
==== Front
Brain Commun
Brain Commun
braincomms
Brain Communications
2632-1297
Oxford University Press UK

10.1093/braincomms/fcae094
fcae094
Original Article
AcademicSubjects/MED00310
AcademicSubjects/SCI01870
Aberrant connectivity in the hippocampus, bilateral insula and temporal poles precedes treatment resistance in first-episode psychosis: a prospective resting-state functional magnetic resonance imaging study with connectivity concordance mapping
https://orcid.org/0000-0003-2288-7821
Skouras Stavros Department of Fundamental Neurosciences, Faculty of Medicine, University of Geneva, CH-1211 Geneva, Switzerland
Department of Neurology, Inselspital University Hospital Bern, CH3010 Bern, Switzerland

Kleinert Maria-Lisa Faculty of Medicine, University of Barcelona, 08036 Barcelona, Spain

https://orcid.org/0000-0002-8909-0676
Lee Edwin H M Department of Psychiatry, University of Hong Kong, Hong Kong, China

Hui Christy L M Department of Psychiatry, University of Hong Kong, Hong Kong, China

https://orcid.org/0000-0003-2766-8476
Suen Yi Nam Department of Psychiatry, University of Hong Kong, Hong Kong, China

https://orcid.org/0000-0001-8658-5780
Camchong Jazmin Department of Psychiatry, University of Minnesota, Minneapolis, MN 55454, USA

Chong Catherine S Y Department of Psychiatry, Kwai Chung Hospital, Hong Kong, China

Chang Wing Chung Department of Psychiatry, University of Hong Kong, Hong Kong, China

Chan Sherry K W Department of Psychiatry, University of Hong Kong, Hong Kong, China

Lo William T L Department of Psychiatry, Kwai Chung Hospital, Hong Kong, China

Lim Kelvin O Department of Psychiatry, University of Minnesota, Minneapolis, MN 55454, USA

Chen Eric Y H Department of Psychiatry, University of Hong Kong, Hong Kong, China

Correspondence to: Edwin H. M. Lee, MBChB (CUHK), MSc (CUHK), MRCPsych, FHKCPsych, FHKAM (Psychiatry) Department of Psychiatry, Queen Mary Hospital, University of Hong Kong 2/F, New Clinical Building Hong Kong, China E-mail: edwinlhm@hku.hk
Stavros Skouras and Maria-Lisa Kleinert contributed equally to this work.

2024
04 5 2024
04 5 2024
6 3 fcae09405 3 2023
04 12 2023
17 4 2024
04 5 2024
© The Author(s) 2024. Published by Oxford University Press on behalf of the Guarantors of Brain.
2024
https://creativecommons.org/licenses/by/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Functional connectivity resting-state functional magnetic resonance imaging has been proposed to predict antipsychotic treatment response in schizophrenia. However, only a few prospective studies have examined baseline resting-state functional magnetic resonance imaging data in drug-naïve first-episode schizophrenia patients with regard to subsequent treatment response. Data-driven approaches to conceptualize and measure functional connectivity patterns vary broadly, and model-free, voxel-wise, whole-brain analysis techniques are scarce. Here, we apply such a method, called connectivity concordance mapping to resting-state functional magnetic resonance imaging data acquired from an Asian sample (n = 60) with first-episode psychosis, prior to pharmaceutical treatment. Using a longitudinal design, 12 months after the resting-state functional magnetic resonance imaging, we measured and classified patients into two groups based on psychometric testing: treatment responsive and treatment resistant. Next, we compared the two groups’ connectivity concordance maps that were derived from the resting-state functional magnetic resonance imaging data at baseline. We have identified consistently higher functional connectivity in the treatment-resistant group in a network including the left hippocampus, bilateral insula and temporal poles. These data-driven novel findings can help researchers to consider new regions of interest and facilitate biomarker development in order to identify treatment-resistant schizophrenia patients early, in advance of treatment and at the time of their first psychotic episode.

Skouras et al. used concordance connectivity mapping, an advanced method of functional connectomics, that highlights differences in the consistency of functional connectivity between two groups. Results show that prior to medication, treatment-resistant patients with first-episode psychosis show more consistent functional connectivity patterns in the hippocampus, insula and temporal poles.

Graphical Abstract

Graphical Abstract

treatment resistant
schizophrenia
resting-state
fMRI
connectivity concordance mapping
Health and Medical Research Fund 10.13039/501100005847 Hong Kong Government Health Bureau Secretariat 11121271
==== Body
pmcIntroduction

Schizophrenia is a lifelong and severely disabling neurodevelopmental disorder,1 with a lifetime prevalence of ∼0.7% worldwide.2 The course of schizophrenia can follow a heterogeneous range of trajectories, from severe cases of repeated relapse and deterioration, to cases in which a single psychotic episode is followed by complete recovery.3 Antipsychotic pharmacotherapy continues to be the frontline treatment for schizophrenia4,5; however, an integrated treatment approach should always consider psychosocial interventions and attention to environmental circumstances.3 A promising window for course-altering interventions is the time of initial diagnosis of the first-episode psychosis (FEP).6 Rapid intervention with antipsychotics decreases the duration of untreated psychosis and, therefore, enhances patients’ quality of life, social functioning and long-term outcomes.7

Nevertheless, about 30–35% of patients with schizophrenia do not respond to first-line antipsychotics and therefore fall into the category of treatment-resistant schizophrenia (TRS).8 Other studies reported this proportion of patients to be even higher, in the range of 40–60%.9,10 Treatment-resistant patients tend to experience more persistent positive, negative and cognitive symptoms, such as wide-ranging deficits in verbal memory and learning and in language functions,11 leading to worsened social functioning, including lower marriage rates and increased likelihood of residence in facilities.12 Furthermore, treatment resistance is associated with a higher socio-economic burden,13 long-term disability12,13 and higher family/caregiver burden.14

TRS can be defined in many ways and comparing studies of TRS can be ‘akin to comparing apples to oranges’.15 This leads to inconsistent and not evidence-based identification and management of TRS in clinical practice.16 As a result, consensus criteria have been proposed to define TRS as characterized by a limited symptom reduction [as assessed by e.g. the Positive and Negative Syndrome Scale (PANSS) or the Brief Psychiatric Rating Scale (BPRS)] with at least moderate functional impairment during a prospective trial or observation of ≥6 weeks and ≥2 past adequate treatment episodes with different non-clozapine antipsychotic drugs.15,16

Therapeutic nihilism can result in sustained ineffective treatment, the delay of appropriate treatment or not offering it at all.17 Having prior knowledge of patients’ antipsychotic treatment response opens up several avenues for improving clinical decision-making and, consequently, patient outcomes.18 Clozapine is considered the first-line pharmacological agent for patients with TRS. Second-line treatment options include pharmacological and non-pharmacological augmentation of clozapine, such as medication combinations, electroconvulsive therapy (ECT), repetitive transcranial magnetic stimulation, deep brain stimulation and cognitive behavioral therapy.19 Early identification of TRS is key because 40–50% of these patients respond to clozapine.20,21 Furthermore, the shorter the delay of clozapine initiation, the better the response to it.22,23

