Brain structural and functional dissociated patterns in schizophrenia

A total of 200 right-handed subjects participated in the study, including 98 patients with schizophrenia and 102 healthy controls. The patients were recruited from Tianjin Aning Hospital and Tianjn Anding Hospital and the healthy controls were recruited from the hospital staff or community residents. Patient diagnoses and illness duration were determined by the consensus of two senior psychiatrists using the Structured Interview for DSM-IV Axis I Disorder (SCID). All healthy controls were screened using the non-patient edition of the SCID to confirm a lifetime absence of psychiatric illnesses. Healthy controls were interviewed to exclude individuals with a known history of psychiatric illness in first-degree relatives. Exclusion criteria for all subjects were a history of head trauma with consciousness disturbances lasting more than 5 min, a history of substance abuse, pregnancy, or any physical illness, such as cardiovascular disease and neurological disorders, as diagnosed by interview or a review of their medical records. A professional radiologist assessed the MRIs slice-by-slice, and three patients and nine healthy controls were excluded owing to poor image quality. Consequently, the results for 95 patients with schizophrenia and 93 controls were analyzed (Fig. 1). The Positive and Negative Syndrome Scale (PANSS) was used to assess the severity of patients’ psychotic symptoms. The Medical Research Ethics Committee of Tianjin Medical University General Hospital approved this study. Written informed consent was gained from each participant following a complete description of the study.

MRI was performed using a 3.0-Tesla MR system (Discovery MR750, General Electric, Milwaukee, WI, USA). Tight, but comfortable, foam padding was used to minimize head motion, and earplugs were used to reduce scanner noise. Sagittal 3D T1-weighted images were acquired using a brain volume sequence with the following parameters: repetition time (TR) = 8.2 ms; echo time (TE) = 3.2 ms; inversion time (TI) = 450 ms; flip angle (FA) = 12°; field of view (FOV) = 256 mm × 256 mm; matrix = 256 × 256; slice thickness = 1 mm, no gap; and 188 sagittal slices). Resting-state fMRI data were acquired using a gradient-echo single-short Echo planar imaging sequence with the following parameters: TR/TE = 2000/45 ms; FOV = 220 mm × 220 mm; matrix = 64 × 64; FA = 90°; slice thickness = 4 mm; gap = 0.5 mm; 32 interleaved transverse slices; and 180 volumes. All subjects were requested to keep their eyes closed, relax, move as little as possible, think of nothing in particular, and not fall asleep during the scanning period.

Resting-state fMRI data were preprocessed using Statistical Parametric Mapping software (SPM8; http://​www.​fil.​ion.​ucl.​ac.​uk/​spm). The first 10 volumes for each subject were discarded to allow the signal to reach equilibrium and the participants to adapt to the scanning noise. The remaining volumes were corrected for the acquisition time delay between slices. Realignment was then performed to correct for motion between time points. All subjects’ fMRI data were within defined motion thresholds (i.e., translational or rotational motion parameters less than 2 mm or 2°). We also calculated frame-wise displacement, which indexes volume-to-volume changes in head position. Several nuisance covariates (six motion parameters, their first time derivations, and average BOLD signals of the ventricular and white matter) were regressed from the data. Recent study have reported that the signal spike caused by head motion significantly contaminates the final resting-state fMRI results even after regressing out the six motion parameters [34]. Therefore, we further regressed out spike volumes when the frame-wise displacement of the specific volume exceeded 0.5. The datasets were then band-pass filtered in a frequency range of 0.01 to 0.08 Hz [35]. In the normalization step, individual structural images were linearly co-registered with the mean functional image, and then linearly co-registered to the Montreal Neurological Institute (MNI) space. Finally, each filtered functional volume was spatially normalized to the MNI space using the co-registration parameters and were resampled into a 3-mm cubic voxel.

The GMV of each voxel was calculated using SPM8 (http://​www.​fil.​ion.​ucl.​ac.​uk/​spm/​software/​spm8/​). Structural MR images were segmented into gray matter (GM), white matter and cerebrospinal fluid using the standard unified segmentation model. After an initial affine registration of the GM concentration map into the MNI space, GM concentration images were nonlinearly warped using the diffeomorphic anatomical registration through exponentiated lie algebra technique, and the results resampled to a voxel size of 3 mm × 3 mm × 3 mm. The GMV of each voxel was obtained by multiplying the GM concentration map by the non-linear determinants derived from the spatial normalization step. Finally, the GMV images were smoothed using a Gaussian kernel of 6 mm × 6 mm × 6 mm full-width at half maximum. After spatial preprocessing, the smoothed GMV maps were used for statistical analyses.

