Background: An imbalance in the inflammatory tumor necrosis factor system, including soluble tumor necrosis factor receptor 2 (sTNFR2), may contribute to the pathophysiology of schizophrenia. Methods: We measured the plasma levels of sTNFR2 in 256 healthy controls and 250 patients with schizophrenia including antipsychotic drug-free patients and treatment-resistant patients. We also explored the possible association between plasma sTNFR2 levels and cognitive performance in healthy controls and patients with schizophrenia using the Wechsler Adult Intelligence Scale, Third Edition, the Wechsler Memory Scale-Revised, and the Rey Auditory Verbal Learning Test. An association between plasma sTNFR2 levels and hippocampal volume in controls and patients with schizophrenia was also investigated via MRI. Received: August 9, 2017; Revised: January 24, 2018; Accepted: February 23, 2018 © The Author(s) 2018. Published by Oxford University Press on behalf of CINP. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http:// creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, 631 provided the original work is properly cited. For commercial re-use, please contact firstname.lastname@example.org Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/631/4907974 by Ed 'DeepDyve' Gillespie user on 03 July 2018 632 | International Journal of Neuropsychopharmacology, 2018 Significance Statement The tumor necrosis factor (TNF) system, including soluble tumor necrosis factor receptor 2 (sTNFR2), might be involved in the pathophysiology of schizophrenia. We found that the plasma levels of sTNFR2 were significantly higher in patients with schizo- phrenia, including antipsychotic drug-free patients. As far as we know, this is the first study to demonstrate that plasma sTNFR2 levels are significantly increased in antipsychotic drugs-free patients with schizophrenia. sTNFR2 are reported to negatively affect both hippocampal structure and cognitive functioning. However, the correlations between sTNFR2 levels and hippocampal structure and cognitive functioning have not been thoroughly investigated in patients with schizophrenia. We found a significant negative association between plasma sTNFR2 levels and cognitive performance in controls and patients with schizophrenia. Hippocampal volume was also negatively associated with plasma sTNFR2 levels in patients with schizophrenia. Results: We found that the plasma levels of sTNFR2 were significantly higher in patients with schizophrenia, including both antipsychotic drug-free patients and treatment-resistant patients. We found a significant negative association between plasma sTNFR2 levels and cognitive performance in controls and patients with schizophrenia. Hippocampal volume was also negatively associated with plasma sTNFR2 levels in patients with schizophrenia. Conclusion: Together, these convergent data suggest a possible biological mechanism for schizophrenia, whereby increased sTNFR2 levels are associated with a smaller hippocampal volume and cognitive impairment. Keywords: soluble tumor necrosis factor receptor 2, cognitive performance, hippocampal volume Introduction Increasing evidence has indicated that inflammation may play in mood disorder patients (Doganavsargil-Baysal et al., 2013; a critical role in the pathophysiology of schizophrenia (Monji Bobińska et al., 2017). However, the correlations between per - et al., 2009; Dean et al., 2013; Na et al., 2014; Hung et al., 2017). ipheral sTNFR2 levels and hippocampal structure and cognitive Tumor necrosis factor-α (TNF-α) is a proinflammatory cytokine functioning have not been thoroughly investigated in patients involved in the innate immune response. TNF-α and the down- with schizophrenia. stream TNF-α signaling pathways of 2 distinct membrane-bound We hypothesized that sTNFR2 are elevated as compensa- receptors, soluble tumor necrosis factor receptor 1 (sTNFR1) and tory response to proinflammatory imbalance in the TNF sys- soluble tumor necrosis factor receptor 2 (sTNFR2), play crit- tem and play protective roles to prevent hippocampal volume ical roles in brain function, including neurotransmission, syn- loss in schizophrenia. The aim of this study was to investigate apse formation, and neurogenesis (McCoy and Tansey, 2008). the plasma levels of sTNFR2 in patients with schizophrenia in sTNFR1 and sTNFR2 are differentially expressed and controlled; a large Japanese cohort that included antipsychotic drug-free sTNFR1 is historically thought to promote apoptosis and cyto- patients and treatment-resistant patients. We further investi- toxicity, while cell survival, proliferation, and protective cellular gated the possible association between plasma levels of sTNFR2 responses induced by TNF-α are attributed primarily to sTNFR2 and hippocampal volume and cognitive functioning in patients (Montgomery and Bowers, 2012). Depletion of sTNFR2, but not with schizophrenia. sTNFR1, has been reported to lead to hippocampal neuron death through different signal transduction pathways than those of Methods sTNFR1 (Yang et al., 2002), suggesting that sTNFR2 plays pro- tective roles in hippocampal neurons. Protective roles of sTNFR2 Subjects signaling were also reported in individuals with neurodegenera- tive disorders, including Alzheimer’s disease (Shen et al., 1997). A total of 250 patients with schizophrenia and 256 healthy Several studies have demonstrated that patients with schizo- controls were included in this study. The patients and healthy phrenia have increased peripheral levels of sTNFR2 (Coelho controls were age and gender matched (Table). 