Since clozapine exhibits low dopamine receptor D2 occupancy compared with other antipsychotics, clozapine’s success in treating TRS patients suggests that its underlying mechanism of action may extend beyond dopamine receptors.24 Aberrant interactions between dopaminergic and glutamatergic signalling pathways in individuals with TRS have been proposed.25-27 In this scenario, an N-methyl-D-aspartate receptor hypofunction leads to decreased activity of cortical GABA-ergic interneurons and a diminished inhibitory control of pyramidal glutamate neurons, which accounts for increased activity of dopaminergic projections from the midbrain to the striatum.28 The pharmacodynamics of clozapine include its ability to modulate glutamate activity. Clozapine’s hypothesized effects involve partial agonist activity of N-methyl-D-aspartate receptor, along with its ability to regulate glutamate transport, leading to a potential enhancement of glutamate transmission.29-31 However, it is probably the collective impact of clozapine’s affinity to a diverse range of neuroreceptors, e.g. D4, 5-HT2A, α1 and m1, that explains its superior effectiveness compared with classical antipsychotics in the setting of TRS. This aberrant synaptic connectivity, leading to a disruption of communication between brain networks, embodies the cornerstone within the disconnection hypothesis of the psychopathology of schizophrenia,32 although it is worth noting that aberrant glutamatergic signalling pathways linked to hyperdopaminergic states are not specific to TRS and are possible in all disorders within the psychosis spectrum. Moreover, there is evidence suggesting that TRS in FEP might be neurobiologically different from TRS developing over time.33,34 Whereas the former features normal dopamine function, while glutamate or other pathways contribute to the distinct neurobiology of treatment resistance,34 the latter can be explained by progressive dopamine hypersensitivity in the striatum induced by continuous antipsychotic dopamine receptor D2 blockage.35 Both models are not mutually exclusive but could explain the neurobiological differences between two clinically different presentations of TRS, i.e. those who do not respond from the beginning and those who respond initially but their response wears off after some time.34

Functional connectivity (FC) analysis of resting-state functional MRI (rs-fMRI) measures temporal correlations of spontaneous blood oxygen levels (BOLD) amongst spatially distributed brain regions, with the assumption that areas with correlated activity form functional networks.36 Recent evidence suggests that FC during rs-fMRI holds promise as a predictive biomarker for TRS.18,23,37 Disrupted FC between different brain regions has been associated with TRS, specifically involving the prefrontal cortex, striatum and other subcortical structures,38 the insula and the anterior cingulate cortex39 and the central opercular cortex and the sensorimotor cortex.40 In a systematic review of rs-fMRI investigations for predicting antipsychotic response in schizophrenia, Mehta et al.18 reported striatal and default mode network functional segregation and integration metrics to be consistent determinants of treatment response (with a pooled odds ratio of 12.66). Nevertheless, research has not yet converged on any specific hypothesis of aberrant FC that distinguishes treatment responders from non-responders.40

The lack of consensus is favoured by several issues. On the one hand, to date, most studies that examine FC rs-fMRI compare responders and non-responders to antipsychotic treatment on a cross-sectional level (see recent review on such cross-sectional studies41). Not only can the results of such studies be considered biased by the type of antipsychotic treatment itself but also comparisons are often made between non-responders and healthy controls, instead of non-responders versus responders. Only few prospective studies exist that examine baseline rs-fMRI data in FEP patients with regard to subsequent treatment response over a time period of weeks/months. On the other hand, the lack of consensus might be a consequence of the broad variety of data analytical approaches that have been used to conceptualize and measure FC. For instance, some researchers used the specific FC of two a priori regions of interest (ROIs) with eight other a priori ROIs.42 Others measured FC as the co-activation of a priori seed ROIs with the rest of the brain.43 Independent component analysis relies on post hoc decisions in order to determine which components are functionally relevant, by measuring the co-activation between brain networks with the rest of the brain.44,45 With few exceptions, graph theory approaches typically examine pairwise co-activations between all ROIs defined by anatomical and functional parcellation units from a brain atlas.46 What most approaches have in common is that they are driven by specific hypotheses about regions or networks of interest, which could bias results and conclusions. Furthermore, the variety of analytical algorithms may very well measure distinct underlying phenomena8 and does not facilitate the convergence of findings across studies.38 Thereby, it is crucial to use voxel-wise, whole-brain algorithms to allow bias-free neuromarker discovery for TRS in FEP patients. To our knowledge, the present study is the first to use connectivity concordance mapping (CCM) in this context. CCM is a computationally intensive method that identifies the brain areas that exhibit the highest consistency in patterns of FC, across subjects (see ‘Materials and methods’). The result of a CCM analysis is a voxel-wise map of concordance values. Regions of high inter-subject concordance can be assumed to be functionally consistent and may thus be used as ROIs for further investigations.47

Materials and methods

Participants

This study was approved by the Ethics Committee of the Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster. Written informed consent was obtained from all the participants (or their parents for those under the age of 18 years) after complete description of the study. The study was performed in accordance with the Declaration of Helsinki. Between July 2014 and July 2017, a total of 106 patients aged between 15 and 25 years with a FEP were recruited from the Early Assessment Service for Young people with psychosis in Hong Kong, which provides a specialized intervention service to FEP. Patients were selected consecutively from all in- and out-patient psychiatric departments at Queen Mary Hospital and Kwai Chung Hospital.

Patients with a DSM-IV diagnosis of schizophrenia, schizophreniform disorder, schizoaffective disorder, brief psychotic disorder, delusional disorder or psychosis not otherwise specified were included in the study. Participants had to be first-episode cases with no antipsychotic treatment history. Exclusion criteria were intellectual disability, substance-induced psychosis, psychotic disorder because of a general medical condition or an inability to speak Cantonese Chinese for the research interview.

All patients underwent MRI scanning at baseline. The study followed a naturalistic observational study design with no influence on the following therapy and clinical assessments determined by the patients’ clinicians that were performed according to standard clinical practice. The majority of patients received second-generation antipsychotics in doses recommended by APA Practice Guidelines. Chlorpromazine equivalent doses were computed for analysis (50–1000 mg/day).48 Out of the 106 participants who entered the study and received the baseline assessment including rs-fMRI scans, only 60 participants could be followed up at 12 months due to a high drop-out rate of 43%.

Study design

The study was performed as a prospective, 1-year follow-up study of treatment response in FEP patients. As shown in Fig. 1, assessments took place at the first intake of antipsychotic treatment and 12 months later. Basic demographics, premorbid functioning and course-related variables were assessed at study entry. Other clinical, functional and neurocognitive predictors were assessed at baseline and at 12 months. See Supplementary Table 1 for a complete list of measurements and questionnaires used. For the purposes of this study, anatomical and functional MRI data were collected in all patients at baseline.

Figure 1 Study design. A total of 106 patients with a FEP were recruited and underwent clinical, functional and neurocognitive assessments at baseline, including anatomical and functional MRI. Over the course of the following 12 months, participants received their individual antipsychotic treatment. Sixty participants participated in follow-up visits after 12 months due to a high drop-out rate. At that time, we classified the remaining participants into responders (n = 19) and non-responders (n = 41), based on their longitudinal neurocognitive assessments. In order to qualify as a non-responder, the following criteria had to be met, in relation to the baseline assessments: <50% symptom reduction in total PANSS score, at least moderate functional impairment assessed with Social and Occupational Functioning Assessment Scale score and ≥2 past adequate treatment episodes with different non-clozapine antipsychotic drugs. To investigate patterns of brain function that relate to responsiveness to antipsychotic medication, we applied CCM to the rs-fMRI data acquired at baseline and compared the connectivity concordance maps of the two groups. See also Supplementary Table 1.

