demjaha A.
      • Salimbeni H.
      • et al.
      A potential biomarker for treatment strati fi cation in psychosis: Evaluation of an [ 18 F ] FDOPA PET imaging approach.
      ,
      • Jauhar S.
      • Nour M.M.
      • Veronese M.
      • Rogdaki M.
      • Bonoldi I.
      • Azis M.
      • et al.
      A test of the transdiagnostic dopamine hypothesis of psychosis using positron emission tomographic imaging in bipolar affective disorder and schizophrenia.
      ,
      • Jauhar S.
      • Veronese M.
      • Nour M.M.
      • Rogdaki M.
      • Hathway P.
      • Turkheimer F.E.
      • et al.
      Determinants of treatment response in first-episode psychosis: An 18F-DOPA PET study.
      ), but the lack of predictive discriminative power of DSC in the present data is in agreement with the 2 so far largest first-episode studies (
      • Jauhar S.
      • Mccutcheon R.
      • Borgan F.
      • Veronese M.
      • Nour M.
      • Pepper F.
      • et al.
      The relationship between cortical glutamate and striatal dopamine in first-episode psychosis: a cross-sectional multimodal PET and magnetic resonance spectroscopy imaging study.
      ,
      • Sigvard A.K.
      • Nielsen M.Ø.
      • Gjedde A.
      • Bojesen K.B.
      • Fuglø D.
      • Tangmose K.
      • et al.
      Dopaminergic activity in antipsychotic-naïve patients assessed with positron emission tomography before and after partial dopamine D 2 receptor agonist treatment: Association with psychotic symptoms and treatment response.
      ) failing to confirm differences in DSC between patients and HCs.
      The interaction between GABA levels in the ACC and Ki4p values in the striatum added with the largest β coefficient to the prediction of subject group. This intriguing finding supports the claim that patients may have a disturbed relationship between the 2 transmitters. The so-called filter hypothesis for schizophrenia proposed that schizophrenia develops as a result of disturbances in interacting dopaminergic, glutamatergic, and GABAergic systems in the cortico-striato-thalamo-cortical loops, leading to disruption of the thalamic filter and an overload of information to the cortex (
      • Carlsson A.
      The current status of the dopamine hypothesis of schizophrenia.
      ). As an extension, Carlsson suggested that the balance between activation and inhibition in the direct and indirect pathways in the macrocircuits plays a key role for the level of subcortical dopamine and for the understanding of how psychotic symptoms evolve (
      • Carlsson A.
      The neurochemical circuitry of schizophrenia.
      ). The direct pathway results in activation of subcortical structures via glutamatergic excitation from cortical neurons, whereas the indirect pathway, involving an odd number of GABAergic interneurons, counterbalances the activation. Hence, if the direct pathway dominates, the net result is disinhibition of the thalamic filter, whereas domination of the indirect pathway results in net inhibition of the cortico-striato-thalamo-cortical loop (
      • Howes O.D.
      • Kapur S.
      The dopamine hypothesis of schizophrenia: version III--the final common pathway.
      ,
      • Carlsson A.
      The neurochemical circuitry of schizophrenia.
      ,
      • Geday J.
      • Østergaard K.
      • Johnsen E.
      • Gjedde A.
      STN-stimulation in Parkinson ’ s disease restores striatal inhibition of thalamocortical projection.
      ). Cortical GABAergic neurons have also been held to exert important effects on the balance between excitatory and inhibitory states by regulating the connectivity of GABA interneurons (
      • Fazzari P.
      • Paternain A.V.
      • Valiente M.
      • Pla R.
      • Luján R.
      • Lloyd K.
      • et al.
      Control of cortical GABA circuitry development by Nrg1 and ErbB4 signalling.
      ,
      • Gonzalez-Burgos G.
      • Fish K.N.
      • Lewis D.A.
      GABA neuron alterations, cortical circuit dysfunction and cognitive deficits in schizophrenia.
      ). The higher β coefficient for the interaction term between DSC and GABA levels than the β coefficients for the individual neurotransmitters indicates that the contribution of the interrelation exceeds the one from each neurotransmitter. See the Supplement for detailed discussion of the impact of the interaction term.
      A recent study found a negative correlation between gray matter volume in the PFC and striatal DSC in treatment-responsive patients with first-episode psychosis (
      • D’Ambrosio E.D.
      • Jauhar S.
      • Kim S.
      • Veronese M.
      • Rogdaki M.
      • Pepper F.
      • et al.
      The relationship between grey matter volume and striatal dopamine function in psychosis: A multimodal 18 F-DOPA PET and voxel-based morphometry study.
      ). The authors suggested that lower PFC volume reflects fewer or disrupted inhibitory projections from the PFC (e.g., a disrupted indirect pathway), potentially leading to downstream disinhibition of striatal dopamine function and elevated DSC (
      • D’Ambrosio E.D.
      • Jauhar S.
      • Kim S.
      • Veronese M.
      • Rogdaki M.