1 Blood sam- et al., 2008; Noto et al., 2013; Hoseth et al., 2017). We also previ- ples were collected, and plasma was used for the analysis. Most ously reported that patients with schizophrenia have increased cases were recruited at the Osaka University Hospital. One set of levels of plasma sTNFR2 (Yamamori et al., 2016). Together, these Osaka samples of 40 patients with schizophrenia and 40 healthy findings indicate that sTNFR2 might be involved in the patho- controls are the same samples used in the previous study that physiology of schizophrenia. showed a significant increase of plasma sTNFR2 in patients Many studies have reported that patients with schizophre- with schizophrenia (Yamamori et al., 2016). In all, 80 patients nia demonstrate smaller hippocampal volumes (Okada et al., with schizophrenia and 80 healthy controls were recruited at 2016) and cognitive decline (Hashimoto et al., 2013 Fujino et ; al., Chiba University Hospital and Tokushima University Hospital. 2014, 2016; Ohi et al., 2015) than those without. Elevated per - Each subject had been diagnosed and assessed by at least 2 ipheral proinflammatory cytokines are reported to negatively trained psychiatrists according to the Diagnostic and Statistical affect both hippocampal structure and cognitive function- Manual of Mental Disorders, Fourth Edition (DSM-IV), criteria ing. Higher peripheral sTNFR2 levels were also reported to be based on a structured clinical interview for DSM-IV axis I dis- associated with smaller hippocampal volume in a community orders (SCID-I). Symptoms of schizophrenia were assessed with sample (Schmidt et al., 2016), as well as with poorer cognitive the Positive and Negative Syndrome Scale (PANSS). The total of test scores in a community-living aged population (Gao et al., prescribed antipsychotics was calculated using chlorpromazine 2016), in Parkinson’s disease patients (Rocha et al., 2014), and (CPZ) equivalents (mg/d) (Inada and Inagaki, 2015). Patients with Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/631/4907974 by Ed 'DeepDyve' Gillespie user on 03 July 2018 Kudo et al. | 633 Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/631/4907974 by Ed 'DeepDyve' Gillespie user on 03 July 2018 Table 1. Demographic Variables of Subjects Treatment-Resistant Variables Osaka-1 Osaka-2 Osaka-3 Drug Free SCZ Tokushima Chiba Total HC SCZ HC SCZ HC SCZ HC SCZ HC SCZ HC SCZ HC SCZ HC SCZ n = 35 n = 32 n = 40 n = 39 n = 40 n = 40 n = 39 n = 37 n = 22 n = 22 n = 40 n = 40 n = 40 n = 40 n = 256 n = 250 Age (y) 54.5 ± 5.9 57.3 ± 11.9 37.8 ± 12.2 38 ± 12.3 42.3 ± 12.2 41.4 ± 11.8 35 ± 12 35.5 ± 13.1 38.1 ± 12.9 39.1 ± 13.3 52 ± 7.2 55.2 ± 6.1 32.4 ± 6.2 31.9 ± 5.7 41.8 ± 12.7 42.5 ± 14 Gender 15/20 11/21 20/20 20/19 20/20 20/20 23/16 23/14 12/10 12/10 20/20 20/20 20/20 20/20 130/126 126/124 (male/ female) Age of onset 36.1 ± 15.2 26.6 ± 11.1 N/A 26.6 ± 11.8 21.8 ± 8.4 N/A 24.8 ± 5.1 27.1 ± 11.4 Duration of 20.5 ± 15.1 11.4 ± 9.9 N/A 9.5 ± 7.4 17.3 ± 11.2 N/A 7.1 ± 4.3 12.3 ± 10.8 illness CPZ dose 608.8 ± 505.4 562.6 ± 398.3 N/A 0 993.2 ± 236.2 N/A N/A 680.4 ± 442 (mg) PANSS 18.4 ± 7.3 17.1 ± 5.6 N/A 20.1 ± 6.5 23 ± 4.4 N/A N/A 19.3 ± 6.4 positive PANSS 17.5 ± 7.4 19.6 ± 5.6 N/A 18.4 ± 6.8 25.3 ± 5.5 N/A N/A 19.8 ± 6.8 negative PANSS 36.8 ± 12.7 42.2 ± 13.1 N/A 41.4 ± 14.6 53.8 ± 9.3 N/A N/A 42.8 ± 13.8 general PANSS total 72.7 ± 25.5 79 ± 23.1 N/A 79.9 ± 25.8 102.1 ± 17.9 N/A N/A 81.9 ± 25.3 Abbreviations: CPZ, chlorpromadine; HC, healthy control; NA, not available; PANSS, Positive and Negative Syndrome Scale; SCZ, schizophrenia. The mean ± SD is shown. 634 | International Journal of Neuropsychopharmacology, 2018 comorbidities of substance-related disorders or mental retard- Original T1-weighted images were checked by visual inspec- ation were excluded. The “drug-free” patients took no antip- tion. We excluded images with insufficient brain coverage (field- sychotics for over 1 week. “Treatment-resistant” was defined of-view problem), those with low signal-to-noise ratios or any according to the following criteria mentioned in CPZ drug infor - artifacts (e.g., motion artifacts and magnetic susceptibility arti- mation in JAPAN: (1) non- or little response to treatment from facts), and those with any abnormal organic findings (e.g., large at least 2 adequately dosed antipsychotic trials for at least 4 cerebellar cysts and cavum septum pellucidum). weeks (including at least 1 second-generation antipsychotic, T1-weighted imaging data of 153 healthy subjects and 82 >600 mg/d of CPZ equivalent) and patients never had the Global patients with schizophrenia were processed with FreeSurfer Assessment of Functioning scores >40. (2) Intolerance to at 5.3 software (http://surfer.nmr.mgh.harvard.edu), as described least 2 second-generation antipsychotics because of uncon- previously (van Erp et al., 2016). Through this procedure, we trolled extrapyramidal symptoms. All subjects included in this obtained images of subcortical segmentation and regional vol- study met the criteria of non- or little response. Controls were umes (for the bilateral lateral ventricles, thalamus, caudate, recruited through local advertisements. Psychiatrically, medic- putamen, globus pallidus, hippocampus, amygdala, accumbens, ally, and neurologically healthy controls were evaluated via the and the ICV), and left hippocampal volume data were used to DSM-IV structured clinical interview, nonpatient version