Responders versus non-responders classification criteria and group characteristics

Treatment responsiveness was assessed with the PANSS49 by experienced psychiatrists. The following criteria were adopted to define TRS: <50% symptom reduction in total PANSS score and at least moderate functional impairment assessed with Social and Occupational Functioning Assessment Scale score, 12 months after FEP and ≥2 past adequate treatment episodes with different non-clozapine antipsychotic drugs and documented adherence. According to these criteria, participants were divided into two groups: treatment responsive (n = 19) and treatment resistant (n = 41). Table 1 shows the demographic and clinical data of both groups.

Table 1 Demographical and clinical data of participants

	Responders (n = 19)	Non-responders (n = 41)	Statistical testing	
Age (y)	26.3 ± 13.2	28.3 ± 12.0	t(58) = 0.804	
P = 0.432	
Gender (M/F)	7/12	13/28	χ 2(1, 60) = 0.154	
P = 0.695	
Education level (y)	12.1 ± 3.2	11.8 ± 3.7	t(58) = 0.794	
P = 0.435	
DUP (d)	220 ± 314	285 ± 458	t(58) = 0.241	
P = 0.819	
PANSS score at baseline	53.68 ± 13.85	39.12 ± 6.66	t(58) = 5.2	
P < 0.001	
Antipsychotic dose (mg/d)a	403.6 ± 267.1	367.4 ± 176.0	t(58) = −0.041	
P = 0.968	
DUP, duration of untreated psychosis; PANSS, Positive and Negative Syndrome Scale.

aDose of antipsychotic medication at 12 m was converted to chlorpromazine equivalent dosages as proposed by Woods.48

MRI data acquisition

MRI data were acquired on a 3.0-T Philips Achieva whole-body MRI scanner (Philips, The Netherlands), at the University of Hong Kong. A high-resolution (1 × 1 × 1 mm) T1-weighted anatomical reference image was acquired from each participant using a rapid acquisition gradient echo sequence featuring 160 slices per volume. During rs-fMRI, continuous echo planar imaging was used with an echo time of 30 ms and a repetition time of 2 s. Slice acquisition was interleaved within the repetition time interval. Two hundred fifty whole-brain volumes were acquired for each participant. The matrix acquired was 80 × 80 voxels with a field of view of 240 mm, resulting in an in-plane resolution of 3 mm. Slice thickness was 3.5 mm (38 slices, whole-brain coverage).

Image processing

Anatomical images

T1 images were processed with the N4 nonparametric non-uniform intensity normalization bias correction function50 of the Advanced Normalization Tools51 and an optimized blockwise non-local means denoising filter.52 To segment anatomical images into grey matter, white matter and CSF, the VBM8 (Structural Brain Mapping Group, University of Jena, Jena, Germany; http://www.neuro.uni-jena.de/vbm/) and SPM12 (Wellcome Department of Imaging Neuroscience Group, London, UK; http://www.fil.ion.ucl.ac.uk/spm) toolboxes were used. The cranium was accurately removed from anatomical brain images using graph-cut53 and FSL (FMRIB, Oxford, UK; https://fsl.fmrib.ox.ac.uk/fsl/). A custom anatomical template was computed for the entire sample using the Advanced Normalization Tools multivariate template construction process.54,55 We computed neuroanatomically plausible symmetric diffeomorphic matrices in order to transform each subject’s anatomical data to the optimal template and afterwards to the Montreal Neurological Institute (MNI) space54,56 as defined by the Chinese brain atlas ‘Chinese_56’.57 Prior to the normalization of data sets and similarly as for the anatomical data, the cranium of the Chinese_56 brain atlas had also been removed. In accordance with best practices as well as to ensure optimal normalization while avoiding multiple interpolations, all transformation matrices were concatenated and applied to the data sets in a single step.

Functional images

Using Matlab 2014b (MathWorks Inc., Natick, MA, USA), SPM12 and the CONN functional connectivity toolbox (v17; www.conn-toolbox.org), functional data were processed according to the following steps: slice time correction, estimation of movement parameters, co-registration, band-pass filtering between 0.1 and 0.01 Hz, detrending, denoising and repairing of artefacts using the 95th percentile settings of the ART artefact detection tools. Only subjects with at least 90% valid volumes were considered further in the analysis and ‘scan nulling’ regressors were applied on any affected volumes.58 Average CSF signal, average white matter signal and 24 Volterra expansion movement parameters were regressed out of each participant's time series. Using FSL, each subject’s functional data was masked by their equivalent grey matter masks. Using Advanced Normalization Tools, the functional data were normalized to MNI space based on their respective diffeomorphic matrices. Functional data sets were smoothed by a 6 mm FWHM kernel, using the Leipzig Image Processing and Statistical Inference Algorithms (version 2.2.7).

Statistical analysis and CCM algorithm

Chi-square goodness-of-fit testing was used to assess the normality of the distribution of each potential confounding variable for our sample. For the variables where the assumption of normality was violated, we utilized non-parametric independent samples t-tests, with 100 000 permutations per test (that do not require the parametric assumptions to be met), to compare between the two groups in our sample (see results in Table 1).

Lohmann et al.47 proposed CCM as a tool for model-free analysis of rs-fMRI data of the human brain. The aim of CCM is to visualize the reproducibility or inter-subject consistency of FC patterns across the entire brain. For this purpose, we computed the correlations of time courses of each voxel with every other voxel in each data set. The result is a correlation pattern of each patient’s brain. In a second step, we looked at how consistent these correlation patterns are across patients within each group. Inter-subject consistency or concordance was measured using Kendall’s W.59 Voxels whose correlation pattern was consistent across all data sets within a patient group (e.g. non-responders) received high values, i.e. high concordance values. The result of a CCM analysis is a voxel-wise map of concordance values ranging from 0 to 1 (0 = no consistency across data sets; 1 = perfect consistency across data sets). For the purpose of this study, patients were divided into two groups according to the TRS criteria outlined above: treatment responsive (na = 19) and treatment resistant (nb = 41). A connectivity concordance map was computed for each of the two groups separately, based on ∼50 000 grey matter voxels, by utilizing the CCM algorithm described by Lohmann et al.47

For each patient data set k and each voxel address i, a connectivity vector was obtained, whose sik=(si,1k,…,si,nk) entries si,jk contain a pairwise similarity measure between the time courses of voxels i and j. In this study, the pairwise similarity measure used was Pearson’s linear correlation coefficient, defined as follows:

rxy=∑t(xt−x¯)(yt−y¯)∑t(xt−x¯)2(yt−y¯)2,

with xt and yt being the time series in a pair of two voxels x and y, and x¯,y¯ being their temporal means. The point of interest in CCM is the concordance of the connectivity vectors sik across the multiple patient data sets of each group. Kendall’s W, which is a non-parametric statistic for rank correlation, was used as a measure of concordance. It yields values between 0 (total disagreement) and 1 (total agreement) across data sets, without a requirement for any parametric assumptions to be met. Via a resource-intensive computational process, one such concordance value is computed across all the data sets of each group while considering the pairwise similarity of each voxel with all other voxels, for each of ∼50 000 grey matter voxels. This concordance value becomes the CCM in voxel I, resulting in two images: one CCM image for the treatment-responsive group and one CCM image for the treatment-resistant group. Each of these images quantified how consistent the connectivity pattern of each voxel was, across the participants belonging to the respective group. Lastly, the CCM image of the treatment-resistant group was subtracted from the CCM image of the treatment-responsive group, to obtain the difference in CCM values between the two groups (ΔCCM). Thresholding of ΔCCM > 0.05 combined with a thresholding of cluster size k > 10 voxels was applied to produce the final results displayed in Fig. 2.