      • Pepper F.
      • et al.
      The relationship between grey matter volume and striatal dopamine function in psychosis: A multimodal 18 F-DOPA PET and voxel-based morphometry study.
      ). Here, GABA levels in the ACC may represent a more direct measure of the prefrontal regulation of inhibitory projections to the striatum. The impact of the interaction between GABA levels in the ACC and striatal DSC in the present multimodal analyses suggests that the diagnosis of psychosis rather depends on disruption of the indirect pathway in the macrocircuits in the brain (
      • Carlsson A.
      The neurochemical circuitry of schizophrenia.
      ,
      • Carlsson A.
      • Waters N.
      • Carlsson M.L.
      Neurotransmitter interactions in schizophrenia–therapeutic implications.
      ) than on dopaminergic or GABAergic abnormalities in themselves, although such a disruption, if present, did not lead to elevated DSC in this study.
      We also showed GABA levels in the ACC to be an important predictor of group. A discrimination based on GABA levels agrees with a previous finding of lower baseline GABA levels in antipsychotic-naïve patients (
      • Bojesen K.B.
      • Broberg B.V.
      • Fagerlund B.
      • Jessen K.
      • Thomas M.B.
      • Sigvard A.
      • et al.
      Associations between cognitive function and levels of glutamatergic metabolites and gamma-aminobutyric acid in antipsychotic-naïve patients with schizophrenia or psychosis.
      ). The lower baseline levels appeared to be driven by nonresponding patients compared with HCs in an overlapping sample of patients (
      • Bojesen K.B.
      • Ebdrup B.H.
      • Jessen K.
      • Sigvard A.
      • Tangmose K.
      • Edden R.A.E.
      • et al.
      Treatment response after 6 and 26 weeks is related to baseline glutamate and GABA levels in antipsychotic-naïve patients with psychosis.
      ). The results also agree with the data from a recent 7T 1H-MRI meta-analysis showing a tendency toward lower GABA levels in patients, although the summary effect size was nonsignificant (
      • Sydnor V.J.
      • Roalf D.R.
      A meta-analysis of ultra-high fi eld glutamate, glutamine, GABA and glutathione 1HMRS in psychosis: Implications for studies of psychosis risk.
      ). In agreement, we consistently found negative β coefficients, indicating lower GABA levels of patients.
      The higher accuracies and lower AICc values of the triple transmitter models (Table 2) compared with the double transmitter models (Supplement and Table S7) suggest that Glx levels contributed to the significance of the overall models. Nevertheless, Glx measures did not serve as a predictor in any of the triple neurotransmitter models. The lack of predictive power implies that potential disturbances of Glx levels are minor in the thalamus and ACC in patients at this stage of psychosis. In agreement with this, a study of an overlapping sample (
      • Bojesen K.B.
      • Broberg B.V.
      • Fagerlund B.
      • Jessen K.
      • Thomas M.B.
      • Sigvard A.
      • et al.
      Associations between cognitive function and levels of glutamatergic metabolites and gamma-aminobutyric acid in antipsychotic-naïve patients with schizophrenia or psychosis.
      ) yielded no initial group difference of Glx levels in the thalamus or ACC. However, patients with a diagnosis of schizophrenia had higher glutamate levels in the thalamus than did the HCs (
      • Bojesen K.B.
      • Broberg B.V.
      • Fagerlund B.
      • Jessen K.
      • Thomas M.B.
      • Sigvard A.
      • et al.
      Associations between cognitive function and levels of glutamatergic metabolites and gamma-aminobutyric acid in antipsychotic-naïve patients with schizophrenia or psychosis.
      ), and as was the case for the abnormal GABA levels in the ACC, with the abnormality seemed to be driven by nonresponders to subsequent treatment (
      • Bojesen K.B.
      • Ebdrup B.H.
      • Jessen K.
      • Sigvard A.
      • Tangmose K.
      • Edden R.A.E.
      • et al.
      Treatment response after 6 and 26 weeks is related to baseline glutamate and GABA levels in antipsychotic-naïve patients with psychosis.
      ).
      Combinations 2 and 3 are discussed in the Supplement.
      Taken together, these findings not only support the importance of dysfunction in specific neurotransmitters and regions in the aforementioned neural circuits for the development of psychosis, but also demonstrate that it is possible to predict patient status based on these. The consequent improvement in diagnostic validity might also, as argued by Tamminga (
      • Tamminga C.A.
      Assessing striatal dopamine in schizophrenia.
      ), add to improvements in the precision and usefulness of (other) biomarker discoveries, particularly when the diagnostic validity is based on neurobiological findings.
      To our knowledge, this study is the first to include both dopamine synthesis measures and GABA and glutamate levels in the same patients and HCs. The strengths of the findings are the presence of strictly antipsychotic-naïve patients in a relatively large sample free of substance abuse and antidepressant medication and further free from acute effects of benzodiazepines. Patients and HCs endured long neuroimaging sessions and completed an extensive examination program. This means that the participants may fail to be representative of the typical patient experiencing psychosis. However, the present patients were still moderately ill and, from that perspective, comparable to patients of other studies. It is a general issue that the most severely ill patients are hard to include in research.