that assess the reproducibility of the association with plasma levels included the interview whether the subject has physical prob- of sTNFR2. lems. Subjects were excluded if they had neurological or med- ical conditions that could potentially affect the central nervous Assessment of Cognitive Functions system, such as an atypical headache, head trauma with loss Cognitive function was assessed with the Wechsler Adult of consciousness, chronic lung disease, kidney disease, chronic Intelligence Scale, Third Edition (WAIS-III) (Wechsler, 1997; hepatic disease, thyroid disease, active stage cancer, cerebro- Committee, 2006), the Wechsler Memory Scale-Revised (WMS- vascular disease, epilepsy, or seizures. Neurological and med- R) (Wechsler, 1987), and the Rey Auditory Verbal Learning Test ical conditions were assessed mainly by interview. In the case (AVLT). The numbers of subjects who underwent testing with of patients, blood tests were also conducted to confirm med- the WAIS-III, WMS-R, and AVLT are shown in Table 3. ical conditions. Written informed consent was obtained from all subjects after the procedures had been fully explained. This study was conducted in accordance with the World Medical Statistical Analysis Association’s Declaration of Helsinki and was approved by The statistical analysis was performed with SPSS 20.0 soft- the Research Ethical Committee of Osaka University, Chiba ware for Windows (SPSS Japan Inc). Variables were expressed as University and Tokushima University. the mean± SD. Data normality was assessed via Kolmogorov- Smirnov and Shapiro-Wilk tests, and the plasma sTNFR2 lev- Determination of Plasma sTNFR2 Levels els did not show a Gaussian distribution. Differences in clinical characteristics between patients and controls were analyzed Plasma levels of sTNFR2 were determined with Quantikine using χ tests for categorical variables and the Mann–Whitney human ELISA kits from R&D Systems. According to the manu- U-test for continuous variables. The groups did not differ with facturer’s instructions in the ELISA kits, the intra-assay and respect to age or gender (P > .05). Differences in the plasma lev- inter-assay CVs for the sTNFR2 ELISAs, as measured in 20 els of sTNFR2, cognitive and memory functions, and hippocam- plasma samples, averaged 3.5% and 4.1%, respectively. The min- pal volume between the patients and controls were analyzed imum detectable dose of human sTNF2 was 0.6 pg/mL, and all via the Mann-Whitney U-test. Correlations were examined with values obtained from our plasma samples were above the mini- Spearman’s rank correlation. Potential confounders of plasma mum detectable dose. All plasma samples were sent to Chiba sTNFR2 (age, sex, smoking status) were investigated with non- University, and the plasma levels of sTNFR2 were measured in parametric tests (Mann-Whitney U-test and Spearman’s rank Chiba University. correlation), and the P values of these confounders were <.05. These confounders were controlled using partial correlations. Imaging Processing In the analysis of patients with schizophrenia, CPZ dose was All MRI data were sent to National Institute for Physiological also controlled. The Bonferroni correction was used for multiple Sciences, and FreeSurfer analysis was done in National Institute comparison corrections. Differences were considered statisti- for Physiological Sciences. Among the subjects, 167 healthy cally significant for P < .05. subjects and 94 patients with schizophrenia underwent MRI. The MR images were processed with SPM12 (Wellcome MRI data were obtained using one of the following 3 scan- Department of imaging Neuroscience, University College ners: (1) 1.5T GE Signa EXCITE system; (2) 3.0T GE Signa HDxt; London, UK, http://www.fil.ion.ucl.ac.uk/spm) in MATLAB R2015b (3) 3.0T GE MR750. A 3-dimensional volumetric acquisition of a (The Mathworks, Natick, MA, USA) for tissue segmentation and T1-weighted gradient echo sequence produced a gapless series anatomical normalization using the diffeomorphic anatomi- of sagittal sections using a spoiled gradient-recalled acquisition cal registration through an exponentiated lie algebra (DARTEL) in the steady-state sequence. The sequence parameters for each algorithm. Custom DARTEL templates were generated from all scanner were as follows: (1) 1.5T GE Signa EXCITE: number of participants’ gray matter and white matter images. After spa- slices, 124; TE/TR, 4.2/12.6 ms; flip angle, 15°; acquisition matrix, tial normalization, the gray matter images were modulated to 256 × 256; 1NEX, FOV, 24 × 24 cm; and slice thickness, 1.4 mm; (2) preserve the volume, followed by smoothing with an 8-mm full 3.0T GE Signa HDxt: number of slices, 172; TE/TR, 2.9/7.2 ms; flip width at half maximum Gaussian kernel. For this preprocessing, angle, 11°; acquisition matrix, 256 × 256; 1NEX, FOV, 24 × 24 cm; default parameters were used. In addition, the total intracranial and slice thickness, 1 mm; and (3) 3.0T GE MR 750: number of volume was calculated by summing up the gray matter volume, slices, 156; TE/TR, 3.1/8.2 ms; flip angle, 11°; acquisition matrix, white matter volume, and cerebrospinal fluid volume using 256 × 256; 1NEX, FOV, 24 × 24 cm; and slice thickness, 1.2 mm. the “Tissue Volumes” function of SPM12. To account for several Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/631/4907974 by Ed 'DeepDyve' Gillespie user on 03 July 2018 Kudo et al. | 635 covariates, we used a multiple regression model and explored sTNFR2 and Cognitive and Memory