Figure 2 fMRI results. To investigate patterns of brain function that relate to responsiveness to antipsychotic medication, CCM was applied to rs-fMRI data acquired data from schizophrenia patients soon after the time of FEP. Two CCM images were computed: one for the group of responders and one for the group of non-responders. Their subtraction resulted in the presented second-level contrast image that depicts areas with higher connectivity concordance in patients who were classified as non-responders (i.e. treatment-resistant patients). The colour map shows the magnitude of differences between the two groups, in CCM values based on Kendall’s W, particularly in the hippocampus, bilateral insula and temporal poles. Kendall’s W is a statistical coefficient of concordance that assumes voxel values in the range between 0 (indicating no consistency of FC across data sets in the group) and 1 (indicating perfect consistency of FC across data sets in the group).

Results

Basic demographic and clinical characteristics

Table 1 shows the demographic information and clinical characteristics of our final sample (N = 60). Thirty-two per cent of the participants responded to antipsychotic treatment, whereas the remaining 68% were classified as treatment resistant. The two groups did not significantly differ in terms of age, gender, education, duration of untreated psychosis and antipsychotic medication dosages. However, there was a significant difference with regard to the PANSS score at baseline [t(58) = 5.2, P < 0.001, (confidence interval 8.7–19.5)]. The treatment-responsive group had a higher PANSS score at baseline (53.68 ± 13.85) compared to non-responders (39.12 ± 6.66).

FC and connectivity concordance maps

Our analyses yielded the clusters depicted in Fig. 2 that show areas of more consistent FC for the treatment-resistant group, in a network including the left hippocampus, the bilateral insula and the temporal poles. Complete details are presented in Table 2.

Table 2 Areas with higher connectivity concordance in non-responders

AAL label	Hemisphere	MNI coordinates	Volume
(mm3)	mean ΔCCM	max ΔCCM	SD ΔCCM	
X	Y	Z	
Supramarginal gyrus	Left	−51	−39	28	4212	0.07	0.12	0.01	
BA13, pars opercularis	Right	39	15	4	1539	0.07	0.12	0.02	
BA47	Left	−54	27	−5	1296	0.06	0.09	0.01	
BA20, inferior temporal lobule	Left	−57	−9	−35	1080	0.08	0.19	0.03	
Insula	Left	−33	24	7	891	0.06	0.08	0.01	
Insula	Left	−45	12	−5	783	0.06	0.09	0.01	
No label available (nearest to lingual gyrus)	Left	−27	−63	7	621	0.07	0.1	0.01	
BA38, temporal pole	Right	39	18	−29	459	0.06	0.09	0.01	
IFG, pars triangularis	Right	54	39	−2	405	0.06	0.08	0.01	
Hippocampus	Left	−27	−21	−17	378	0.06	0.07	0.01	
IFG, pars triangularis	Right	51	27	22	324	0.06	0.07	0.01	
MNI coordinates specify the peak coordinate of each cluster. Labels correspond to the peak coordinates, according to the Automated Anatomical Labeling atlas and the xjview version 10.0 visualization tool (www.alivelearn.net/xjview/). The mean, max and standard deviation (SD) of each cluster are displayed.

AAL, Automated Anatomical Labelling; BA, Brodmann area; CCM, concordance connectivity mapping; IFG, inferior frontal gyrus; MNI, Montreal Neurological Institute.

Discussion

The purpose of this study was to utilize an advanced data-driven, voxel-wise, whole-brain analysis approach to investigate rs-fMRI data from drug-naïve FEP patients, in order to discover potential ROIs to guide future research of TRS biomarkers. We identified a network of areas showing consistently increased FC, including the left hippocampus, bilateral insula and temporal poles in the treatment-resistant group. These findings are supported by previous studies that have linked TRS to increased FC in cortical and subcortical networks including the insula60,61 and the temporal poles.60 However, other studies have associated treatment resistance in schizophrenia patients with decreased FC of those same regions62,63 as well as the hippocampus.64 This highlights the relevance of our findings in relation to an ongoing debate. Our results suggest that these areas differ most between treatment-resistant and treatment-responsive patients, in terms of their connectivity with the rest of the brain, thereby explaining why previous studies have generated conflicting results.

In schizophrenia patients, the insula has been proposed as a potential brain area that may shed light on the neurobiological underpinnings of treatment resistance.39 Some of the various functions the insula is involved in include homeostasis,65-67 taste perception,68,69 auditory perception,70,71 multimodal sensory processing,72 pain,73 interoception,74-76 self-consciousness77-79 and social emotions80,81 This broad range of functions ensues from (i) extensive viscerosensory inputs into the region and (ii) the insula’s anatomical location that allows for strong reciprocal connections with the prefrontal, somatosensory and temporal areas, as well as with the limbic system.82 The anterior insula and the anterior cingulate cortex form the salience network. The salience network modulates both bottom-up and top-down processing of stimuli and, moreover, controls the switch from a resting-state and the default mode network to an attentive state and the central executive network.83 Altered auditory evoked potentials and abnormal sensory gating are key findings that help to explain auditory hallucinations, one of the most common positive symptoms found in schizophrenia patients.84,85 Strikingly, Alonso-Solís et al.60 found treatment-resistant auditory verbal hallucinations linked to higher FC in a network including the insular cortex. Likewise, alterations across various interoceptive systems have been identified in schizophrenia.86 Furthermore, aberrant activation and FC of the salience network39,87 and the default mode network16,84 have been associated with TRS. Interestingly, inter-individual diversity in the insula’s FC explains variability in some of the clinical symptoms of schizophrenia.88

Structural and functional abnormalities of the hippocampus are amongst the most consistent findings in schizophrenia neuroscience research.89 Hippocampal dysfunction may therefore explain two of the most prominent features of the schizophrenia clinical spectrum, i.e. cognitive deficits (including memory function) and reality distortion.90,91 A systematic review suggested that hippocampal hyperactivity (measured by BOLD, cerebral blood volume and cerebral blood flow) is caused by a decrease in hippocampal inhibitory interneurons in schizophrenia patients.92 In fact, hippocampal hyperactivity is one the most consistent functional aberrations in schizophrenia and it predates the onset of psychosis and hippocampal volume reduction.92,93 Interestingly, post-mortem studies heavily support a reduction in hippocampal subfield volumes, total neuron counts and neuron size but preserved neuron density, predominantly lateralized to the left hemisphere of schizophrenia patients.94 Moreover, these changes seem to be accentuated in TRS. When compared with matched treatment-responsive patients, TRS patients showed poorer performance in working memory and smaller hippocampal volume.95 The hippocampus can be considered as a cornerstone of the glutamate hypothesis of schizophrenia; hence, N-methyl-D-aspartate receptor hypofunction and decreased inhibitory interneuron activity support the idea of hippocampal hyperactivity. Dysfunctional connectivity, including connections to the striatum and the frontal lobes, could result from such interneuron abnormalities.96