      The PET procedures were carried out with arterial input functions and steady-state kinetics that previously proved superior to the less comprehensive tissue reference method applied in most studies of this field. The 4P model procedure came with the price of a slightly decreased sample size, but relative to previous studies, the sample size is large. A priori knowledge-guided data selection and constraints in the number of combinations with requirements of information on all 3 neurotransmitters proved fruitful and superior for the purpose of identification of psychosis.
      Cross-validation is an estimate of the out-of-sample generalizability. Combination 1, the best-fitted neurotransmitter model, reached significance with respect to both accuracy and AUC on the test set assessed with cross-validation and therefore seems to be generalizable to an independent sample.

      Conclusions

      We found that a combination of 3 key biological markers of psychosis (striatal DSC, levels of GABA in the ACC, and levels of Glx in the thalamus or ACC) predicted patient status when single neurotransmitters failed to do so. The findings indicate that the diagnosis of psychosis rather depends on disruption of the indirect pathway in the macrocircuits in the brain than on isolated abnormalities in the involved neurotransmitters. The data add new perspectives to the stratification of patients according to a combination of neurotransmitter disturbances, particularly prefrontal GABAergic and striatal dopaminergic neurotransmission. Thus, future treatment strategies may benefit from focusing on the combined dopaminergic and GABAergic disturbances.

      Acknowledgments and Disclosures

      This study was funded by Ph.D. grants and a postdoctoral grant from the Mental Health Services in the Capital Region of Denmark (to AKS, KT, and MØN); a Ph.D. grant from the Faculty of Health and Medical Sciences, University of Copenhagen (to KBB); an independent grant from the Lundbeck Foundation (Grant No. R155-2013-16337) to the Lundbeck Foundation Centre of Excellence for Clinical Intervention and Neuropsychiatric Schizophrenia Research (to BYG); grants from the Wørzner and Gerhard Linds Foundations; and support from the Mental Health Services, Capital Region of Denmark (to BYG).
      This project applies tools developed under National Institutes of Health ( NIH ) grants (Grant Nos. P41EB031771, R01EB016089, and R01EB023963). RE also received support from these grants.
      The funding sources had no role in the design or conduction of the study design; in the collection, analyses, and interpretation of data; or in the writing, review approval, and submission of the manuscript for publication.
      We thank the staff at Center for Neuropsychiatric Research and Centre for Clinical Intervention and Neuropsychiatric Schizophrenia Research, Mental Health Centre Glostrup, University of Copenhagen; the positron emission tomography PET center, Herlev Hospital, University of Copenhagen; and Functional Imaging Unit, Rigshospitalet, Glostrup for help.
      BYG has been the leader of a Lundbeck Foundation Centre of Excellence for Clinical Intervention and Neuropsychiatric Schizophrenia Research (January 2009–December 2021), which was partially financed by an independent grant from the Lundbeck Foundation based on international review and partially financed by the Mental Health Services in the Capital Region of Denmark, the University of Copenhagen, and other foundations. All grants are the property of the Mental Health Services in the Capital Region of Denmark and administrated by them. She has no other conflicts to disclose. BHE received lecture fees and/or is part of the advisory board at Bristol-Myers Squibb, Eli Lilly and Company, Janssen-Cilag, Otsuka Pharma Scandinavia AB, Takeda Pharmaceutical Company, Boehringer Ingelheim, and Lundbeck Pharma A/S. KBB received lecture fee from Lundbeck Pharma A/S. All other authors report no biomedical financial interests or potential conflicts of interest.
      ClinicalTrials.gov: The Pan European Collaboration on Antipsychotic-Naïve Schizophrenia II (PENCASII) Study; https://www.clinicaltrials.gov/ct2/show/NCT02339844?term=Glenth%C3%B8j&cond=Schizophrenia&rank=7; NCT02339844.

      References

        • Howes O.D.
        • Kapur S.
        The dopamine hypothesis of schizophrenia: version III--the final common pathway.
        Schizophr Bull. 2009; 35: 549-562
        • Carlsson A.
        The neurochemical circuitry of schizophrenia.
        Pharmacopsychiatry. 2006; 39: S10-S14
        • Reith J.
        • Benkelfat C.
        • Sherwin A.
        • Yasuhara Y.
        • Kuwabara H.
        • Andermann F.
        • et al.
        Elevated dopa decarboxylase activity in living brain of patients with psychosis.
        Proc Natl Acad Sci U S A. 1994; 91: 11651-11654
        • Howes O.D.
        • Kambeitz J.
        • Kim E.
        • Stahl D.
        • Slifstein M.
        • Abi-Dargham A.
        • Kapur S.
        The nature of dopamine dysfunction in schizophrenia and what this means for treatment.