Function the correlation between the serum level of sTNFR and gray mat- Cognitive function was investigated using the WAIS-III, and ter volume, treating age, gender, and total intracranial volume memory function was investigated using the WMS-R and AVLT. as confounding covariates. Since we used images from multi- Cognitive and memory functions were compared between the ple scanners, scanner site was modeled as a factor as previous schizophrenia patient group and the control group. Data of cog- studies employed (Stonnington et al., 2008). However, since we nitive and memory functions were available in 3 sets of samples were interested in the correlation between sTNFR2 and grey from Osaka, 1 set of antipsychotic drug-free patient samples matter, we modified the design matrix so that we could explore from Osaka, and 1 set of treatment-resistant schizophrenia the correlation accounting for the scanner difference (supple- samples from Osaka. In all categories of cognitive and memory mentary Figure 1). Detail is explained in the supplementary functions, scores were significantly lower in patients than in Information. Beside whole brain analysis, we performed small controls (supplementary Table 1b). When we see the difference volume correction, since we were interested in cognitive func- in each drug-free, treatment-resistant, and remaining patients tion, especially memory. For this analysis, we generated a mask with schizophrenia (OSAKA 1–3), and healthy controls, signifi- of bilateral hippocampi using the Neuromorphometrics atlas cant decline in schizophrenia was observed in all sets of sam- implemented in SPM12. The significance threshold was also set ples (supplementary Table 1a). to family-wise error corrected P value < .05. At the same time, The relationship between plasma sTNFR2 levels and cogni- we extracted the volume of whole right and left hippocampus tive function was investigated via the WAIS-III. Plasma sTNFR2 using “get_totals” script by Ged Ridgway (http://www0.cs.ucl. levels were significantly and negatively correlated with the WAIS ac.uk/staff/g.ridgway/vbm/get_totals.m_x0001). We employed Verbal IQ (VIQ), Performance IQ (PIQ), Full-scale IQ (FIQ) (Fig. 2a), the Neuromorphometric atlas implemented in SPM12 to define Picture Completion Block Design and Matrix Reasoning (PO), and both left and right hippocampus. Block Design and Matrix Reasoning (PS) in all samples (patients with schizophrenia and healthy controls) after Bonferroni cor - Results –3 rection (Table 3 , VIQ; r = -0.19, P = 2.4 × 10 , corrected P = .031, PIQ; –4 –4 r = -0.21, P = 8.1 × 10 , corrected P = .011, FIQ; r = -0.22,P = 3.5 × 10 , Plasma Levels of sTNFR2 –3 –3 corrected P = 4.5 × 10 , PO; r = -0.2, P = 1.5 × 10 , corrected P = .019, –3 PS; r = -0.2, P = 1.0 × 10 , corrected P = .013). No significant correla- The plasma levels of sTNFR2 were compared between the tion was observed when correlations between sTNFR2 levels and schizophrenia patient group (total n = 250) and the contr ol cognitive functions assessed via the WAIS-III were investigated group (total n = 256). Seven sets of samples were investigated, in patients with schizophrenia or in healthy controls (Tabl ).e T 3 o including 3 sets of samples from Osaka, 2 sets of samples from exclude the effects of age, gender, and smoking status, we con- Tokushima and Chiba, 1 set of antipsychotic drug-free patient ducted a partial correlation analysis between the plasma sTNFR2 samples from Osaka, and 1 set of treatment-resistant schizo- levels and cognitive function assessed via WAIS-III using age, phrenia samples from Osaka (Table ). 1 All sample sets were age gender, and smoking status as control variables. In patients with and gender matched. In all sets of samples, the plasma levels schizophrenia, CPZ dose was also controlled. Plasma sTNFR2 of sTNFR2 were significantly higher in patients than in controls levels were significantly and negatively correlated with VIQ, PIQ, –16 (Figure 1; Table 2, Mann-Whitney U-test; z = -8.21, P = 2.3 × 10 ). FIQ, and all 4 secondary indices of the WAIS-III (VC; Vocabulary, Similarities and Information, PO, WM; Arithmetic, Digit Span and Letter Number Sequencing, and PS) in all samples (patients with schizophrenia and healthy controls) as shown in supple- mentary Table 2. Then, we see the correlation in each drug-free, treatment-resistant, and remaining patients with schizophrenia (OSAKA 1–3) and healthy controls. Significant and negative cor - relation was observed between plasma sTNFR2 levels and FIQ in the remaining patients with schizophrenia and healthy controls –3 (supplementary Table 3a, FIQ; r = -0.23,P = 2.8 × 10 , corrected P = .04) in all samples (patients with schizophrenia and healthy controls). We also investigated the relationship between the plasma sTNFR2 levels and memory functions using the WMS-R and AVLT. Plasma sTNFR2 levels were significantly and negatively correlated with the WMS-R Verbal Memory (VerM), Visual Memory (ViM), and General Memory (GM) subscales (Figure 2b), as well as Delayed recall (DR) and total recall in the AVLT in all samples (patients with schizophrenia and healthy controls) after –5 Bonferroni correction (Table ; 3 VerM; r = -0.26, P = 1.3 × 10 , cor- –4 –4 – rected P = 1.7 × 10 , ViM; r = -0.22, P = 2.2 × 10 , corrected P = 2.8 × 10 3 –6 –5 , GM; r = -0.28, P = 2.2 × 10 , corrected P = 2.9 × 10 , DR; r = -0.26, –5 –4 –9 P = 2.2 × 10 , corrected P = 2.9 × 10 , AVLT; r = -0.36, P = 2.4 × 10 , –8 corrected P = 3.2 × 10 ). No