Additional evidence for the hippocampus’ and the insula’s involvement in TRS comes from research in the field of ECT. There is growing support for the addition of ECT to antipsychotic treatment regimes in TRS.97-99 A recent systematic review of the neural effects of ECT in schizophrenia patients found the hippocampus and insula to be key regions of modulation after ECT.100 This applied to morphometry, FC and symptom association measures. The temporal pole is an integral component of the paralimbic circuit, together with the insula and the orbitofrontal cortex.101 The temporal pole plays an important role in various cognitive functions, olfaction and affectional–sensory integration.102 Abnormal connectivity patterns have been reported for the temporal pole as well as other limbic and paralimbic regions in schizophrenia.103 Goswami et al.104 found increased FC in schizophrenia patients between the areas of the right temporal pole and the left hippocampus.

One of the advantages of the present study is its naturalistic prospective design that renders a high ecological validity. Nevertheless, there are also some important limitations. The entire patient sample did not undergo an identical antipsychotic treatment, and even though this did not bias the rs-fMRI data (obtained before treatment onset), it may have had an effect on treatment response. Furthermore, the patient group classified as treatment responders had a significantly higher PANSS score and, hence, more severe psychotic symptoms at baseline, in comparison with the TRS patient group. Similar findings have been reported elsewhere105 and support the idea that current antipsychotic treatment options are more effective for more severe, pan-symptomatic patients. Nevertheless, this difference raises the possibility that the severity of the symptoms could be a confounding factor, explaining the differences in the neuroimaging results. Due to the mathematical specifics of CCM, it is not possible to add covariates to the analysis to eliminate the possibility of confounding effects. This is a limitation of our study. Future studies should investigate whether the observed neural differences can be explained by differences in schizophrenia symptom severity, as well as whether they are related to specific subtypes of FEP and whether they are related to specific types of symptoms. Moreover, it is worth noting that our results may not generalize to patients who develop TRS over time, who may harbour different biological pathways. Despite these limitations, our results reveal a network of key brain regions that can be further explored in FC studies of TRS research. Future research in the field should move towards precision medicine approaches to individualized treatment of schizophrenia.106 This endeavour would require the merging of neuroimaging with big data and machine learning techniques in order to transition from between-group comparisons to inferences on the individual level.38,107

Conclusion

In summary, TRS represents the most chronic form of psychotic illness, a personal tragedy with a high caregiver and socioeconomic burden.13 Schizophrenia patients resistant to antipsychotic treatment comprise the lowest community functioning performance amongst psychiatric conditions12 and often experience worse clinical outcomes19 when compared with treatment-responsive patients. The present results contribute to a growing body of evidence that portrays TRS as a singular neurobiological sub-type of schizophrenia with a distinctive neural FC signature even at illness onset, before starting antipsychotic treatment. Hence, it is not surprising that TRS patients respond to fundamentally different treatment regimens than treatment-responsive patients.34 The current results contribute to disentangling the intricate neural patterns of treatment resistance in order to elucidate key predictors of TRS. Integrating functional magnetic resonance imaging in the assessment of psychotic patients at the time of their first episode, in order to decide on a certain line of treatment or another, opens up very promising future avenues. Having prior knowledge of the response to antipsychotic treatment of patients creates several possibilities to improve clinical decision-making at an early stage of the disease and, consequently, to improve patients’ quality of life.

Supplementary Material

fcae094_Supplementary_Data

Acknowledgements

We would like to thank all the participants for their participation in this project. We would also like to thank the Health and Medical Research Fund for supporting this project.

Supplementary material

Supplementary material is available at Brain Communications online.

Funding

This work has received funding from the Health and Medical Research Fund of the Hong Kong Government Health Bureau Secretariat (project number 11121271). Publication was made possible in part by support from the University of Hong Kong (HKU) Libraries Open Access Author Fund sponsored by the HKU Libraries.

Competing interests

The authors report no competing interests.

Data availability

Available data can be shared for non-commercial research purposes, upon reasonable request to the authors.
==== Refs
References