        Arch Gen Psychiatry. 2012; 69: 776-786
        • Lindström L.H.
        • Gefvert O.
        • Hagberg G.
        • Lundberg T.
        • Bergström M.
        • Hartvig P.
        • Långström B.
        Increased dopamine synthesis rate in medial prefrontal cortex and striatum in schizophrenia indicated by L-(beta-11C) DOPA and PET.
        Biol Psychiatry. 1999; 46: 681-688
        • Hietala J.
        • Syvälahti E.
        • Vilkman H.
        • Vuorio K.
        • Räkköläinen V.
        • Bergman J.
        • et al.
        Depressive symptoms and presynaptic dopamine function in neuroleptic-naive schizophrenia.
        Schizophr Res. 1999; 35: 41-50
        • Mcgowan S.
        • Lawrence A.D.
        • Sales T.
        • Quested D.
        • Grasby P.
        Presynaptic dopaminergic dysfunction in schizophrenia: A positron emission tomographic [18F]fluorodopa study.
        Arch Gen Psychiatry. 2004; 61: 134-142
        • Cheng P.W.C.
        • Chang W.C.
        • Lo G.G.
        • Chan K.W.S.
        • Lee H.M.E.
        • Hui L.M.C.
        • et al.
        The role of dopamine dysregulation and evidence for the transdiagnostic nature of elevated dopamine synthesis in psychosis: A positron emission tomography (PET) study comparing schizophrenia, delusional disorder, and other psychotic disorders.
        Neuropsychopharmacology. 2020; 45: 1870-1876
        • Brugger S.P.
        • Angelescu I.
        • Abi-dargham A.
        • Mizrahi R.
        • Shahrezaei V.
        • Howes O.D.
        Heterogeneity of striatal dopamine function in schizophrenia: Meta-analysis of variance.
        Biol Psychiatry. 2020; 87: 215-224
        • Hietala J.
        • Syvälahti E.
        • Vuorio K.
        • Räkköläinen V.
        • Bergman J.
        • Haaparanta M.
        • et al.
        Presynaptic dopamine function in striatum of neuroleptic-naive schizophrenic patients.
        Lancet. 1995; 346: 1130-1131
        • Dao-Castellana M.H.
        • Paillère-Martinot M.L.
        • Hantraye P.
        • Attar-Lévy D.
        • Rémy P.
        • Crouzel C.
        • et al.
        Presynaptic dopaminergic function in the striatum of schizophrenic patients.
        Schizophr Res. 1997; 23: 167-174
        • Kumakura Y.
        • Cumming P.
        • Vernaleken I.
        • Buchholz H.G.
        • Siessmeier T.
        • Heinz A.
        • et al.
        Elevated [18F]fluorodopamine turnover in brain of patients with schizophrenia: An [18F]fluorodopa/positron emission tomography study.
        J Neurosci. 2007; 27: 8080-8087
        • Nozaki S.
        • Kato M.
        • Takano H.
        • Ito H.
        • Takahashi H.
        • Arakawa R.
        • et al.
        Regional dopamine synthesis in patients with schizophrenia using L-[β-11C]DOPA PET.
        Schizophr Res. 2009; 108: 78-84
        • Katthagen T.
        • Kaminski J.
        • Heinz A.
        • Buchert R.
        • Schlagenhauf F.
        Striatal dopamine and reward prediction error signaling in unmedicated schizophrenia patients.
        Schizophr Bull. 2020; 46: 1535-1546
        • Egerton A.
        • Murphy A.
        • Donocik J.
        • Anton A.
        • Barker G.J.
        • Collier T.
        • et al.
        Dopamine and glutamate in antipsychotic-responsive compared with antipsychotic-nonresponsive psychosis: A multicenter positron emission tomography and magnetic resonance spectroscopy study (STRATA).
        Schizophr Bull. 2021; 47: 505-516
        • Jauhar S.
        • Mccutcheon R.
        • Borgan F.
        • Veronese M.
        • Nour M.
        • Pepper F.
        • et al.
        The relationship between cortical glutamate and striatal dopamine in first-episode psychosis: a cross-sectional multimodal PET and magnetic resonance spectroscopy imaging study.
        Lancet Psychiatry. 2018; 5: 816-823
        • Sigvard A.K.
        • Nielsen M.Ø.
        • Gjedde A.
        • Bojesen K.B.
        • Fuglø D.
        • Tangmose K.
        • et al.
        Dopaminergic activity in antipsychotic-naïve patients assessed with positron emission tomography before and after partial dopamine D 2 receptor agonist treatment: Association with psychotic symptoms and treatment response.
        Biol Psychiatry. 2022; 91: 236-245
        • Nakazawaa K.
        • Zsirosa V.
        • Jianga Z.
        • Nakao K.
        • Kolata S.
        • Zhanga S.
        • Belforte J.E.
        GABAergic interneuron origin of schizophrenia pathophysiology.
        Neuropharmacology. 2013; 62: 1574-1583
        • Schmidt M.J.
