Treatment effects on neurometabolite levels in schizophrenia: A meta-analysis dataset of proton magnetic resonance spectroscopy

This article describes a dataset for a meta-analysis that aimed to investigate the effects of treatment on the neurometabolite status in patients with schizophrenia (DOI of original article: https://doi.org/10.1016/j.schres.2020.03.069[1]). The data search was performed with MEDLINE, Embase, and PsycINFO. The neurometabolites investigated include glutamate, glutamine, glutamate + glutamine, gamma-aminobutyric acid, N-acetylaspartate, and myo-inositol, and the regions of interest (ROIs) include the frontal cortex, temporal cortex, parieto-occipital cortex, thalamus, basal ganglia, and hippocampus. The meta-analysis was conducted with a random-effects model, and the use of the standardized mean difference method between pre- and post-treatment of subjects for neurometabolites in each ROI of three patient groups or more. The dataset covers raw data of 39 patient groups (773 patients with schizophrenia at follow-up) with neurometabolite levels measured by magnetic resonance spectroscopy both before and after treatment. Furthermore, it contains details of clinical characteristics and treatment types for each group. Therefore, the data would be useful for a reinvestigation of treatment effects on the neurometabolite status from diverse points of view, as well as for the development of future treatment strategies for psychiatric diseases.


a b s t r a c t
This article describes a dataset for a meta-analysis that aimed to investigate the effects of treatment on the neurometabolite status in patients with schizophrenia (DOI of original article: https://doi.org/10.1016/j.schres.2020.03.069 [1] ). The data search was performed with MEDLINE, Embase, and PsycINFO. The neurometabolites investigated include glutamate, glutamine, glutamate + glutamine, gammaaminobutyric acid, N -acetylaspartate, and myo-inositol, and the regions of interest (ROIs) include the frontal cortex, temporal cortex, parieto-occipital cortex, thalamus, basal ganglia, and hippocampus. The meta-analysis was conducted with a random-effects model, and the use of the standardized mean difference method between pre-and post-treatment of subjects for neurometabolites in each ROI of three patient groups or more. The dataset covers raw data of 39 patient groups (773 patients with schizophrenia at followup) with neurometabolite levels measured by magnetic resonance spectroscopy both before and after treatment. Furthermore, it contains details of clinical characteristics and treatment types for each group. Therefore, the data would be useful for a reinvestigation of treatment effects on the neurometabolite status from diverse points of view, as well as for the development of future treatment strategies for psychiatric diseases.
© 2020 The Author(s

Value of the data
The dataset covers 39 patient groups (773 patients with schizophrenia at follow-up) with neurometabolite data for both before and after treatment, which allows for a reinvestigation of the treatment effects on the neurometabolite status from diverse points of view. The dataset includes details of the clinical backgrounds and treatment types for each patient group, facilitating the development of future treatment strategies for psychiatric diseases. The dataset would be useful for conducting a future meta-analysis associated with treatment intervention in various diseases. Table 1 demonstrates scores of Quality Assessment conducted by modified Newcastle -Ottawa Quality Assessment Scale. Fig. 1 demonstrates meta-regression analyses to investigate the effects of clinico-demographic variables on neurometabolites. Fig. 2 depicts a funnel plot of frontal Glx, thalamic NAA, and thalamic MI. Datasheet 1 includes the comprehensive data with a summary sheet for our meta-analysis.

Literature search
Two authors (MK and SM) initially screened the titles and abstracts of articles to identify potentially relevant data. These authors then assessed the eligibility of these data for our metaanalysis, which required full-text screening. Discrepancies in data selection were resolved by discussions. We excluded articles in which only figures for MRS data were reported and metabolite values were not available despite our inquiry. We

Data extraction
Two authors, MK and SM, independently extracted data. The data were then cross-checked and discrepancies were resolved by discussion between these two authors. When different articles reported data on the same metabolite from the same sample, we chose the article with the larger sample size for inclusion in our meta-analysis. The extracted data included the name of the first author, year of publication, number of patients at pre-and post-treatment, mean age of patients at pre-treatment, sex ratio of patients, clinical characteristics and symptom severity of patients, baseline treatment, duration of illness, detailed type of treatment, duration of treatment, strength of magnetic field, MRS acquisition sequence, echo time, repetition time, value of neurometabolites (mean, sd, number) at pre-and post-treatment, and scaling (creatine or water scaling, corrected by cerebrospinal fluid or not).