1 Insel  T . Rethinking schizophrenia. Nature. 2010;468 (7321 ):187–193.21068826
2 McGrath  J, Saha  S, Chant  D, Welham  J. Schizophrenia: A concise overview of incidence, prevalence, and mortality. Epidemiol Rev.  2008;30 (1 ):67–76.18480098
3 Vita  A, Barlati  S. Recovery from schizophrenia. Curr Opin Psychiatry.  2018;31 (3 ):246–255.29474266
4 Keepers  G, Fochtmann  L, Anzia  J, et al  The American Psychiatric Association practice guideline for the treatment of patients with schizophrenia. Am J Psychiatry. 2020;177 (9 ):868–872.32867516
5 Sommer  I, Tiihonen  J, van Mourik  A, Tanskanen  A, Taipale  H. The clinical course of schizophrenia in women and men—A nation-wide cohort study. NPJ Schizophr.  2020;6 (1 ):12.32358572
6 Millan  M, Andrieux  A, Bartzokis  G, et al  Altering the course of schizophrenia: Progress and perspectives. Nat Rev Drug Discove. 2016;15 (7 ):485–515.
7 Marino  L, Nossel  I, Choi  J, et al  The RAISE connection program for early psychosis. J Nerv Ment Dis. 2015;203 (5 ):365–371.25900546
8 Kahn  R, Winter van Rossum  I, Leucht  S, et al  Amisulpride and olanzapine followed by open-label treatment with clozapine in first-episode schizophrenia and schizophreniform disorder (OPTiMiSE): A three-phase switching study. Lancet Psychiatry. 2018;5 (10 ):797–807.30115598
9 Juarez-Reyes  MG, Hargreaves  WA, Hansen  MS, Bacchetti  P, Battl  M, Shumway  M. Restricting clozapine use: The impact of stringent criteria on eligibility for clozapine among public mental health clients. Psychiatr Serv. 1995;46 (8 ):801–806.7583481
10 Essock  SM, Hargreaves  WA, Dohm  F-A, Goethe  J, Carver  L, Hipshman  L. Clozapine eligibility among state hospital patients. Schizophr Bull.  1996;22 (1 ):15–25.8685657
11 Millgate  E, Hide  O, Lawrie  S, Murray  R, MacCabe  J, Kravariti  E. Neuropsychological differences between treatment-resistant and treatment-responsive schizophrenia: A meta-analysis. Psychol Med.  2021;52 (1 ):1–13.
12 Iasevoli  F, Giordano  S, Balletta  R, et al  Treatment resistant schizophrenia is associated with the worst community functioning among severely-ill highly-disabling psychiatric conditions and is the most relevant predictor of poorer achievements in functional milestones. Prog Neuro-Psychopharmacol Biol Psychiatry. 2016;65 :34–48.
13 Kennedy  J, Altar  C, Taylor  D, Degtiar  I, Hornberger  J. The social and economic burden of treatment-resistant schizophrenia. Int Clin Psychopharmacol.  2014;29 (2 ):63–76.23995856
14 Brain  C, Kymes  S, DiBenedetti  D, Brevig  T, Velligan  D. Experiences, attitudes, and perceptions of caregivers of individuals with treatment-resistant schizophrenia: A qualitative study. BMC Psychiatry. 2018;18 (1 ):1–13.29304757
15 Howes  O, McCutcheon  R, Agid  O, et al  Treatment-resistant schizophrenia: Treatment response and resistance in psychosis (TRRIP) working group consensus guidelines on diagnosis and terminology. Am J Psychiatry. 2017;174 (3 ):216–229.27919182
16 Kane  JM, Agid  O, Baldwin  ML, et al  Clinical guidance on the identification and management of treatment-resistant schizophrenia. J Clin Psychiatry.  2019;80 (2 ):2783.
17 Marcus  SC, Zummo  J, Pettit  AR, Stoddard  J, Doshi  JA. Antipsychotic adherence and rehospitalization in schizophrenia patients receiving oral versus long-acting injectable antipsychotics following hospital discharge. J Manag Care Spec Pharm. 2015;21 (9 ):754–769.26308223
18 Mehta  UM, Ibrahim  FA, Sharma  MS, et al  Resting-state functional connectivity predictors of treatment response in schizophrenia—A systematic review and meta-analysis. Schizophr Res.  2021;237 :153–165.34534947
19 Nucifora  FC, Woznica  E, Lee  BJ, Cascella  N, Sawa  A. Treatment resistant schizophrenia: Clinical, biological, and therapeutic perspectives. Neurobiol Dis.  2019;131 :104257.30170114
20 Seppälä  A, Pylvänäinen  J, Lehtiniemi  H, et al  Predictors of response to pharmacological treatments in treatment-resistant schizophrenia—A systematic review and meta-analysis. Schizophr Res.  2021;236 :123–134.34496316
21 Siskind  D, Siskind  V, Kisely  S. Clozapine response rates among people with treatment-resistant schizophrenia: Data from a systematic review and meta-analysis. Can J Psychiatry. 2017;62 (11 ):772–777.28655284
22 Shah  P, Iwata  Y, Brown  EE, et al  Clozapine response trajectories and predictors of non-response in treatment-resistant schizophrenia: A chart review study. Eur Arch Psychiatry Clin Neurosci.  2019;270 (1 ):11–22.31428862
23 Chan  SK, Chan  HY, Honer  WG, et al  Predictors of treatment-resistant and clozapine-resistant schizophrenia: A 12-year follow-up study of first-episode schizophrenia-spectrum disorders. Schizophr Bull.  2020;47 (2 ):485–494.
24 Nucifora  FC, Mihaljevic  M, Lee  BJ, Sawa  A. Clozapine as a model for antipsychotic development. Neurotherapeutics. 2017;14 (3 ):750–761.28653280
25 Brennan  AM, Harris  AWF, Williams  LM. Functional dysconnectivity in schizophrenia and its relationship to neural synchrony. Expert Rev Neurother.  2013;13 (7 ):755–765.23898848
26 Caravaggio  F, Hahn  M, Nakajima  S, Gerretsen  P, Remington  G, Graff-Guerrero  A. Reduced insulin-receptor mediated modulation of striatal dopamine release by basal insulin as a possible contributing factor to hyperdopaminergia in schizophrenia. Med Hypotheses.  2015;85 (4 ):391–396.26118462
27 Veerman  S, Schulte  P, de Haan  L. The glutamate hypothesis: A pathogenic pathway from which pharmacological interventions have emerged. Pharmacopsychiatry. 2014;47 (04/05 ):121–130.25002292
28 Schwartz  TL, Sachdeva  S, Stahl  SM. Glutamate neurocircuitry: Theoretical underpinnings in schizophrenia. Front Pharmacol.  2012;3 :195.23189055
29 Tibrewal  P, Nair  PC, Gregory  KJ, et al  Does clozapine treat antipsychotic-induced behavioural supersensitivity through glutamate modulation within the striatum? . Mol Psychiatry. 2023;28 :1839–1842.36932159
30 de Bartolomeis  A, Vellucci  L, Barone  A, et al  Clozapine’s multiple cellular mechanisms: What do we know after more than fifty years? A systematic review and critical assessment of translational mechanisms relevant for innovative strategies in treatment-resistant schizophrenia. Pharm Ther. 2022;236 :108236.
31 Melone  M, Vitellaro-Zuccarello  L, Vallejo-Illarramendi  A, et al  The expression of glutamate transporter GLT-1 in the rat cerebral cortex is down-regulated by the antipsychotic drug clozapine. Mol Psychiatry. 2001;6 :380–386.11443521
32 Friston  K, Brown  HR, Siemerkus  J, Stephan  KE. The dysconnection hypothesis (2016). Schizophr Res.  2016;176 (2–3 ):83–94.27450778
33 Takeuchi  H, Siu  C, Remington  G, et al  Does relapse contribute to treatment resistance? Antipsychotic response in first- vs. second-episode schizophrenia. Neuropsychopharmacology. 2018;44 (6 ):1036–1042.30514883
34 Potkin  SG, Kane  JM, Correll  CU, et al  The neurobiology of treatment-resistant schizophrenia: Paths to antipsychotic resistance and a roadmap for future research. NPJ Schizophr.  2020;6 (1 ):1.31911624
35 Suzuki  T, Kanahara  N, Yamanaka  H, et al  Dopamine supersensitivity psychosis as a pivotal factor in treatment-resistant schizophrenia. Psychiatry Res.  2015;227 (2–3 ):278–282.25863824
36 Woodward  ND, Cascio  CJ. Resting-state functional connectivity in psychiatric disorders. JAMA Psychiatry. 2015;72 (8 ):743.26061674
37 Kottaram  A, Johnston  LA, Tian  Y, et al  Predicting individual improvement in schizophrenia symptom severity at 1-year follow-up: Comparison of connectomic, structural, and clinical predictors. Hum Brain Mapp.  2020;41 (12 ):3342–3357.32469448
38 Tarcijonas  G, Sarpal  DK. Neuroimaging markers of antipsychotic treatment response in schizophrenia: An overview of magnetic resonance imaging studies. Neurobiol Dis.  2019;131 :104209.29953933
39 Paul  S, Sharfman  N. Functional connectivity as a means to delineate differences between treatment-resistant and treatment-responsive schizophrenia. J Neurophysiol.  2016;116 (2 ):229–231.26763784
40 Chan  NK, Kim  J, Shah  P, et al  Resting-state functional connectivity in treatment response and resistance in schizophrenia: A systematic review. Schizophr Res.  2019;211 :10–20.31331784
41 Tuovinen  N, Hofer  A. Resting-state functional MRI in treatment-resistant schizophrenia. Front Neuroimaging. 2023;2 :1127508.37554635
42 Vercammen  A, Knegtering  H, den Boer  JA, Liemburg  EJ, Aleman  A. Auditory hallucinations in schizophrenia are associated with reduced functional connectivity of the temporo-parietal area. Biol Psychiatry.  2010;67 (10 ):912–918.20060103
43 Sarpal  DK, Robinson  DG, Fales  C, et al  Relationship between duration of untreated psychosis and intrinsic corticostriatal connectivity in patients with early phase schizophrenia. Neuropsychopharmacology. 2017;42 (11 ):2214–2221.28294137
44 Ganella  EP, Seguin  C, Pantelis  C, et al  Resting-state functional brain networks in first-episode psychosis: A 12-month follow-up study. Aust N Z J Psychiatry. 2018;52 (9 ):864–875.29806483
45 Wolf  ND, Sambataro  F, Vasic  N, et al  Dysconnectivity of multiple resting-state networks in patients with schizophrenia who have persistent auditory verbal hallucinations. J Psychiatry Neurosci. 2011;36 (6 ):366–374.21791169
46 McNabb  CB, Tait  RJ, McIlwain  ME, et al  Functional network dysconnectivity as a biomarker of treatment resistance in schizophrenia. Schizophr Res.  2018;195 :160–167.29042073
47 Lohmann  G, Ovadia-Caro  S, Jungehülsing  GJ, Margulies  DS, Villringer  A, Turner  R. Connectivity concordance mapping: A new tool for model-free analysis of fmri data of the human brain. Front Syst Neurosci.  2012;6 :13.22470320
48 Woods  SW . Chlorpromazine equivalent doses for the newer atypical antipsychotics. J Clin Psychiatry.  2003;64 (6 ):663–667.12823080
49 Kay  SR, Fiszbein  A, Opler  LA. The Positive and Negative Syndrome Scale (PANSS) for schizophrenia. Schizophr Bull.  1987;13 (2 ):261–276.3616518
50 Tustison  NJ, Gee  J. N4ITK: Nick's N3 ITK implementation for MRI bias field correction. IEEE Trans Med Imaging. 2010;29 (6 ):1310–1320.20378467
51 Avants  B, Tustison  NJ, Song  G. Advanced normalization tools: V1.0. Insight Journal. 2009;2 (365 ):1–35.
52 Coupe  P, Yger  P, Prima  S, Hellier  P, Kervrann  C, Barillot  C. An optimized blockwise nonlocal means denoising filter for 3-D magnetic resonance images. IEEE Trans Med Imaging. 2008;27 (4 ):425–441.18390341
53 Sadananthan  SA, Zheng  W, Chee  MWL, Zagorodnov  V. Skull stripping using graph cuts. NeuroImage. 2010;49 (1 ):225–239.19732839
54 Avants  BB, Tustison  NJ, Song  G, Cook  PA, Klein  A, Gee  JC. A reproducible evaluation of ants similarity metric performance in brain image registration. NeuroImage. 2011;54 (3 ):2033–2044.20851191
55 Avants  BB, Yushkevich  P, Pluta  J, et al  The optimal template effect in hippocampus studies of diseased populations. NeuroImage. 2010;49 (3 ):2457–2466.19818860
56 Tustison  NJ, Avants  BB. Explicit B-spline regularization in diffeomorphic image registration. Front Neuroinform.  2013;7 :39.24409140
57 Tang  Y, Hojatkashani  C, Dinov  ID, et al  The construction of a Chinese MRI brain atlas: A morphometric comparison study between Chinese and Caucasian cohorts. NeuroImage. 2010;51 (1 ):33–41.20152910
58 Lemieux  L, Salek-Haddadi  A, Lund  TE, Laufs  H, Carmichael  D. Modelling large motion events in fmri studies of patients with epilepsy. Magn Reson Imaging.  2007;25 (6 ):894–901.17490845
59 Kendall  MG, Smith  BB. The problem of m rankings. Ann Math Statist. 1939;10 (3 ):275–287.
60 Alonso-Solís  A, Vives-Gilabert  Y, Grasa  E, et al  Resting-state functional connectivity alterations in the default network of schizophrenia patients with persistent auditory verbal hallucinations. Schizophr Res.  2015;161 (2–3 ):261–268.25468173
61 McNabb  CB, Sundram  F, Soosay  I, Kydd  RR, Russell  BR. Increased sensorimotor network connectivity associated with clozapine eligibility in people with schizophrenia. Psychiatry Res Neuroimaging. 2018;275 :36–42.29650266
62 Ganella  EP, Bartholomeusz  CF, Seguin  C, et al  Functional brain networks in treatment-resistant schizophrenia. Schizophr Res.  2017;184 :73–81.28011131
63 Blessing  EM, Murty  VP, Zeng  B, Wang  J, Davachi  L, Goff  DC. Anterior hippocampal–cortical functional connectivity distinguishes antipsychotic naïve first-episode psychosis patients from controls and may predict response to second-generation antipsychotic treatment. Schizophr Bull.  2019;46 (3 ):680–689.
64 Kraguljac  NV, White  DM, Hadley  N, et al  Aberrant hippocampal connectivity in unmedicated patients with schizophrenia and effects of antipsychotic medication: A longitudinal resting state functional MRI study. Schizophr Bull.  2016;42 (4 ):1046–1055.26873890
65 Oppenheimer  SM, Gelb  A, Girvin  JP, Hachinski  VC. Cardiovascular effects of human insular cortex stimulation. Neurology. 1992;42 (9 ):1727.1513461
66 Critchley  HD . Neural mechanisms of autonomic, affective, and cognitive integration. J Comp Neurol.  2005;493 (1 ):154–166.16254997
67 Paulus  MP, Khalsa  SS. When you don’t feel right inside: Homeostatic dysregulation and the mid-insular cortex in psychiatric disorders. Am J Psychiatry. 2021;178 (8 ):683–685.34383566
68 Pritchard  TC, Macaluso  DA, Eslinger  PJ. Taste perception in patients with insular cortex lesions. Behav Neurosci.  1999;113 (4 ):663–671.10495075
69 Avery  JA, Liu  AG, Ingeholm  JE, Riddell  CD, Gotts  SJ, Martin  A. Taste quality representation in the human brain. J Neurosci.  2019;40 (5 ):1042–1052.31836661