        • Mirnics K.
        Neurodevelopment, GABA system dysfunction, and schizophrenia.
        Neuropsychopharmacology. 2015; 40: 190-206
        • Fazzari P.
        • Paternain A.V.
        • Valiente M.
        • Pla R.
        • Luján R.
        • Lloyd K.
        • et al.
        Control of cortical GABA circuitry development by Nrg1 and ErbB4 signalling.
        Nature. 2010; 464: 1376-1380
        • Gonzalez-Burgos G.
        • Fish K.N.
        • Lewis D.A.
        GABA neuron alterations, cortical circuit dysfunction and cognitive deficits in schizophrenia.
        Neural Plast. 2011; 2011723184
        • Öngür D.
        • Prescot A.P.
        • Mccarthy J.
        • Cohen B.M.
        • Renshaw P.F.
        Elevated gamma-aminobutyric acid levels in chronic schizophrenia.
        Biol Psychiatry. 2010; 68: 667-670
        • Kegeles L.S.
        • Mao X.
        • Stanford A.D.
        • Girgis R.
        • Ojeil N.
        • Xu X.
        • et al.
        Elevated prefrontal cortex γ-aminobutyric acid and glutamate-glutamine levels in schizophrenia measured in vivo with proton magnetic resonance spectroscopy.
        Arch Gen Psychiatry. 2012; 69: 449-459
        • de la Fuente-Sandoval C.
        • Reyes-Madrigal F.
        • Mao X.
        • León-Ortiz P.
        • Rodríguez-Mayoral O.
        • Jung-Cook H.
        • et al.
        Prefrontal and striatal gamma-aminobutyric acid levels and the effect of antipsychotic treatment in first-episode psychosis patients.
        Biol Psychiatry. 2018; 83: 475-483
        • Yang Z.
        • Zhu Y.
        • Song Z.
        • Mei L.
        • Zhang J.
        • Chen T.
        • et al.
        Comparison of the density of gamma-aminobutyric acid in the ventromedial prefrontal cortex of patients with first-episode psychosis and healthy controls.
        Shanghai Arch Psychiatry. 2015; 27: 341-347
        • De La Fuente-Sandoval C.
        • Reyes-Madrigal F.
        • Mao X.
        • León-Ortiz P.
        • Rodríguez-Mayoral O.
        • Solís-Vivanco R.
        • et al.
        Cortico-striatal GABAergic and glutamatergic dysregulations in subjects at ultra-high risk for psychosis investigated with proton magnetic resonance spectroscopy.
        Int J Neuropsychopharmacol. 2015; 19 (pyv105)
        • Goto N.
        • Yoshimura R.
        • Kakeda S.
        • Moriya J.
        • Hori H.
        • Hayashi K.
        • et al.
        No alterations of brain GABA after 6 months of treatment with atypical antipsychotic drugs in early-stage first-episode schizophrenia.
        Prog Neuropsychopharmacol Biol Psychiatry. 2010; 34: 1480-1483
        • Tayoshi S.
        • Nakataki M.
        • Sumitani S.
        • Taniguchi K.
        • Shibuya-Tayoshi S.
        • Numata S.
        • et al.
        GABA concentration in schizophrenia patients and the effects of antipsychotic medication: A proton magnetic resonance spectroscopy study.
        Schizophr Res. 2010; 117: 83-91
        • Sydnor V.J.
        • Roalf D.R.
        A meta-analysis of ultra-high fi eld glutamate, glutamine, GABA and glutathione 1HMRS in psychosis: Implications for studies of psychosis risk.
        Schizophr Res. 2020; 226: 61-69
        • Marsman A.
        • Mandl R.C.W.
        • Klomp D.W.J.
        • Bohlken M.M.
        • Boer V.O.
        • Andreychenko A.
        • et al.
        GABA and glutamate in schizophrenia: A 7 T 1H-MRS study.
        Neuroimage Clin. 2014; 6: 398-407
        • Rowland L.M.
        • Kontson K.
        • West J.
        • Edden R.A.
        • Zhu H.
        • Wijtenburg S.A.
        • et al.
        In vivo measurements of glutamate, GABA, and NAAG in schizophrenia.
        Schizophr Bull. 2013; 39: 1096-1104
        • Rowland L.M.
        • Krause B.W.
        • Wijtenburg S.A.
        • Mcmahon R.P.
        • Chiappelli J.
        • Nugent K.L.
        • et al.
        Medial frontal GABA is lower in older schizophrenia: A MEGA-PRESS with macromolecule suppression study.
        Mol Psychiatry. 2016; 21: 198-204
        • Marques T.R.
        • Ashok A.H.
        • Angelescu I.
        • Borgan F.
        • Myers J.
        • Lingford-Hughes A.L.
        • et al.
        GABA-A receptor differences in schizophrenia: A positron emission tomography study using [11C]Ro154513.
        Mol Psychiatry. 2021; 26: 2616-2625
        • Théberge J.
        • Bartha R.