significant correlations were observed Figure 1. The plasma levels of sTNFR2 were measured in healthy controls (HC) between sTNFR2 levels and memory function assessed via the (n = 256) and schizophrenia (SCZ) patients (n = 250). The bars represent mean val- ues ± SD. ***P < .001 by Mann-Whitney U-test. WMS-R and AVLT in patients with schizophrenia. In healthy Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/631/4907974 by Ed 'DeepDyve' Gillespie user on 03 July 2018 636 | International Journal of Neuropsychopharmacology, 2018 Table 2. sTNFR2 Expression in Plasma Treatment Osaka-1 Osaka-2 Osaka-3 Drug free resistant SCZ Tokushima Chiba Total Variables (35vs32) (40vs39) (40vs40) (39vs37) (22vs22) (40vs40) (40vs40) (256vs250) HC (ng/mL) 1.88 ± 0.44 1.68 ± 0.42 1.81 ± 0.37 1.93 ± 0.45 2.01 ± 0.66 2.50 ± 0.64 1.98 ± 0.44 1.97 ± 0.54 SCZ (ng/mL) 2.65 ± 0.89 1.96 ± 0.54 2.46 ± 0.68 2.28 ± 0.66 2.99 ± 0.94 3.57 ± 1.25 2.27 ± 0.60 2.57 ± 0.96 –4 –7 –4 –6 –16 P value 1.9 × 10 *** .014* 5.1 × 10 *** .017* 4.7 × 10 *** 2.5 × 10 *** .019* 2.3 × 10 *** Effect size 1.70 0.73 1.29 1.07 2.11 2.68 0.94 1.68 Abbreviations: HC, healthy control; SCZ, schizophrenia. *P < .05, **P < .01, ***P < .005. The mean ± SD is shown. Figure 2. Correlation of WAIS Full-Scale IQ (FIQ) (a), WMS General Memory (GM) (b), and plasma levels of sTNFR2. sTNFR2 was negatively correlated with both FIQ and GM. controls, plasma sTNFR2 levels were significantly and nega- sTNFR2 and Hippocampal Volume tively correlated with total recall of the AVLT (Tabl ; r e = 3 -0.32, –5 –4 Hippocampal volume was compared between the schizophre- P = 1.2 × 10 , corrected P = 1.6 × 10 ). To exclude the effects of age, nia patient group and the control group. Left and right hip- gender, and smoking status, we conducted a partial correlation pocampal volume was assessed by voxel-based morphometry analysis between the plasma sTNFR2 levels and memory func- (VBM) and FreeSurfer analysis. MRI data were available in 3 sets tions assessed via the WMS-R and AVLT using age, gender, and of samples from Osaka, 1 set of antipsychotic drug-free patient smoking status as control variables. In patients with schizophre- samples from Osaka, and 1 set of treatment-resistant schizo- nia, CPZ dose was also controlled. Plasma sTNFR2 levels were phrenia samples from Osaka. Both left and right hippocampal significantly and negatively correlated with VerM, ViM, GM, A/C volume were significantly smaller in patients than in controls (Attention and Concentration), DR, and total recall in the AVLT (supplementary T able 4). When we see the difference in each drug- in all samples (patients with schizophrenia and healthy con- free, treatment-resistant, and remaining patients with schizo- trols) as shown in supplementary Table 2. Plasma sTNFR2 levels phrenia and healthy controls, significant decrease of hippocampal were significantly and negatively correlated with total recall in volume in schizophrenia was observed in all sets of samples the AVLT only in healthy controls by partial correlation ana- (supplementary Table 4). lysis using age, gender, and smoking status as control variables Whole brain correlational analysis between sTNFR2 and grey (supplementary Table 2). Then, we see the correlation in each matter volume by VBM analysis revealed no significant region. drug-free, treatment-resistant, and remaining patients with However, small volume correction revealed negative correl- schizophrenia (OSAKA 1–3) and healthy controls. Significant ation between sTNFR2 and left hippocampus. (supplementary and negative correlation was observed between plasma sTNFR2 Figure 2, P = .031, family-wise error corrected for multiple com- levels and VerM, ViM, GM, DR, and total recall in the AVLT in parison, k = 28, x = -33, y = -26, z = -9). Since we found significant the remaining patients with schizophrenia and healthy con- correlation between STNFR2 and a cluster in left hippocampus, –4 trols (supplementary Table 3a, VerM; r = -0.27, P = 2.8 × 10 , cor- –3 –3 we investigated the relationships between plasma sTMFR2 lev- rected P = 3.7 × 10 , ViM; r = -0.24, P = 1.3 × 10 , corrected P = .02, els and left whole hippocampal volume. Plasma sTNFR2 levels –5 4 GM; r = -0.31, P = 3.2 × 10 , corrected P = 4.1 × 10–, DR; r = -0.27, –4 –3 –8 were significantly and negatively correlated with left whole P = 3.7 × 10 , corrected P = 4.8 × 10 , AVLT; r = -0.42, P = 1.9 × 10 , hippocampal volume (patients with schizophrenia and healthy –7 corrected P = 2.5 × 10 ). Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/631/4907974 by Ed 'DeepDyve' Gillespie user on 03 July 2018 Kudo et al. | 637 controls) (Figure 3a, total; n = 261, r = -0.14, P = .02). Anatomical negatively correlated with hippocampal volume in all sam- atlases from which the signal was extracted are shown in ples (patients with schizophrenia and healthy controls) and in Figure 3b. Plasma sTNFR2 levels remained significantly and patients with schizophrenia using a partial correlation analysis –5 negatively correlated with left hippocampal volume measured (total; n = 239, r = -0.26, P = 4.8 × 10 , SCZ; n = 71, r = -0.27, P = .02). by VBM analysis only in patients with schizophrenia (SCZ; n = 94, To confirm the negative correlation between the plasma r = -0.22, P = .03). No significant correlation was observed between sTNFR2 levels and hippocampal volume, FreeSurfer analysis sTNFR2 levels and left hippocampal volume in healthy controls was also conducted. A significant negative correlation