Eligibility criteria
Eligibility criteria were as follows: 1) Data for patients meeting the Diagnostic and Statistical Manual of Mental Disorders, 3rd, 4th, or 5th edition criteria for psychotic disorders including schizophrenia, schizoaffective, and schizophreniform; or patients meeting the Comprehensive Assessment of At-Risk Mental States criteria for being at ultra-high risk (UHR) for onset of first psychotic disorder 2) Data with neurometabolite levels (Glu, Gln, Glx, GABA, NAA or MI) for both pre-and posttreatment using 1 H-MRS 3) Data with at least five patients at each time point 4) Data sufficient to obtain mean differences between two time points 5) Data from English-language articles Exclusion criteria were as follows: 1) Cross-sectional data (only one time point) 2) Data without sufficient information for the meta-analysis regardless of our inquiry from authors 3) Non-English articles and conference abstracts   5017 articles were identified through our initial database search. Among them, 32 articles met the eligibility criteria for our meta-analysis. From these article records, we retrieved data for 39 patient groups.

Outcomes
We investigated changes in neurometabolite levels between pre-and post-treatment in the following ROIs: 1) frontal cortex including frontal white matter, anterior cingulate cortex, medial prefrontal cortex, and dorsolateral prefrontal cortex 2) temporal cortex 3) parieto-occipital cortex 4) thalamus 5) basal ganglia 6) hippocampus MRS data in the cerebellum were not investigated because of a lack of sufficient data size. When bilateral data were reported, only those of the left hemisphere were included, as it was examined in most research. In case the metabolite data of the same sample were reported from two different sub-regions within the same ROI, we included data for the sub-region more frequently used by others. If an article reported two or more kinds of measures of metabolites, we prioritized an absolute metabolite value with cerebrospinal fluid ratio correction, and if not available, we used the ratio of a metabolite to the creatine level. If data of three or more time points were reported from the same publication, we used data at the first follow-up as well as at baseline to minimize other effects on metabolites.

Meta-analysis
1) Software used for meta-analysis: Review Manager Version 5.3 ( http://tech.cochrane.org/ revman ) 2) Statistical methods: random-effects model, standardized mean difference (SMD) method between pre-and post-treatment 3) Index of heterogeneity: Q-test and I 2 index Characteristics of the included data of the meta-analysis are shown in Datasheet 1.

Effects of treatment type
Because type of treatment could influence changes in neurometabolite status, we performed the meta-analysis by dividing the patient groups into two subgroups: AP subgroups and non-AP subgroups.

Effects of ROI location within the frontal cortex
Because previous MRS research indicated that the location of ROIs within the frontal cortex might affect the neurometabolite status, we divided the frontal ROIs based on 1) whether they were principally composed of gray matter or white matter, and 2) whether they were located in the ventral part or dorsal part of the frontal lobe, and investigated neurometabolite changes in these frontal sub-regions separately.
Subgroup analyses were conducted if they were based on three or more patient groups.

Meta-regression analyses
Using Comprehensive Meta-Analysis version 3, we performed meta-regression analyses to investigate the effects of clinico-demographic variables on neurometabolites. The analyses were conducted for ROIs of five patient groups or more. The variables investigated include age, gender, illness duration, treatment duration, and symptom severity at baseline measured by Positive and Negative Syndrome Scale [3] ( Fig. 1 ).

Publication bias
For regions and metabolites in which significant treatment effect was found, we investigated publication bias by visual inspection of a funnel plot and by using Begg and Mazumdar rank correlation, with Comprehensive Meta-Analysis version 3. We did not find any publication bias for frontal cortex Glx, thalamus NAA and thalamic MI (tau = 0.15, p = 0.49; tau = −0.07, p = 0.85; tau = 1.00, p = 0.11, respectively) ( Fig. 2 ).

Quality assessment
Risk of bias was assessed by modified Newcastle -Ottawa Quality Assessment Scale [4] . Participants' selection (case definition, representativeness, ascertainment of exposure, definition of controls) and exposure (outcome assessment, follow-up period, adequacy of follow-up) were independently scored by two review authors, MK and SM. Discrepancies were resolved by discussion between the two. Higher scores indicate better quality (maximum total score = 14).

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships which have, or could be perceived to have influenced the work reported in this article.