        • Drost D.J.
        • Menon R.S.
        • Malla A.
        • Takhar J.
        • et al.
        Glutamate and glutamine measured with 4.0 T proton MRS in never-treated patients with schizophrenia and healthy volunteers.
        Am J Psychiatry. 2002; 159: 1944-1946
        • Théberge J.
        • Williamson K.E.
        • Aoyama N.
        • Drost D.J.
        • Manchanda R.
        • Malla A.K.
        • et al.
        Longitudinal grey-matter and glutamatergic losses in first-episode schizophrenia.
        Br J Psychiatry. 2007; 191: 325-334
        • Bustillo J.R.
        • Rowland L.M.
        • Mullins P.
        • Jung R.
        • Chen H.
        • Qualls C.
        • et al.
        1H-MRS at 4 Tesla in minimally treated early schizophrenia.
        Mol Psychiatry. 2010; 15: 629-636
        • Wang J.
        • Tang Y.
        • Zhang T.
        • Cui H.
        • Xu L.
        • Zeng B.
        • et al.
        Reduced γ-aminobutyric acid and glutamate+glutamine levels in drug-naïve patients with first-episode schizophrenia but not in those at ultrahigh risk.
        Neural Plast. 2016; 20163915703
        • Jeon P.
        • Limongi R.
        • Ford S.D.
        • Mackinley M.
        • Dempster K.
        • Théberge J.
        • Palaniyappan L.
        Progressive changes in glutamate concentration in early stages of schizophrenia: A longitudinal 7-Tesla MRS study.
        Schizophr Bull Open. 2021; 2 (sgaa072)
        • Merritt K.
        • McGuire P.K.
        • Egerton A.
        • Aleman A.
        • Block W.
        • et al.
        • 1H-MRS in Schizophrenia Investigators
        Association of age, antipsychotic medication, and symptom severity in schizophrenia with proton magnetic resonance spectroscopy brain glutamate level: A mega-analysis of individual participant-level data.
        JAMA Psychiatry. 2021; 78: 667-681
        • Godlewska B.R.
        • Minichino A.
        • Emir U.
        • Angelescu I.
        • Lennox B.
        • Micunovic M.
        • et al.
        Brain glutamate concentration in men with early psychosis: A magnetic resonance spectroscopy case-control study at 7 T.
        Transl Psychiatry. 2021; 11: 367
        • Aoyama N.
        • Théberge J.
        • Drost D.J.
        • Manchanda R.
        • Northcott S.
        • Neufeld R.W.J.
        • et al.
        Grey matter and social functioning correlates of glutamatergic metabolite loss in schizophrenia.
        Br J Psychiatry. 2011; 198: 448-456
        • Chen T.
        • Wang Y.
        • Zhang J.
        • Wang Z.
        • Xu J.
        • Li Y.
        • et al.
        Abnormal concentration of GABA and glutamate in the prefrontal cortex in schizophrenia.-An in vivo 1H-MRS study.
        Shanghai Arch Psychiatry. 2017; 29: 277-286
        • Merritt K.
        • Perez-Iglesias R.
        • Sendt K.V.
        • Goozee R.
        • Jauhar S.
        • Pepper F.
        • et al.
        Remission from antipsychotic treatment in first episode psychosis related to longitudinal changes in brain glutamate.
        npj Schizophr. 2019; 5: 12
        • Maximo J.O.
        • Briend F.
        • Armstrong W.P.
        • Kraguljac N.V.
        • Lahti A.C.
        Salience network glutamate and brain connectivity in medication-naïve first episode patients – A multimodal magnetic resonance spectroscopy and resting state functional connectivity MRI study.
        NeuroImage Clin. 2021; 32102845
        • Demjaha A.
        • Egerton A.
        • Murray R.M.
        • Kapur S.
        • Howes O.D.
        • Stone J.M.
        • McGuire P.K.
        Antipsychotic treatment resistance in schizophrenia associated with elevated glutamate levels but normal dopamine function.
        Biol Psychiatry. 2014; 75: e11-e13
        • Gleich T.
        • Deserno L.
        • Lorenz R.C.
        • Boehme R.
        • Pankow A.
        • Buchert R.
        • et al.
        Prefrontal and striatal glutamate differently relate to striatal dopamine: Potential regulatory mechanisms of striatal presynaptic dopamine function?.
        J Neurosci. 2015; 35: 9615-9621
        • Bojesen K.B.
        • Ebdrup B.H.
        • Jessen K.
        • Sigvard A.
        • Tangmose K.
        • Edden R.A.E.
        • et al.
        Treatment response after 6 and 26 weeks is related to baseline glutamate and GABA levels in antipsychotic-naïve patients with psychosis.
        Psychol Med. 2020; 50: 2182-2193
        • Bojesen K.B.
        • Broberg B.V.
        • Fagerlund B.
        • Jessen K.
        • Thomas M.B.
        • Sigvard A.
        • et al.