70 Blenkmann  AO, Collavini  S, Lubell  J, et al  Auditory deviance detection in the human insula: An intracranial EEG study. Cortex. 2019;121 :189–200.31629197
71 Zhang  Y, Zhou  W, Wang  S, et al  The roles of subdivisions of human insula in emotion perception and auditory processing. Cerebral Cortex. 2018;29 (2 ):517–528.
72 Bushara  KO, Hanakawa  T, Immisch  I, Toma  K, Kansaku  K, Hallett  M. Neural correlates of cross-modal binding. Nat Neurosci.  2002;6 (2 ):190–195.
73 Bastuji  H, Frot  M, Perchet  C, Hagiwara  K, Garcia-Larrea  L. Convergence of sensory and limbic noxious input into the anterior insula and the emergence of pain from nociception. Sci Rep.  2018;8 (1 ):13360.30190593
74 Allen  M . Unravelling the neurobiology of interoceptive inference. Trends Cogn Sci (Regul Ed).  2020;24 (4 ):265–266.
75 Gehrlach  DA, Dolensek  N, Klein  AS, et al  Aversive state processing in the posterior insular cortex. Nat Neurosci.  2019;22 (9 ):1424–1437.31455886
76 Livneh  Y, Sugden  AU, Madara  JC, et al  Estimation of current and future physiological states in insular cortex. Neuron. 2020;105 (6 ):1094–1111.e10.31955944
77 Craig  AD . How do you feel—Now? The anterior insula and human awareness. Nat Rev Neurosci. 2009;10 (1 ):59–70.19096369
78 Farrer  C, Frith  CD. Experiencing oneself vs another person as being the cause of an action: The neural correlates of the experience of agency. NeuroImage. 2002;15 (3 ):596–603.11848702
79 Park  H-D, Blanke  O. Coupling inner and outer body for self-consciousness. Trends Cogn Sci (Regul Ed).  2019;23 (5 ):377–388.
80 Singer  T . The neuronal basis and ontogeny of empathy and mind reading: Review of literature and implications for future research. Neurosci Biobehav Rev. 2006;30 (6 ):855–863.16904182
81 Lamm  C, Singer  T. The role of anterior insular cortex in social emotions. Brain Struct Funct. 2010;214 (5–6 ):579–591.20428887
82 Uddin  LQ, Nomi  JS, Hébert-Seropian  B, Ghaziri  J, Boucher  O. Structure and function of the human insula. J Clin Neurophysiol.  2017;34 (4 ):300–306.28644199
83 Menon  V, Uddin  LQ. Saliency, switching, attention and control: A network model of Insula function. Brain Struct Funct. 2010;214 (5–6 ):655–667.20512370
84 Smith  DM, Grant  B, Fisher  DJ, Borracci  G, Labelle  A, Knott  VJ. Auditory verbal hallucinations in schizophrenia correlate with P50 gating. Clin Neurophysiol.  2013;124 (7 ):1329–1335.23490551
85 Thoma  RJ, Meier  A, Houck  J, et al  Diminished auditory sensory gating during active auditory verbal hallucinations. Schizophr Res.  2017;188 :125–131.28109666
86 Yao  B, Thakkar  K. Interoception abnormalities in schizophrenia: A review of preliminary evidence and an integration with Bayesian accounts of psychosis. Neurosci Biobehav Rev. 2022;132 :757–773.34823914
87 Molent  C, Olivo  D, Wolf  RC, Balestrieri  M, Sambataro  F. Functional neuroimaging in treatment resistant schizophrenia: A systematic review. Neurosci Biobehav Rev. 2019;104 :178–190.31276716
88 Tian  Y, Zalesky  A, Bousman  C, Everall  I, Pantelis  C. Insula functional connectivity in schizophrenia: Subregions, gradients, and symptoms. Biol Psychiatry Cogn Neurosci Neuroimaging. 2019;4 (4 ):399–408.30691966
89 Heckers  S, Konradi  C. Hippocampal neurons in schizophrenia. J Neural Transm.  2002;109 (5–6 ):891–905.12111476
90 Heckers  S, Konradi  C. Hippocampal pathology in schizophrenia. Curr Top Behav Neurosci. 2010;4 :529–553.21312412
91 Tamminga  CA, Stan  AD, Wagner  AD. The hippocampal formation in schizophrenia. Am J Psychiatry. 2010;167 (10 ):1178–1193.20810471
92 Heckers  S, Konradi  C. GABAergic mechanisms of hippocampal hyperactivity in schizophrenia. Schizophr Res.  2015;167 (1–3 ):4–11.25449711
93 Lieberman  JA, Girgis  RR, Brucato  G, et al  Hippocampal dysfunction in the pathophysiology of schizophrenia: A selective review and hypothesis for early detection and intervention. Mol Psychiatry.  2018;23 (8 ):1764–1772.29311665
94 Roeske  MJ, Konradi  C, Heckers  S, Lewis  AS. Hippocampal volume and hippocampal neuron density, number and size in schizophrenia: A systematic review and meta-analysis of postmortem studies. Mol Psychiatry.  2020;26 (7 ):3524–3535.32724199
95 Huang  J, Zhu  Y, Fan  F, et al  Hippocampus and cognitive domain deficits in treatment-resistant schizophrenia: A comparison with matched treatment-responsive patients and healthy controls. Psychiatry Res Neuroimaging. 2020;297 :111043.32062167
96 Lodge  DJ, Grace  AA. Aberrant hippocampal activity underlies the dopamine dysregulation in an animal model of schizophrenia. J Neurosci. 2007;27 (42 ):11424–11430.17942737
97 Rosenquist  PB, Youssef  NA, Surya  S, McCall  WV. When all else fails: The use of electroconvulsive therapy for conditions other than major depressive episode. Psychiatr Clin North Am. 2018;41 (3 ):355–371.30098650
98 Grover  S, Sahoo  S, Rabha  A, Koirala  R. ECT in schizophrenia: A review of the evidence. Acta Neuropsychiatr.  2018;31 (03 ):115–127.30501675
99 Sinclair  DJM, Zhao  S, Qi  F, Nyakyoma  K, Kwong  JSW, Adams  CE. Electroconvulsive therapy for treatment-resistant schizophrenia. Cochrane Database of Systematic Reviews. 2019;3 .
100 Moon  S-Y, Kim  M, Lho  SK, Oh  S, Kim  SH, Kwon  JS. Systematic review of the neural effect of electroconvulsive therapy in patients with schizophrenia: Hippocampus and insula as the key regions of modulation. Psychiatry Investig.  2021;18 (6 ):486–499.
101 Mesulam  M-M, Mufson  EJ. Insula of the old world monkey. Architectonics in the insulo-orbito-temporal component of the paralimbic brain. J Comp Neurol.  1982;212 (1 ):1–22.7174905
102 Nakamura  K, Kubota  K. The primate temporal pole: Its putative role in object recognition and memory. Behav Brain Res.  1996;77 (1–2 ):53–77.8762159
103 Crespo-Facorro  B, Nopoulos  PC, Chemerinski  E, Kim  J-J, Andreasen  NC, Magnotta  V. Temporal pole morphology and psychopathology in males with schizophrenia. Psychiatry Res Neuroimaging. 2004;132 (2 ):107–115.
104 Goswami  S, Beniwal  RP, Kumar  M, et al  A preliminary study to investigate resting state fmri as a potential group differentiator for schizophrenia. Asian J Psychiatr.  2020;52 :102095.32339919
105 Rabinowitz  J, Werbeloff  N, Caers  I, et al  Determinants of antipsychotic response in schizophrenia. J Clin Psychiatry.  2014;75 (04 ):e308–e316.24813414
106 Abi-Dargham  A, Horga  G. The search for imaging biomarkers in psychiatric disorders. Nat Med.  2016;22 (11 ):1248–1255.27783066
107 Bzdok  D, Meyer-Lindenberg  A. Machine learning for precision psychiatry: Opportunities and challenges. Biol Psychiatry Cogn Neurosci Neuroimaging. 2018;3 (3 ):223–230.29486863