The FCD of each voxel was calculated using an in-house Linux script according to the method described by Tomasi and Volkow [23‐26]. Pearson’s linear correlation evaluated the strength of the functional connectivity between voxels [36, 37]. According to the most of the previous studies, voxel pairs with a correlation coefficient of R > 0.6 were considered significantly connected, thereafter, in current study, we also adopted R > 0.6 as the cut-off point. [23‐26, 38‐40], FCD calculations were restricted to the cerebral gray matter mask regions. The gFCD at a given voxel, x₀, was computed as the global number of functional connections, k(x₀), between x₀ and all other voxels. This calculation was repeated for all x₀ voxels in the brain. The FCD maps were spatially smoothed using a 6 mm × 6 mm × 6 mm Gaussian kernel to minimize differences in the functional brain anatomy across subjects.

Group differences in GMV and gFCD were compared in a voxel-wise manner using a two-sample t-test with age and sex as nuisance variables. Multiple comparisons were corrected using a false discovery rate method with a corrected threshold of P < 0.05.

The demographic and clinical characteristics of the subjects are summarized in Table 1. There were no significant group differences in sex (χ2 = 1.35, P = 0.25) or age (t = 0.48, P = 0.63). Eighty-seven patients received medication during the MRI examinations, with the remaining eight patients being treatment naïve. The mean antipsychotic dose (cholorpromazin equivalents) was 446.5 ± 341.6 mg/d for the patients with schizophrenia. The mean duration of illness was 121.4 ± 92.8 months. Mean scores on the PANSS positive, negative, and general psychopathology subscales were 17.1 ± 7.9, 20.3 ± 9.1, and 34.1 ± 10.8, respectively.

Our results revealed that the strongest gFCD hubs were consistently located in the precuneus, inferior parietal lobe, superior temporal gyrus, medial prefrontal cortex, and dorsal lateral prefrontal cortex in our two groups. Most of the gFCD hubs are located in the default mode network and sensory cortices (Fig. 2). The distribution of the gFCD hubs did not differ between the two groups and was also consistent with the results of Tomasi et al. [24].

Voxel-wise analysis showed increased gFCD in the sub-cortical and limbic system areas (including the bilateral putamen, caudate, pallidum, and thalami), the hippocampal/parahippocampal complex, and the supplementary motor area (SMA) (Fig. 3a, Table 2). Decreased gFCD was seen in the posterior cortices, including the bilateral postcentral gyri, precentral gyri, right calcarine sulcus, and the left inferior occipital gyrus lobule in patients with schizophrenia (Fig. 3b, Table 2). ROI-based analyses showed that the differences in gFCD between schizophrenia patients and controls persisted even further regress out the GMV effect, suggesting that gFCD alterations in these ROIs are relatively independent characteristics of schizophrenia.

Decreased GMV was seen only in the bilateral frontal, parietal, temporal, occipital lobule, cingulate cortex, insular, striatum and thalamus in patients with schizophrenia (Fig. 3c and d and Table 3).

As previously mentioned, increased gFCD were primarily located in subcortical regions, including the basal ganglia and some components of the limbic system and SMA. Decreased gFCD were primarily located in the posterior cortical regions. However, in this study, only decreased GMV was observed. Notably, the aberrant patterns of gFCD and GMV differ between regions. In the caudate, thalami and hippocampus complex (HPC), gFCD increased but GMV decreased, while in the posterior cortices, decreased gFCD accompanied decreased GMV. These findings indicated a complex association between gFCD aberrances and GMV changes in schizophrenia. To further investigate whether gFCD correlates with GMV changes, we also performed correlation. In multiple regions, we found that although gFCD and GMV were both affected, there was no statistical correlation between gFCD and GMV either in patients with schizophrenia patients or controls (Fig. 3).

We did not find any statistical correlation between gFCD or GMV and antipsychotic dose, illness duration, or any PANSS scores (negative, positive, general psychopathology, or total scores).