        Associations between cognitive function and levels of glutamatergic metabolites and gamma-aminobutyric acid in antipsychotic-naïve patients with schizophrenia or psychosis.
        Biol Psychiatry. 2021; 89: 278-287
        • Alexander G.E.
        • DeLong M.R.
        • Strick P.L.
        Parallel organization of functionally segregated circuits linking basal ganglia and cortex.
        Annu Rev Neurosci. 1986; 9: 357-381
        • Haber S.N.
        The place of dopamine in the cortico-basal ganglia circuit.
        Neuroscience. 2014; 282: 248-257
        • Modinos G.
        • Pettersson-yeo W.
        • Allen P.
        • Mcguire P.K.
        • Aleman A.
        • Mechelli A.
        Multivariate pattern classification reveals differential brain activation during emotional processing in individuals with psychosis proneness.
        Neuroimage. 2012; 59: 3033-3041
        • Soh P.
        • Narayanan B.
        • Khadka S.
        • Calhoun V.D.
        • Keshavan M.S.
        • Tamminga C.A.
        • et al.
        Joint coupling of awake EEG frequency activity and MRI gray matter volumes in the psychosis dimension: A BSNIP study.
        Front Psychiatry. 2015; 6: 162
        • Valli I.
        • Marquand A.F.
        • Mechelli A.
        • Raffin M.
        • Allen P.
        • Seal M.L.
        • McGuire P.
        Identifying individuals at high risk of psychosis: Predictive utility of support vector machine using structural and functional MRI data.
        Front Psychiatry. 2016; 7: 52
        • Ramyead A.
        • Studerus E.
        • Kometer M.
        • Uttinger M.
        • Gschwandtner U.
        • Fuhr P.
        • Riecher-Rössler A.
        Prediction of psychosis using neural oscillations and machine learning in neuroleptic-naïve at-risk patients.
        World J Biol Psychiatry. 2016; 17: 285-295
        • Lei D.
        • Pinaya W.H.L.
        • Young J.
        • Van Amelsvoort T.
        • Marcelis M.
        • Donohoe G.
        • et al.
        Integrating machining learning and multimodal neuroimaging to detect schizophrenia at the level of the individual.
        Hum Brain Mapp. 2020; 41: 1119-1135
        • Ebdrup B.H.
        • Axelsen M.C.
        • Bak N.
        • Fagerlund B.
        • Oranje B.
        • Raghava J.M.
        • et al.
        Accuracy of diagnostic classification algorithms using cognitive-, electrophysiological-, and neuroanatomical data in antipsychotic-naïve schizophrenia patients.
        Psychol Med. 2019; 49: 2754-2763
        • Ambrosen K.S.
        • Skjerbæk M.W.
        • Foldager J.
        • Axelsen M.C.
        • Bak N.
        • Arvastson L.
        • et al.
        A machine-learning framework for robust and reliable prediction of short- and long-term treatment response in initially antipsychotic-naïve schizophrenia patients based on multimodal neuropsychiatric data.
        Transl Psychiatry. 2020; 10: 276
        • Wing J.K.
        • Babor T.
        • Brugha T.
        • Burke J.
        • Cooper J.E.
        • Giel R.
        • et al.
        SCAN. Schedules for clinical assessment in neuropsychiatry.
        Arch Gen Psychiatry. 1990; 47: 589-593
        • Yung A.R.
        • Phillips L.J.
        • McGorry P.D.
        • McFarlane C.A.
        • Francey S.
        • Harrigan S.
        • et al.
        Prediction of psychosis. A step towards indicated prevention of schizophrenia.
        Br J Psychiatry Suppl. 1998; 172: 14-20
        • Tan K.R.
        • Rudolph U.
        • Lüscher C.
        Hooked on benzodiazepines: GABA A receptor subtypes and addiction.
        Trends Neurosci. 2011; 34: 188-197
        • Dale A.M.
        • Fischl B.
        • Sereno M.I.
        Cortical surface-based analysis. I. Segmentation and surface reconstruction.
        Neuroimage. 1999; 9: 179-194
        • Fischl B.
        • Sereno M.I.
        • Dale A.M.
        Cortical surface-based analysis. II: Inflation, flattening, and a surface-based coordinate system.
        Neuroimage. 1999; 9: 195-207
        • Flagstad P.
        • Mørk A.
        • Glenthøj B.Y.
        • van Beek J.
        • Michael-Titus A.T.
        • Didriksen M.
        Disruption of neurogenesis on gestational Day 17 in the rat causes behavioral changes relevant to positive and negative schizophrenia symptoms and alters amphetamine-induced dopamine release in nucleus accumbens.
        Neuropsychopharmacology. 2004; 29: 2052-2064
        • Mikell C.B.
        • Mckhann G.M.
        • Segal S.
        • Mcgovern R.A.
        • Wallenstein M.B.
        • Moore H.
        The hippocampus and nucleus accumbens as potential therapeutic targets for neurosurgical intervention in schizophrenia.
        Stereotact Funct Neurosurg. 2009; 87: 256-265
        • Nielsen M.O.
        • Rostrup E.
        • Wulff S.
        • Bak N.
        • Broberg B.V.
        • Lublin H.
        • et al.
        Improvement of brain reward abnormalities by antipsychotic monotherapy in schizophrenia.
        Arch Gen Psychiatry. 2012; 69: 1195-1204
        • Wulff S.
        • Nielsen M.Ø.
        • Rostrup E.
        • Svarer C.
        • Jensen L.T.
        • Pinborg L.
        • Glenthøj B.Y.
        The relation between dopamine D2 receptor blockade and the brain reward system: A longitudinal study of first-episode schizophrenia patients.
        Psychol Med. 2020; 50: 220-228
        • Provencher S.
        LCModel & LCMgui User’s Manual, Version 6.3-1L.
        (Available at:)
        http://s-provencher.com/lcm-manual.shtml
        Date: 2021
        Date accessed: November , 2019
        • Edden R.
        GABA Analysis toolkit.
        (Available at:)
        https://github.com/richardedden/Gannet3.1
        Date: 2020
        Date accessed: November , 2019
        • Box G.E.P.
        • Tidwell P.W.
        Transformation of the independent variables.
        Technometrics. 1962; 4: 531-550
        • Wagenmakers E.J.
        • Farrell S.
        AIC model selection using Akaike weights.
        Psychon Bull Rev. 2004; 11: 192-196
        • Phipson B.
        • Smyth G.K.
        Permutation P-values should never be zero: calculating exact P-values when permutations are randomly drawn.
        Stat Appl Genet Mol Biol. 2010; 9 (Article 39)
        • Hosmer Jr., D.W.
        • Lemeshow S.
        • Sturdivant R.X.
        Applied Logistic Regression.
        Wiley, Hoboken2013
        • Benjamini Y.
        • Hochberg Y.
        Controlling the false discovery rate: A practical and powerful approach to multiple testing.
        J R Stat Soc B. 1995; 57: 289-300
        • Seabold S.
        • Perktold J.
        Statsmodels: Econometric and statistical modeling with Python.
        in: Presented at the Proceedings of the 9th Python in Science Conference, June 28–July 3. 2010 (Austin, Texas)
        • Pedregosa F.
        • Varoquaux G.
        • Gramfort A.
        • Michel V.
        • Thirion B.
        • Grisel O.
        • et al.
        Scikit-learn: Machine learning in Python.
        J Mach Learn Res. 2011; 12: 2825-2830
        • Fusar-Poli P.
        • Meyer-Lindenberg A.
        Striatal presynaptic dopamine in schizophrenia, part II: Meta-analysis of [(18)F/(11)C]-DOPA PET studies.
        Schizophr Bull. 2013; 39: 33-42
        • Veronese M.
        • Santangelo B.
        • Jauhar S.
        • D’Ambrosio E.
        • Demjaha A.
        • Salimbeni H.
        • et al.
        A potential biomarker for treatment strati fi cation in psychosis: Evaluation of an [ 18 F ] FDOPA PET imaging approach.
        Neuropsychopharmacology. 2021; 46: 1122-1132
        • Jauhar S.
        • Nour M.M.
        • Veronese M.
        • Rogdaki M.
        • Bonoldi I.
        • Azis M.
        • et al.
        A test of the transdiagnostic dopamine hypothesis of psychosis using positron emission tomographic imaging in bipolar affective disorder and schizophrenia.
        JAMA Psychiatry. 2017; 74: 1206-1213
        • Jauhar S.
        • Veronese M.
        • Nour M.M.
        • Rogdaki M.
        • Hathway P.
        • Turkheimer F.E.
        • et al.
        Determinants of treatment response in first-episode psychosis: An 18F-DOPA PET study.
        Mol Psychiatry. 2019; 24: 1502-1512
        • Carlsson A.
        The current status of the dopamine hypothesis of schizophrenia.
        Neuropsychopharmacology. 1988; 1: 179-186
        • Geday J.
        • Østergaard K.
        • Johnsen E.
        • Gjedde A.
        STN-stimulation in Parkinson ’ s disease restores striatal inhibition of thalamocortical projection.
        Hum Brain Mapp. 2009; 30: 112-121
        • D’Ambrosio E.D.
        • Jauhar S.
        • Kim S.
        • Veronese M.
        • Rogdaki M.
        • Pepper F.
        • et al.
        The relationship between grey matter volume and striatal dopamine function in psychosis: A multimodal 18 F-DOPA PET and voxel-based morphometry study.
        Mol Psychiatry. 2021; 26: 1332-1345
        • Carlsson A.
        • Waters N.
        • Carlsson M.L.
        Neurotransmitter interactions in schizophrenia–therapeutic implications.
        Biol Psychiatry. 1999; 46: 1388-1395
        • Tamminga C.A.
        Assessing striatal dopamine in schizophrenia.
        Biol Psychiatry. 2022; 91: 170-172