between (HC; n = 167, r = -0.04, P = .61). To exclude the effects of age, gender, plasma sTNFR2 levels and left hippocampal volume measured and smoking status, we conducted a partial correlation analysis by FreeSurfer analysis was observed in all samples (patients with between the plasma sTNFR2 levels and left hippocampal vol- schizophrenia and healthy controls) (total; n = 235, r= -0.16, P = .016, ume using age, gender, and smoking status as control variables. SCZ; n = 82, r = -0.19, P = .09, HC; n = 153, r = -0.04, P = .63). A partial In patients with schizophrenia, CPZ dose was also controlled. correlation using age, gender, and smoking status as control vari- –4 Plasma sTNFR2 levels were still found to be significantly and ables did not change the results (total; n = 214, r= -0.24, P = 4.3 × 10 ). Table 3. Correlation between Plasma sTNFR2 Levels and Cognitive Function Total Schizophrenia Healthy Control Corrected Corrected Corrected n r P P value n r P P value n r P P value –3 IQ WAIS VIQ 263 -0.19 2.4 × 10 ** .031* 87 0.14 n.s. n.s. 176 -0.19 .012* n.s. –4 PIQ 263 -0.21 8.1 × 10 *** .011* 87 0.04 n.s. n.s. 176 -0.13 n.s. n.s. –4 –3 FIQ 263 -0.22 3.5 × 10 *** 4.5 × 10 *** 87 0.09 n.s. n.s. 176 -0.19 .012* n.s. VC 263 -0.16 .011* n.s. 87 0.16 n.s. n.s. 176 -0.19 .014* n.s. –3 PO 263 -0.2 1.5 × 10 *** .019* 87 0.03 n.s. n.s. 176 -0.16 .036* n.s. –3 WM 260 -0.16 9.5 × 10 ** n.s. 84 0.08 n.s. n.s. 176 -0.15 .049* n.s. –3 PS 260 -0.2 1.0 × 10 *** .013* 84 0.01 n.s. n.s. 176 -0.1 n.s. n.s. –5 –4 Memory WMS VerM 267 -0.26 1.3 × 10 *** 1.7 × 10 *** 91 -0.02 n.s. n.s. 176 -0.15 .049* n.s. –4 –3 ViM 267 -0.22 2.2 × 10 *** 2.8 × 10 *** 91 -0.13 n.s. n.s. 176 -0.13 n.s. n.s. –6 –5 GM 267 -0.28 2.2 × 10 *** 2.9 × 10 *** 91 -0.06 n.s. n.s. 176 -0.16 .03* n.s. A/C 267 -0.15 .018* n.s. 91 0.02 n.s. n.s. 176 -0.06 n.s. n.s. –5 –4 DR 267 -0.26 2.2 × 10 *** 2.9 × 10 *** 91 -0.11 n.s. n.s. 176 -0.11 n.s. n.s. –9 –8 –5 –4 AVLT total 257 -0.36 2.4 × 10 *** 3.2 × 10 *** 81 -0.23 .03* n.s. 176 -0.32 1.2 × 10 *** 1.6 × 10 *** recall *P < .05, **P < .01, ***P < .005 Abbreviations: A/C, Attention and Concentration; DR, delayed recall; FIQ, Full-scale IQ; GM, General Memory; n.s., not significant; VIQ, Verbal IQ; PIQ, Performance IQ; PO, Picture Completion Block Design and Matrix Reasoning; PS, Block Design and Matrix Reasoning; VerM, Verbal Memory; ViM, Visual Memory; WM, Airthmetic, Digit Span and Letter Number Sequencing. Figure 3. Correlation of the left hippocampus volume and plasma levels of sTNFR2 (a). sTNFR2 was negatively correlated with the left hippocampus volume. Anatomical atlas from which the signal was extracted are shown in (b). Y axis shows the volume of left hippocampus (mL). Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/631/4907974 by Ed 'DeepDyve' Gillespie user on 03 July 2018 638 | International Journal of Neuropsychopharmacology, 2018 and hippocampal volume in patients with schizophrenia at least Correlation Between Clinical Variables and Plasma in VBM analysis. Although we found significant negative correla- Level of sTNFR2 tion of sTNFR2 levels and cognition mainly in healthy controls, a We also observed the correlations between clinical variables significant correlation of sTNFR2 levels and hippocampal volume and plasma level of sTNFR2. PANSS positive subscale scores and was observed mainly in patients with schizophrenia, and we did duration of illness were positively and significantly correlated not detect a significant association in healthy controls only. The with plasma level of sTNFR2 (supplementary Table 5, PANSS- results showing discrepancy between cognition and brain vol- positive; r = 0.23, P = .01, duration of illness; r = 0.35, P = 7.9 × 10–5). ume. It is possible that sTNFR2 levels have a more direct effect PANSS scores were also negatively and significantly correlated on hippocampal volume than cognition in pathophysiology of with many of cognitive functions (supplementary Table 5). schizophrenia. Experimental data suggest that administration of TNF-α attenuate hippocampal growth (Seguin et al., 2009), whereas depletion of sTNFR2 lead to hippocampal neuron death Discussion (Yang et al., 2002). It is possible that sTNFR2 play protective roles In this study, we confirmed that the plasma levels of sTNFR2 in hippocampus against activation of TNF system that could were significantly increased in patients with schizophrenia in potentially contribute to a decrease in hippocampal volume. a large Japanese cohort that include antipsychotic drugs-free The results of this study should be evaluated in light of cer - patients and treatment-resistant patients. The involvement of tain limitations. First, plasma sampling conditions, such as the inflammatory-related pathways, including TNF-α-related path- time of sampling and diet, might modulate the plasma levels ways, in schizophrenia has been implied by many studies (Dean of investigated markers. However, these conditions were not et al., 2013). Our data showing higher plasma sTNFR2 levels in controlled for in this study. In addition, sTNFRs (sTNFR1 and patients with schizophrenia are consistent with the results of sTMFR2) have been reported to be correlated with body mass previous studies (Coelho et al., 2008; Noto et al., 2013; Hoseth index (BMI) (Hinze-Selch et al., 2000; Himmerich et al., 2006), and et al., 2017). Many studies have shown that proinflammatory this association indicates that BMI should be controlled for in cytokines including TNF-α are etiological factors for schizo- such studies. However, the BMI of our subjects was not available phrenia (Na et al., 2014). Considering the neuroprotective roles and was therefore not controlled for in this study. of sTNFR2, it is possible that high levels of sTNFR2 in schizo- The results that increased plasma sTNFR2 levels are associ- phrenia are the compensatory response to proinflammatory ated with a smaller hippocampal volume and cognitive impair - imbalance in the TNF system. To the best of our knowledge, ment in schizophrenia suggest that sTNFR2 might play an this is the first study to demonstrate that plasma sTNFR2 levels important role in the pathophysiology in schizophrenia. Further are significantly increased in antipsychotic drugs-free patients studies are needed to elucidate the mechanism how sTNFR2 in with schizophrenia, suggesting that higher levels of sTNFR2 in schizophrenia have effect on hippocampal volume loss and cog- patients with schizophrenia are not the result of medication. It nitive impairment. has been reported that sTNFR1, but not sTNFR2, levels are higher in treatment-resistant patients than in controls (Noto et al., Supplementary Material 2013). However, our results showed that sTNFR2 levels were also higher in treatment-resistant patients than in controls. Supplementary data are available at International Journal of It has been reported that higher peripheral levels of sTNFR2 Neuropsychopharmacology online. are associated with worse cognitive function in community sam- ples (Gao et al., 2016), Parkinson’s disease patients (Rocha et al., Funding 2014), and mood disorder patients (Doganavsargil-Baysal et al., 2013; Bobińska et al., 2017). The plasma TNF-α /sTNFRs (sTNFR1 This work was supported by a Grant-in-Aid for Scientific Research and sTNFR2) ratio was also reported to be significantly associ- (B) (25293250, 16H05375), a Grant-in-Aid for Scientific Research (C) ated with working memory assessed via the WAIS-III in patients (25461730), and a Grant-in-Aid for Young Scientists (B) (15K19726) with schizophrenia. However, an association between plasma from the Japan Society for the Promotion of Science; the Health sTNFR2 levels and cognitive function has not been investigated and Labour Sciences Research Grants for Comprehensive in patients with schizophrenia. We found that higher plasma Research on Persons with Disabilities from the Japan Agency for sTNFR2 levels were associated with worse cognitive function. Medical Research and Development; Brain Mapping by Integrated A significant association was observed mainly in all samples Neurotechnologies for Disease Studies from the Ministry of (patients with schizophrenia and healthy controls), and we did Education, Science, Sports, Culture and Technology in Japan. The not detect a significant association in patients with schizophre- funders had no role in the study design, data collection and ana- nia only. The effect of sTNFR2 on cognitive function might be the lysis, decision to publish, or preparation of the manuscript. indirect one in the pathophysiology of schizophrenia. Most sub- categories of both the WAIS and WMS-R were associated with Acknowledgments plasma sTNFR2 levels, suggesting that sTNFR2 is associated with general cognitive and memory functions rather than specific We thank all the individuals who participated in this study. functions. Higher peripheral sTNFR2 levels were reported to be associ- Statement of interest ated with smaller hippocampal volume in a community sample (Schmidt et al., 2016). However, an association between periph- None. eral sTNFR2 levels and hippocampal volume in patients with schizophrenia has not yet been investigated. An association References between the plasma levels of sTNFR1 and hippocampal volume was investigated in patients with schizophrenia, and no signifi- Bobińska K, Gałecka E, Szemraj J, Gałecki P, Talarowska M (2017) cant association was observed (Hoseth et al., 2016). We found the Is there a link between TNF gene expression and cognitive significant negative association between plasma level of sTNFR2 deficits in depression? Acta Biochim Pol 64:65–73. Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/631/4907974 by Ed 'DeepDyve' Gillespie user on 03 July 2018 Kudo et al. | 639 Coelho FM, Reis HJ, Nicolato R, Romano-Silva MA, Teixeira Monji A, Kato T, Kanba S (2009) Cytokines and schizophre- MM, Bauer ME, Teixeira AL (2008) Increased serum levels of nia: microglia hypothesis of schizophrenia. Psychiatry Clin inflammatory markers in chronic institutionalized patients Neurosci 63:257–265. with schizophrenia. Neuroimmunomodulation 15:140–144. Montgomery SL, Bowers WJ (2012) Tumor necrosis factor-alpha Committee JW-IP (2006) Japanese Wechsler Adult Intelligence and the roles it plays in homeostatic and degenerative pro- Scale. Tokyo: Nihon Bunka Kagakusya. cesses within the central nervous system. J Neuroimmune Dean B, Gibbons AS, Tawadros N, Brooks L, Everall IP, Scarr E Pharmacol 7:42–59. (2013) Different changes in cortical tumor necrosis factor-α- Na KS, Jung HY, Kim YK (2014) The role of pro-inflammatory related pathways in schizophrenia and mood disorders. Mol cytokines in the neuroinflammation and neurogenesis of Psychiatry 18:767–773. schizophrenia. Prog Neuropsychopharmacol Biol Psychiatry Doganavsargil-Baysal O, Cinemre B, Aksoy UM, Akbas H, Metin 48:277–286. O, Fettahoglu C, Gokmen Z, Davran F (2013) Levels of TNF-α, Noto C, Gadelha A, Belangero SI, Spindola LM, Rocha NP, de soluble TNF receptors (stnfr1, stnfr2), and cognition in bipolar Miranda AS, Teixeira AL, Cardoso Smith MA, de Jesus Mari J, disorder. Hum Psychopharmacol 28:160–167. Bressan RA, Brietzke E (2013) Circulating levels of stnfr1 as a Fujino H, Sumiyoshi C, Sumiyoshi T, Yasuda Y, Yamamori H, Ohi marker of severe clinical course in schizophrenia. J Psychiatr K, Fujimoto M, Umeda-Yano S, Higuchi A, Hibi Y, Matsuura Y, Res 47:467–471. Hashimoto R, Takeda M, Imura O (2014) Performance on the Ohi K, Hashimoto R, Ikeda M, Yamamori H, Yasuda Y, Fujimoto M, wechsler adult intelligence scale-III in Japanese patients with Umeda-Yano S, Fukunaga M, Fujino H, Watanabe Y, Iwase M, schizophrenia. Psychiatry Clin Neurosci 68:534–541. Kazui H, Iwata N, Weinberger DR, Takeda M (2015) Glutamate Fujino H, et al. for COCORO (2017) Estimated cognitive decline in networks implicate cognitive impairments in schizophrenia: patients with schizophrenia: a multicenter study. Psychiatry genome-wide association studies of 52 cognitive phenotypes. Clin Neurosci 71:294–300. Schizophr Bull 41:909–918. Gao Q, Camous X, Lu YX, Lim ML, Larbi A, Ng TP (2016) Novel Okada N, et al. (2016) Abnormal asymmetries in subcortical inflammatory markers associated with cognitive perform- brain volume in schizophrenia. Mol Psychiatry 21:1460–1466. ance: Singapore longitudinal ageing studies. Neurobiol Aging Rocha NP, Teixeira AL, Scalzo PL, Barbosa IG, de Sousa MS, Morato IB, 39:140–146. Vieira EL, Christo PP, Palotás A, Reis HJ (2014) Plasma levels of sol- Hashimoto R, Ikeda M, Ohi K, Yasuda Y, Yamamori H, Fukumoto uble tumor necrosis factor receptors are associated with cogni- M, Umeda-Yano S, Dickinson D, Aleksic B, Iwase M, Kazui H, tive performance in Parkinson’s disease. Mov Disord 29:527–531. Ozaki N, Weinberger DR, Iwata N, Takeda M (2013) Genome- Schmidt MF, Freeman KB, Windham BG, Griswold ME, Kullo IJ, wide association study of cognitive decline in schizophrenia. Turner ST, Mosley TH Jr (2016) Associations between serum Am J Psychiatry 170:683–684. inflammatory markers and hippocampal volume in a com- Himmerich H, Fulda S, Linseisen J, Seiler H, Wolfram G, munity sample. J Am Geriatr Soc 64:1823–1829. Himmerich S, Gedrich K, Pollmächer T (2006) TNF-alpha, sol- Seguin JA, Brennan J, Mangano E, Hayley S (2009) Proinflammatory uble TNF receptor and interleukin-6 plasma levels in the gen- cytokines differentially influence adult hippocampal cell pro- eral population. Eur Cytokine Netw 17:196–201. liferation depending upon the route and chronicity of admin- Hinze-Selch D, Schuld A, Kraus T, Kuhn M, Uhr M, Haack M, istration. Neuropsychiatr Dis Treat 5:5–14. Pollmacher T (2000) Effects of antidepressants on weight and Shen Y, Li R, Shiosaki K (1997) Inhibition of p75 tumor necrosis on the plasma levels of leptin, TNF-alpha and soluble TNF factor receptor by antisense oligonucleotides increases hyp- receptors: a longitudinal study in patients treated with ami- oxic injury and beta-amyloid toxicity in human neuronal cell triptyline or paroxetine. Neuropsychopharmacology 23:13–19. line. J Bio Chem 272:3550–3553. Hoseth EZ, Westlye LT, Hope S, Dieset I, Aukrust P, Melle I, Stonnington CM, Tan G, Klöppel S, Chu C, Draganski B, Jack CR Haukvik UK, Agartz I, Ueland T, Ueland T, Andreassen OA Jr, Chen K, Ashburner J, Frackowiak RS (2008) Interpreting (2016) Association between cytokine levels, verbal memory scan data acquired from multiple scanners: a study with and hippocampus volume in psychotic disorders and healthy Alzheimer’s disease. Neuroimage 39:1180–1185. controls. Acta Psychiatr Scand 133:53–62. van Erp TG, et al (2016) Subcortical brain volume abnormalities Hoseth EZ, Ueland T, Dieset I, Birnbaum R, Shin JH, Kleinman JE, in 2028 individuals with schizophrenia and 2540 healthy con- Hyde TM, Mørch RH, Hope S, Lekva T, Abraityte AJ, Michelsen trols via the ENIGMA consortium. Mol Psychiatry 21:547–553. AE, Melle I, Westlye LT, Ueland T, Djurovic S, Aukrust P, Wechsler D (1987) WMS-R: Wechsler memory scale-revised. San Weinberger DR, Andreassen OA (2017) A study of TNF path- Antonio, TX: Psychological Corporation. way activation in schizophrenia and bipolar disorder in Wechsler D (1997) WAIS-III: Wechsler adult intelligence scale. plasma and brain tissue. Schizophr Bull 43:881–890. San Antonio, TX: Psychological Corporation. Hung YP, Wang CS, Yen CN, Chang HC, Chen PS, Lee IH, Chen Yamamori H, Ishima T, Yasuda Y, Fujimoto M, Kudo N, Ohi K, KC, Yang YK, Lu RB, Wang TY (2017) Role of cytokine changes Hashimoto K, Takeda M, Hashimoto R (2016) Assessment of a in clozapine-induced fever: a cohort prospective study. multi-assay biological diagnostic test for mood disorders in a Psychiatry Clin Neurosci 71:395–402. Japanese population. Neurosci Lett 612:167–171. Inada T, Inagaki A (2015) Psychotropic dose equivalence in Japan. Yang L, Lindholm K, Konishi Y, Li R, Shen Y (2002) Target Psychiatry Clin Neurosci 69:440–447. depletion of distinct tumor necrosis factor receptor sub- McCoy MK, Tansey MG (2008) TNF signaling inhibition in the types reveals hippocampal neuron death and survival CNS: implications for normal brain function and neurode- through different signal transduction pathways. J Neurosci generative disease. J Neuroinflammation 5:45. 22:3025–3032. Downloaded from https://academic.oup.com/ijnp/article-abstract/21/7/631/4907974 by Ed 'DeepDyve' Gillespie user on 03 July 2018
International Journal of Neuropsychopharmacology – Oxford University Press
Published: Feb 24, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera