Cerebrospinal Fluid Neuromarker Levels in Patients With Schizophrenia and Healthy Control Participants: A Proximity Extension Assay Study Authors: Shinsuke Hidese, Kotaro Hattori, Takako Enokida, Megumi Tatsumi, Ryo Matsumura, Hiroshi Kunugi Published: 2026 ABSTRACT AbstractBackgroundTo further evaluate the association between cerebrospinal fluid (CSF) neuromarkers and schizophrenia, we measured CSF neuromarker levels using proximity extension assay (PEA) in patients with schizophrenia and healthy control (HC) participants.MethodsParticipants included 97 patients with schizophrenia (mean age [SD] = 38.8 [10.3] years; 60 men and 37 women) and 71 HC participants (mean age [SD] = 46.9 [15.4] years; 42 men and 29 women). Symptoms in patients with schizophrenia were assessed using the Positive and Negative Syndrome Scale (PANSS). CSF protein levels were measured using the Olink PEA Neurology Panel.ResultsCSF levels of cadherin 3, FcRL2, granzyme A, NAAA, plexin-B1, and sFRP-3 were lower in patients with schizophrenia than in HC participants (p < .05). Levels of BCAN, beta-NGF, carboxypeptidases A2 and M, CD200, DDR1, GDF-8, MANF, MDGA1, NAAA, neprilysin, neuropilin-2, neutral ceramidase, Nr-CAM, NTRK2, NTRK3, PDGF-R-alpha, plexin-B3, RGMA, ROBO2, testican-1, THY1, and WFIKKN1 positively correlated with PANSS positive scores, whereas protogenin levels negatively correlated with PANSS negative and general psychopathology scores in patients with schizophrenia (p < .05).ConclusionsThis PEA study identified candidate CSF trait and state markers in schizophrenia; however, these findings require confirmation in further studies, as this was a primarily exploratory and hypothesis-generating study.Keywords: Biomarkers, Cerebrospinal fluid, Proteomics, Proximity extension assay, Schizophrenia, Symptom severity Evidence from clinical biomarkers across multiple domains has been accumulated to discover molecular signatures and develop therapeutic targets in schizophrenia (1). Cerebrospinal fluid (CSF) biomarker studies have been conducted to distinguish major psychiatric disorders from neurodegenerative disorders, including Alzheimer’s disease, Parkinson’s disease, dementia with Lewy bodies, frontotemporal dementia, and young-onset dementia (2, 3, 4). However, CSF biomarker research in schizophrenia remains limited, including in first-episode psychosis (5,6). A meta-review suggested that CSF evidence of microglial activation and neuroplastic changes is involved in the pathogenesis of acute psychosis and schizophrenia (7). CSF studies have also suggested blood-brain barrier pathology in schizophrenia (8). An individual patient data meta-analysis showed that the albumin CSF-to-serum ratio and other CSF alterations, as markers of blood-CSF barrier disruption, were associated with positive symptom severity in male patients with first-episode psychosis (9). Furthermore, inflammatory processes and folate metabolism were associated with the antipsychotic effect on monoamine metabolism in the CSF of patients with schizophrenia (10). A systematic review and meta-analysis found no significant differences in CSF neurofibrillary tangles or tau but lower CSF amyloid-β levels in patients with schizophrenia (11). However, in our previous multiplex immunoassay study with a large sample, we identified several candidate trait (including decreased CSF total tau) and state neuromarkers and found no difference in CSF amyloid-β1–42 levels between patients with schizophrenia and healthy control (HC) participants (12). Although amyloid-β and tau are representative CSF biomarkers in Alzheimer’s disease, this study focused on other neuromarkers due to the relatively young age of the patient cohort (mean approximately 39 years). CSF proteomics has been a useful tool to identify potential biomarker targets for psychiatric and neurodegenerative disorders such as schizophrenia, Parkinson’s disease, and Alzheimer’s disease, providing their unique and common molecular pathways (13). Multiplex immunoassays have been used to investigate CSF trait and state markers in major psychiatric disorders, and proximity extension assay (PEA) was developed to overcome the cross-reactivity and signal crosstalk limitations of multiplex immunoassays (14). Using the Olink Inflammation Panel and the PEA platform, one study found higher CSF glial cell line–derived neurotrophic factor levels in patients with first-episode schizophrenia (n = 29) than in HC participants (n = 44) (15), whereas another found lower CSF cluster of differentiation 5 (CD5) levels in patients with first-episode schizophrenia (n = 77) than in HC participants (n = 56) (16). However, PEA-based findings in schizophrenia remain scarce, underscoring the need for further studies using additional Olink panels. In such exploratory PEA studies, including ours, useful findings of additional diagnosis-related trait markers beyond glial cell line–derived neurotrophic factor (15) and CD5 (16), as well as symptom severity–related state markers, could be identified in the CSF of patients with schizophrenia. To further examine the association of CSF biomarkers related to neurodegeneration, neuroinflammation, and neurotrophic signaling (i.e., neuromarkers) with schizophrenia (12), we measured CSF neuromarker levels using the Olink PEA Neurology Panel in patients with schizophrenia and HC participants. We largely aimed to generate new data given that CSF analysis in schizophrenia with PEA assays is in its infancy. Therefore, in this study, we used a largely agnostic, hypothesis-free approach. Namely, the main value of this study is not hypothesis testing but rather opening a window into more CSF deep proteomic analysis in biological psychiatry. Methods and Materials Participants This study included 97 patients with schizophrenia (mean age [SD] = 38.8 [10.3] years; 60 men and 37 women) and 71 HC participants (mean age [SD] = 46.9 [15.4] years; 42 men and 29 women) matched for sex and ethnicity (Japanese). The sample size was not based on a priori power analysis because no prior effect size estimates were available. This was an exploratory sample largely to generate hypotheses given a paucity of prior work in this space. All participants were recruited at the National Center of Neurology and Psychiatry (NCNP) through advertisements at NCNP Hospital, on the institutional website, and in free local magazines. Participants were screened for psychiatric disorders by trained psychiatrists using the Japanese version of the Mini-International Neuropsychiatric Interview (MINI) (17,18). Consensus diagnoses were established according to DSM-5 (19), based on the MINI, additional unstructured interviews, and medical records, when available. HC participants had no history of contact with psychiatric services. Participants were excluded if they had a history of central nervous system (CNS) disease, severe head injury, substance abuse, or intellectual disability. Written informed consent was obtained from all participants after they received a description of the study protocol. The study protocol was approved by the Ethics Committee of the NCNP and the Teikyo University Ethical Review Board for Medical and Health Research Involving Human Subjects. This study was conducted in accordance with the Declaration of Helsinki (20). Clinical Assessments Symptoms in patients with schizophrenia were assessed using the Positive and Negative Syndrome Scale (PANSS) (21). Board-certified psychiatrists performed the assessments. Daily antipsychotic doses were converted to chlorpromazine-equivalent doses according to a published guideline (22). Lumbar Puncture CSF was collected as described in our previous studies (23, 24, 25) and was provided by the NCNP biobank, which is accredited under ISO 20387. After a single thaw on ice for sample set preparation, the samples were refrozen at −80 °C until Olink PEA was performed; thus, each sample underwent 2 freeze-thaw cycles. Olink PEA CSF samples were analyzed using the commercially available Olink Target 96 Neurology Panel, which was designed for protein biomarker discovery (Olink Proteomics). The Olink platform is based on the PEA technique, in which paired oligonucleotide-labeled antibodies bind to target proteins and generate a quantitative polymerase chain reaction readout (26). The Neurology Panel includes 92 proteins related to neurobiological processes, such as neural development, axon guidance, and synaptic function, as well as broader processes including cellular regulation, immunology, development, and metabolism. Samples from the patient and HC groups were randomly distributed across assay plates. Laboratory technicians were blinded to the clinical information. Initial preprocessing and quality control were performed by Olink using the NPX Manager, and values were exported on a log2 scale (26). Reference intra-assay and interassay coefficients of variation were set at <15% and <25%, respectively. Normalized protein expression (NPX) values were median centered for each protein, and batch-to-batch variation was corrected by bridging normalization between the 2 plates using 8 common CSF samples (7 from patients and 1 from an HC participant). Olink Insights Stat Analyzer statistical tests indicated that there was a statistically significant difference between the means of the 2 assay groups for 25 of 92 assays (Figure S1). Proteins for which >25% of values were below the limit of detection in both groups were excluded from further analyses. Of the 92 proteins, 83 were successfully measured, whereas the remaining 9 proteins—C-type lectin domain family 10 member A, C-type lectin domain family 1 member B, glial cell line–derived neurotrophic factor, granulocyte colony-stimulating factor, interleukin-5 receptor subunit alpha, kynureninase, latexin, linker for activation of T cells family member 1, and nicotinamide/nicotinic acid mononucleotide adenylyltransferase 1—did not meet the criteria for successful PEA measurement. Statistical Analyses Categorical variables (sex) and continuous variables (age and body mass index [BMI]) were compared in patients with schizophrenia and HC participants using χ2 tests and Student’s or Welch’s t tests, respectively. CSF neuromarker levels were compared between patients with schizophrenia and HC participants using multivariate analysis of covariance (MANCOVA), adjusting for age, sex, BMI, and psychotropic medication use. Correlations between CSF neuromarker levels and symptom scores were assessed using Pearson's partial correlation coefficients, adjusting for age, sex, BMI, and psychotropic medication use in patients with schizophrenia. Correlations between CSF neuromarker levels and continuous clinical variables were assessed using Pearson’s correlation coefficients. CSF neuromarker levels were also compared by sex and by medication status (drug free vs. medicated) using unpaired Student’s or Welch’s t tests. Correlation matrices for CSF neuromarker levels were also generated using Pearson’s correlation coefficients. Effect sizes were reported as ϕ for χ2 tests, Cohen’s d for unpaired t tests, and ηp2 for MANCOVAs. Normality and homoscedasticity of continuous variables were tested using Kolmogorov-Smirnov and Levene’s tests, respectively, and the results were used to justify parametric analyses. Bonferroni corrections for multiple testing were applied for group comparisons in t tests, MANCOVAs, and correlational analyses on CSF neuromarker levels successfully measured (p < .05/83 = .0006). All statistical tests were 2-tailed, and p < .05 was considered statistically significant. Statistical analyses were performed using IBM SPSS Statistics, version 30.0 (IBM Japan, Ltd.). Results Demographic and clinical characteristics of the study participants are presented in Table 1. Patients with schizophrenia were significantly younger and had a higher BMI than HC participants (p < .05). Analysis of covariance showed that CSF levels of sFRP-3 (secreted frizzled-related protein 3), cadherin 3, granzyme A, FcRL2 (Fc receptor-like 2), NAAA (N-acylethanolamine-hydrolyzing acid amidase), and plexin-B1 were significantly lower in patients with schizophrenia than in HC participants (uncorrected p < .05) (Figure 1). Table 1Demographic and Clinical Characteristics of the ParticipantsSchizophrenia, n = 97Healthy Control, n = 71Statistical ComparisonsAge, Years38.8 ± 10.3 [18–61]46.9 ± 15.4 [19–82]Welch's t114.2 = 3.84, p = .0002∗, Cohen's d = 0.64Sex, Male60 (61.9%)42 (59.2%)χ21 = 0.13, p = .72, ϕ = −0.027Body Mass Index24.5 ± 4.6 [15.2–46.7]22.4 ± 3.4 [16.9–32.5]Student's t116 = −3.34, p = .001∗, Cohen's d = −0.52Age of Onset, Years24.3 ± 9.1 [13–56]––Duration of Psychotropic Medication Use, Years12.3 ± 12.2 [0–47]––Chlorpromazine-equivalent Dose, mg/day665.2 ± 539.3 [0–2336.4]––Psychotropic Medication Free23 (23.7%)––PANSS Positive15.4 ± 5.6 [7–31]––PANSS Negative17.6 ± 5.6 [7–41]––PANSS General Psychopathology34.2 ± 8.7 [16–60]––PANSS Total67.2 ± 16.6 [37–115]––CSF Neuropilin-2 (NPX)6.36 ± 0.37 [5.47–7.14]6.52 ± 0.44 [5.33–7.39]F1,162 = 0.00, p = .98, ηp2 = 0.000CSF MAPT (NPX)2.69 ± 0.49 [1.60–3.90]2.91 ± 0.62 [1.79–4.50]F1,162 = 0.09, p = .76, ηp2 = 0.001CSF CADM3 (NPX)5.75 ± 0.17 [5.26–6.28]5.80 ± 0.15 [5.44–6.08]F1,162 = 0.00, p = .99, ηp2 = 0.000CSF UNC5C (NPX)3.46 ± 0.46 [2.21–4.66]3.60 ± 0.54 [2.05–4.72]F1,162 = 0.03, p = .86, ηp2 = 0.000CSF Brorin (NPX)4.43 ± 0.47 [3.03–5.28]4.76 ± 0.56 [3.36–5.62]F1,162 = 0.98, p = .32, ηp2 = 0.006CSF Siglec-9 (NPX)2.29 ± 0.39 [1.34–3.36]2.49 ± 0.42 [1.48–3.37]F1,162 = 0.62, p = .43, ηp2 = 0.004CSF CLM-6 (NPX)3.57 ± 0.50 [2.41–4.77]3.63 ± 0.59 [2.74–4.98]F1,162 = 0.48, p = .49, ηp2 = 0.003CSF Ezrin (NPX)3.20 ± 0.30 [2.66–3.89]3.22 ± 0.37 [2.43–4.35]F1,162 = 0.40, p = .84, ηp2 = 0.000CSF SMOC2 (NPX)6.23 ± 0.36 [5.30–7.02]6.39 ± 0.34 [5.58–7.13]F1,162 = 0.71, p = .40, ηp2 = 0.004CSF NBL1 (NPX)4.92 ± 0.17 [4.45–5.31]4.85 ± 0.21 [4.23–5.20]F1,162 = 0.36, p = .55, ηp2 = 0.002CSF Ephrin-A4 (NPX)1.69 ± 0.32 [1.01–2.78]1.81 ± 0.34 [1.15–2.54]F1,162 = 0.47, p = .49, ηp2 = 0.003CSF SCARB2 (NPX)5.17 ± 0.39 [4.24–5.97]5.41 ± 0.48 [4.51–6.30]F1,162 = 1.67, p = .20, ηp2 = 0.010CSF NCAN (NPX)9.38 ± 0.12 [8.96–9.68]9.37 ± 0.10 [9.08–9.60]F1,162 = 3.13, p = .08, ηp2 = 0.019CSF Protogenin (NPX)4.13 ± 0.29 [3.54–4.78]4.27 ± 0.35 [3.42–5.11]F1,162 = 0.89, p = .35, ηp2 = 0.005CSF ROBO2 (NPX)6.79 ± 0.56 [5.33–8.19]7.08 ± 0.61 [5.71–8.27]F1,162 = 0.04, p = .85, ηp2 = 0.000CSF CRTAM (NPX)0.85 ± 0.26 [0.19–1.38]0.97 ± 0.42 [021–3.57]F1,162 = 3.70, p = .056, ηp2 = 0.022CSF RGMA (NPX)10.68 ± 0.23 [10.15–11.20]10.75 ± 0.26 [10.08–11.29]F1,162 = 0.01, p = .94, ηp2 = 0.000CSF Plexin B3 (NPX)4.80 ± 0.47 [3.79–6.12]5.01 ± 0.56 [4.09–6.09]F1,162 = 0.02, p = .88, ηp2 = 0.000CSF Carboxypeptidase A2 (NPX)2.96 ± 0.52 [1.54–4.27]3.02 ± 0.52 [1.89–4.26]F1,162 = 0.47, p = .49, ηp2 = 0.003CSF CD38 (NPX)5.41 ± 0.81 [3.35–7.43]5.70 ± 0.92 [2.36–7.57]F1,162 = 1.36, p = .25, ηp2 = 0.008CSF SMPD1 (NPX)3.91 ± 0.39 [3.11–5.02]4.10 ± 0.48 [2.84–5.18]F1,162 = 0.03, p = .86, ηp2 = 0.000CSF MSR1 (NPX)3.71 ± 0.39 [2.88–4.92]3.86 ± 0.43 [2.95–4.94]F1,162 = 0.18, p = .67, ηp2 = 0.001CSF Alpha-2-MRAP (NPX)5.57 ± 0.39 [4.50–6.43]5.73 ± 0.40 [4.94–6.50]F1,162 = 0.13, p = .72, ηp2 = 0.001CSF sFRP-3 (NPX)6.80 ± 0.19 [5.77–7.28]6.85 ± 0.20 [5.77–7.16]F1,162 = 4.33, p = .039∗, ηp2 = 0.026CSF EPHB6 (NPX)7.25 ± 0.68 [5.41–8.86]7.62 ± 0.79 [6.24–9.12]F1,162 = 0.01, p = .94, ηp2 = 0.000CSF RGMB (NPX)7.96 ± 0.46 [6.75–8.87]8.21 ± 0.51 [7.11–9.07]F1,162 = 0.18, p = .67, ηp2 = 0.001CSF Sialoadhesin (NPX)2.61 ± 0.35 [1.78–3.31]2.67 ± 0.51 [1.77–4.41]F1,162 = 0.02, p = .67, ηp2 = 0.001CSF Contactin-5 (NPX)6.63 ± 0.50 [5.34–7.71]6.88 ± 0.54 [5.79–8.01]F1,162 = 0.04, p = .83, ηp2 = 0.000CSF ADAM 22 (NPX)9.43 ± 0.45 [8.44–10.78]9.64 ± 0.49 [8.48–10.82]F1,162 = 1.48, p = .23, ηp2 = 0.009CSF ADAM 23 (NPX)3.84 ± 0.31 [2.94–4.61]3.97 ± 0.35 [3.22–4.65]F1,162 = 0.43, p = .51, ηp2 = 0.003CSF Matrilin-3 (NPX)10.63 ± 0.47 [9.47–11.71]10.92 ± 0.53 [9.53–12.10]F1,162 = 1.96, p = .16, ηp2 = 0.012CSF R-spondin-1 (NPX)1.00 ± 0.26 [0.38–2.03]1.10 ± 0.26 [0.41–1.80]F1,162 = 0.07, p = .80, ηp2 = 0.000CSF HAGH (NPX)1.12 ± 0.28 [0.51–2.23]1.21 ± 0.31 [0.51–1.92]F1,162 = 1.36, p = .25, ηp2 = 0.008CSF Galectin-8 (NPX)4.70 ± 0.45 [3.44–5.71]4.81 ± 0.51 [3.54–5.92]F1,162 = 0.00, p = .97, ηp2 = 0.000CSF BCAN (NPX)9.90 ± 0.13 [9.58–10.25]9.94 ± 0.11 [9.61–10.22]F1,162 = 0.50, p = .48, ηp2 = 0.003CSF Layilin (NPX)10.27 ± 0.61 [9.00–11.70]10.63 ± 0.69 [8.97–11.94]F1,162 = 0.09, p = .77, ηp2 = 0.001CSF Neprilysin (NPX)2.65 ± 0.44 [1.56–3.94]2.63 ± 0.51 [1.33–3.82]F1,162 = 0.31, p = .58, ηp2 = 0.002CSF GDF-8 (NPX)2.20 ± 0.33 [1.50–3.01]2.28 ± 0.31 [1.45–2.88]F1,162 = 0.80, p = .37, ηp2 = 0.005CSF THY 1 (NPX)10.95 ± 0.10 [10.63–11.27]10.94 ± 0.10 [10.74–11.20]F1,162 = 0.02, p = .90, ηp2 = 0.000CSF WFIKKN1 (NPX)1.17 ± 0.22 [0.63–1.67]1.21 ± 0.19 [0.73–1.64]F1,162 = 2.89, p = .09, ηp2 = 0.018CSF TMPRSS5 (NPX)6.73 ± 0.62 [4.87–8.35]6.88 ± 0.53 [5.39–7.97]F1,162 = 0.10, p = .75, ηp2 = 0.001CSF Cadherin 3 (NPX)5.69 ± 0.24 [4.91–6.12]5.74 ± 0.29 [4.35–6.19]F1,162 = 7.13, p = .008∗, ηp2 = 0.042CSF GFR-alpha-1 (NPX)6.79 ± 0.40 [5.78–7.80]6.85 ± 0.39 [5.96–7.64]F1,162 = 0.35, p = .55, ηp2 = 0.002CSF GM-CSF-R-alpha (NPX)4.36 ± 0.75 [2.71–5.71]4.43 ± 0.81 [2.98–6.30]F1,162 = 0.47, p = .49, ηp2 = 0.003CSF Beta-NGF (NPX)2.26 ± 0.09 [2.08–2.57]2.25 ± 0.10 [1.98–2.57]F1,162 = 0.17, p = .68, ηp2 = 0.001CSF SCARA5 (NPX)2.69 ± 0.36 [1.50–3.69]2.87 ± 0.43 [1.82–3.74]F1,162 = 0.92, p = .34, ηp2 = 0.006CSF CD200 (NPX)11.70 ± 0.49 [10.26–12.92]11.92 ± 0.50 [10.77–12.95]F1,162 = 0.09, p = .76, ηp2 = 0.001CSF NTRK2 (NPX)6.50 ± 0.32 [5.67–7.18]6.64 ± 0.34 [5.89–7.36]F1,162 = 0.28, p = .60, ηp2 = 0.002CSF Granzyme A (NPX)4.98 ± 0.82 [0.47–6.31]5.23 ± 0.72 [3.59–8.68]F1,162 = 4.72, p = .031∗, ηp2 = 0.028CSF Draxin (NPX)9.51 ± 0.52 [8.26–10.58]9.69 ± 0.52 [7.91–10.56]F1,162 = 1.42, p = .24, ηp2 = 0.009CSF SCARF2 (NPX)1.53 ± 0.37 [0.61–2.43]1.64 ± 0.37 [0.77–2.41]F1,162 = 0.10, p = .75, ηp2 = 0.001CSF GDNFR-alpha-3 (NPX)5.89 ± 0.48 [4.33–6.99]6.02 ± 0.45 [4.89–6.85]F1,162 = 0.30, p = .59, ηp2 = 0.002CSF Poliovirus Receptor (NPX)6.13 ± 0.47 [4.98–7.42]6.32 ± 0.54 [5.21–7.58]F1,162 = 0.25, p = .62, ηp2 = 0.002CSF TNFRSF12A (NPX)3.24 ± 0.46 [1.94–4.63]3.45 ± 0.52 [2.34–4.80]F1,162 = 0.03, p = .86, ηp2 = 0.000CSF SKR3 (NPX)5.93 ± 0.41 [4.98–7.17]6.01 ± 0.45 [4.82–6.84]F1,162 = 0.48, p = .49, ηp2 = 0.003CSF FLRT2 (NPX)3.43 ± 0.32 [2.56–4.21]3.54 ± 0.39 [2.48–4.32]F1,162 = 0.12, p = .73, ηp2 = 0.001CSF Carboxypeptidase M (NPX)6.74 ± 0.53 [5.57–8.06]6.89 ± 0.59 [5.53–8.36]F1,162 = 0.00, p = .99, ηp2 = 0.000CSF Glypican-5 (NPX)6.10 ± 0.48 [4.88–7.09]6.35 ± 0.56 [4.87–7.57]F1,162 = 0.04, p = .84, ηp2 = 0.000CSF BMP-4 (NPX)4.43 ± 0.53 [3.27–6.16]4.59 ± 0.57 [3.08–5.65]F1,162 = 0.30, p = .58, ηp2 = 0.002CSF FcRL2 (NPX)1.06 ± 0.27 [0.46–1.74]1.21 ± 0.31 [0.53–1.94]F1,162 = 7.07, p = .009∗, ηp2 = 0.042CSF MDGA1 (NPX)5.34 ± 0.62 [4.02–6.88]5.70 ± 0.58 [4.32–6.77]F1,162 = 0.28, p = .60, ηp2 = 0.002CSF PDGF-R-alpha (NPX)5.38 ± 0.40 [4.54–6.18]5.48 ± 0.47 [4.43–6.70]F1,162 = 0.01, p = .95, ηp2 = 0.000CSF CTSC (NPX)5.51 ± 0.55 [4.33–6.88]5.63 ± 0.48 [3.97–6.55]F1,162 = 0.00, p = .98, ηp2 = 0.000CSF Cadherin-6 (NPX)1.75 ± 0.39 [0.92–2.60]1.95 ± 0.48 [1.10–6.55]F1,162 = 0.39, p = .53, ηp2 = 0.002CSF DDR1 (NPX)8.31 ± 0.42 [7.47–9.39]8.50 ± 0.44 [7.62–9.45]F1,162 = 0.57, p = .45, ηp2 = 0.004CSF JAM-B (NPX)12.52 ± 0.32 [11.88–13.38]12.60 ± 0.37 [11.68–13.24]F1,162 = 1.13, p = .72, ηp2 = 0.001CSF Cathepsin S (NPX)10.73 ± 0.29 [9.88–11.33]10.77 ± 0.29 [10.23–11.48]F1,162 = 0.07, p = .80, ηp2 = 0.000CSF Neutral Ceramidase (NPX)2.24 ± 0.39 [1.54–3.25]2.31 ± 0.40 [1.39–3.12]F1,162 = 0.02, p = .88, ηp2 = 0.000CSF NAAA (NPX)1.54 ± 0.32 [0.57–2.27]1.72 ± 0.39 [0.79–2.58]F1,162 = 6.20, p = .014∗, ηp2 = 0.037CSF N2DL-2 (NPX)4.89 ± 0.70 [3.34–7.41]5.06 ± 0.68 [3.36–6.84]F1,162 = 0.00, p = .95, ηp2 = 0.000CSF Plexin B1 (NPX)1.74 ± 0.29 [1.16–2.60]2.00 ± 0.38 [1.18–2.76]F1,162 = 6.50, p = .012∗, ηp2 = 0.039CSF TNFRSF21 (NPX)10.70 ± 0.19 [10.05–11.18]10.75 ± 0.16 [10.30–11.04]F1,162 = 1.36, p = .25, ηp2 = 0.008CSF CLM-1 (NPX)5.97 ± 0.94 [3.15–7.65]6.32 ± 0.92 [3.15–7.90]F1,162 = 1.95, p = .16, ηp2 = 0.012CSF Testican-1 (NPX)7.12 ± 0.36 [6.27–8.03]7.38 ± 0.35 [6.55–8.03]F1,162 = 0.03, p = .87, ηp2 = 0.000CSF Interleukin-12 (NPX)3.18 ± 0.40 [2.30–4.78]3.28 ± 1.01 [1.86–10.91]F1,162 = 1.04, p = .31, ηp2 = 0.006CSF Dkk-4 (NPX)2.19 ± 0.39 [1.10–3.10]2.35 ± 0.45 [1.48–3.32]F1,162 = 0.87, p = .35, ηp2 = 0.005CSF EDA2R (NPX)9.31 ± 0.49 [8.22–10.15]9.55 ± 0.63 [8.08–10.78]F1,162 = 0.15, p = .70, ηp2 = 0.001CSF NTRK3 (NPX)6.53 ± 0.34 [5.65–7.31]6.69 ± 0.39 [5.86–7.44]F1,162 = 0.04, p = .84, ηp2 = 0.000CSF LAIR-2 (NPX)7.98 ± 0.87 [3.47–9.64]8.45 ± 0.78 [6.94–9.96]F1,162 = 0.12, p = .73, ηp2 = 0.001CSF MANF (NPX)2.89 ± 0.33 [1.19–3.75]3.03 ± 0.41 [2.03–3.90]F1,162 = 0.63, p = .43, ηp2 = 0.004CSF Tenascin-R (NPX)6.43 ± 0.32 [5.76–7.14]6.56 ± 0.29 [5.88–7.12]F1,162 = 0.31, p = .58, ηp2 = 0.002CSF CD200R1 (NPX)1.64 ± 0.27 [0.91–2.50]1.67 ± 0.31 [0.88–2.80]F1,162 = 0.30, p = .59, ηp2 = 0.002CSF Nr-CAM (NPX)11.62 ± 0.10 [11.38–11.94]11.63 ± 0.09 [11.45–11.88]F1,162 = 0.89, p = .35, ηp2 = 0.005Values are presented as mean ± SD [range] or n (%).CSF, cerebrospinal fluid; NPX, normalized protein expression; PANSS, Positive and Negative Syndrome Scale.∗Nominally significant p values. Demographic and Clinical Characteristics of the Participants Values are presented as mean ± SD [range] or n (%). CSF, cerebrospinal fluid; NPX, normalized protein expression; PANSS, Positive and Negative Syndrome Scale. Nominally significant p values. Differences in CSF neuromarker levels between patients with schizophrenia and HC participants. CSF levels of sFRP-3 (A), cadherin 3 (B), granzyme A (C), FcRL2 (D), NAAA (E), and plexin-B1 (F) were lower in patients with schizophrenia than in HC participants (∗p < .05). CSF, cerebrospinal fluid; HC, healthy control. Correlations between CSF neuromarker levels and symptom scores in patients with schizophrenia are presented in Table 2. CSF levels of neuropilin-2, ROBO2 (roundabout guidance receptor 2), RGMA (repulsive guidance molecule A), plexin-B3, carboxypeptidase A2, BCAN (brevican core protein), neprilysin, GDF-8 (growth/differentiation factor 8), THY1 (Thy-1 membrane glycoprotein), WAP, Kazal, immunoglobulin, Kunitz, NTR domain-containing protein 1 (WFIKKN1), beta-NGF (beta–nerve growth factor), OX-2 membrane glycoprotein (CD200), BDNF/NT-3 growth factor receptor (NTRK2), carboxypeptidase M, MDGA1 (MAM domain-containing glycosylphosphatidylinositol anchor protein 1), platelet-derived growth factor receptor alpha (PDGF-R-alpha), epithelial DDR1 (discoidin domain-containing receptor 1), neutral ceramidase, NAAA, testican-1, NTRK3 (NT-3 growth factor receptor), MANF (mesencephalic astrocyte-derived neurotrophic factor), and Nr-CAM (neuronal cell adhesion molecule) were positively correlated with PANSS positive scores (uncorrected p < .05) (Figure 2), whereas CSF protogenin levels were negatively correlated with PANSS negative and general psychopathology scores (uncorrected p < .05) (Figure 3). Table 2Correlations Between CSF Neuromarker Levels and Symptom Scores in Patients With SchizophreniaPANSS PositivePANSS NegativePANSS GeneralPANSS TotalrprprprpCSF Neuropilin-20.27.009∗−0.06.59−0.04.680.05.61CSF MAPT0.11.30−0.07.49−0.12.24−0.05.63CSF CADM30.17.099−0.03.810.05.600.08.44CSF UNC5C0.14.20−0.10.33−0.12.25−0.05.63CSF Brorin0.16.12−0.12.26−0.02.820.00.97CSF Siglec-9−0.03.81−0.13.23−0.12.27−0.11.28CSF CLM-60.00.96−0.04.69−0.12.26−0.07.48CSF Ezrin−0.03.78−0.07.51−0.13.22−0.10.34CSF SMOC20.05.610.05.61−0.07.500.00.99CSF NBL10.07.510.10.350.17.100.15.15CSF Ephrin-A40.10.32−0.12.27−0.12.24−0.07.52CSF SCARB20.12.25−0.07.53−0.11.31−0.04.73CSF NCAN−0.06.60−0.06.59−0.07.53−0.07.48CSF Protogenin0.05.61−0.22.037∗−0.21.047∗−0.16.12CSF ROBO20.22.037∗−0.18.09−0.09.41−0.03.77CSF CRTAM0.10.33−0.04.740.01.940.03.79CSF RGMA0.24.023∗−0.04.70−0.01.890.06.55CSF Plexin B30.26.012∗−0.13.21−0.05.660.02.83CSF Carboxypeptidase A20.28.007∗−0.12.240.01.940.06.57CSF CD380.01.91−0.10.35−0.14.19−0.10.33CSF SMPD10.14.18−0.16.13−0.07.53−0.06.55CSF MSR10.08.430.01.90−0.04.710.01.89CSF Alpha-2-MRAP0.16.13−0.04.74−0.08.420.00.99CSF sFRP-30.04.68−0.07.480.10.320.04.68CSF EPHB60.15.15−0.12.25−0.06.54−0.02.83CSF RGMB0.19.07−0.18.08−0.08.45−0.04.72CSF Sialoadhesin0.11.300.00.990.04.670.06.56CSF Contactin-50.20.06−0.13.23−0.03.770.01.93CSF ADAM 220.10.33−0.01.950.00.970.04.73CSF ADAM 230.16.118−0.15.16−0.13.22−0.06.56CSF Matrilin-30.19.07−0.06.59−0.03.740.03.79CSF R-spondin-10.20.06−0.06.540.12.270.11.30CSF HAGH0.14.18−0.12.260.02.860.02.87CSF Galectin-80.00.96−0.06.57−0.14.17−0.09.37CSF BCAN0.21.043∗−0.13.22−0.05.650.00.96CSF Layilin0.16.14−0.05.61−0.03.800.02.83CSF Neprilysin0.26.011∗−0.02.84−0.06.580.06.59CSF GDF-80.22.030∗−0.17.10−0.07.48−0.02.86CSF THY 10.23.026∗0.08.420.15.150.19.07CSF WFIKKN10.21.040∗−0.03.770.04.710.09.42CSF TMPRSS50.07.49−0.17.11−0.15.15−0.11.29CSF Cadherin 3−0.08.46−0.06.56−0.03.79−0.06.55CSF GFR-alpha-10.14.17−0.04.69−0.14.19−0.03.74CSF GM-CSF-R-alpha0.11.30−0.07.51−0.13.21−0.05.61CSF Beta-NGF0.23.028∗0.01.960.09.390.13.22CSF SCARA50.08.44−0.15.16−0.14.18−0.09.37CSF CD2000.22.034∗−0.20.06−0.08.45−0.03.76CSF NTRK20.25.017∗−0.11.28−0.07.510.01.91CSF Granzyme A0.16.13−0.19.07−0.09.37−0.07.51CSF Draxin−0.03.76−0.14.18−0.20.06−0.16.12CSF SCARF20.03.78−0.03.80−0.08.47−0.04.71CSF GDNFR-alpha-30.06.54−0.12.27−0.11.27−0.08.46CSF Poliovirus Receptor0.17.10−0.14.18−0.07.50−0.02.82CSF TNFRSF12A0.12.24−0.08.46−0.08.43−0.03.80CSF SKR30.12.23−0.07.53−0.08.45−0.02.85CSF FLRT20.14.17−0.13.21−0.11.31−0.05.63CSF Carboxypeptidase M0.21.048∗−0.14.18−0.02.830.01.91CSF Glypican-50.18.08−0.16.13−0.07.51−0.03.80CSF BMP-4−0.01.91−0.08.47−0.13.23−0.10.36CSF FcRL20.16.13−0.05.610.03.790.05.62CSF MDGA10.22.038∗−0.16.12−0.05.61−0.01.94CSF PDGF-R-alpha0.22.037∗−0.06.560.00.990.05.60CSF CTSC0.01.89−0.07.48−0.13.23−0.09.41CSF Cadherin-60.20.06−0.10.33−0.08.45−0.01.94CSF DDR10.30.004∗−0.19.07−0.05.630.01.91CSF JAM-B0.13.21−0.05.60−0.04.710.01.95CSF Cathepsin S0.14.19−0.09.38−0.08.45−0.02.82CSF Neutral Ceramidase0.28.006∗−0.14.18−0.02.870.04.69CSF NAAA0.31.003∗−0.20.0530.03.810.05.63CSF N2DL-20.13.22−0.19.07−0.12.27−0.08.44CSF Plexin B10.11.29−0.18.09−0.10.33−0.08.47CSF TNFRSF210.15.15−0.13.21−0.05.66−0.02.87CSF CLM-10.01.94−0.09.38−0.05.63−0.06.60CSF Testican-10.22.035∗−0.12.25−0.01.890.03.79CSF Interleukin-12−0.01.93−0.14.17−0.11.31−0.11.30CSF Dkk-40.19.08−0.06.57−0.08.450.00.97CSF EDA2R0.09.41−0.03.76−0.15.15−0.06.58CSF NTRK30.24.018∗−0.12.23−0.04.700.02.84CSF LAIR-20.07.52−0.13.23−0.09.38−0.07.52CSF MANF0.21.046∗−0.09.40−0.06.550.01.92CSF Tenascin-R0.19.06−0.11.30−0.02.850.02.84CSF CD200R10.20.06−0.02.88−0.02.850.05.60CSF Nr-CAM0.23.028∗−0.03.770.09.370.12.26r, Pearson's partial correlation coefficient (covariates: age, sex, body mass index, and psychotropic medication use).CSF, cerebrospinal fluid; NPX, normalized protein expression; PANSS, Positive and Negative Syndrome Scale.∗Nominally significant cases. Correlations Between CSF Neuromarker Levels and Symptom Scores in Patients With Schizophrenia r, Pearson's partial correlation coefficient (covariates: age, sex, body mass index, and psychotropic medication use). CSF, cerebrospinal fluid; NPX, normalized protein expression; PANSS, Positive and Negative Syndrome Scale. Nominally significant cases. Positive correlations between CSF neuromarker levels and PANSS positive scores in patients with schizophrenia. CSF levels of neuropilin-2 (A), ROBO2 (B), RGMA (C), plexin-B3 (D), carboxypeptidase A2 (E), BCAN (F), neprilysin (G), GDF-8 (H), THY1 (I), WFIKKN1 (J), beta-NGF (K), CD200 (L), NTRK2 (M), carboxypeptidase M (N), MDGA1 (O), PDGF-R-alpha (P), DDR1 (Q), neutral ceramidase (R), NAAA (S), testican-1 (T), NTRK3 (U), MANF (V), and Nr-CAM (W) were positively correlated with PANSS positive scores (p < .05). CSF, cerebrospinal fluid; PANSS, Positive and Negative Syndrome Scale. Negative correlations between CSF protogenin levels and PANSS negative and general psychopathology scores in patients with schizophrenia. CSF protogenin levels were negatively correlated with PANSS negative (left) and general psychopathology (right) scores (p < .05). CSF, cerebrospinal fluid; PANSS, Positive and Negative Syndrome Scale. Correlations between CSF neuromarker levels and continuous clinical variables are presented in Table S1. Age was positively correlated with multiple CSF neuromarkers in both patients with schizophrenia and HC participants, as detailed in Table S1 (uncorrected or corrected p < .05). In patients with schizophrenia, BMI was positively correlated with CSF HAGH levels and negatively correlated with CSF LAIR-2 levels (uncorrected p < .05). In HC participants, BMI was negatively correlated with CSF N2DL-2 (NKG2D ligand 2) levels and positively correlated with CSF cell surface glycoprotein CD200R1 (CD200 receptor 1) levels (uncorrected p < .05). Age at onset was negatively correlated with CSF CD38 and WFIKKN1 levels in patients with schizophrenia (uncorrected p < .05). Duration of psychotropic medication use was positively correlated with several CSF neuromarkers in patients with schizophrenia, as detailed in Table S1 (uncorrected p < .05). Chlorpromazine-equivalent dose was positively correlated with CSF SMOC2, CD38, MSR1, and GFR-alpha-1 levels and negatively correlated with CSF NBL1, beta-NGF, FcRL2, and Nr-CAM levels in patients with schizophrenia (uncorrected p < .05). Comparisons of CSF neuromarker levels across nominal clinical variables are presented in Table S2. Among the schizophrenia group, CSF SMOC2, galectin-8, and TNFRSF12A levels were higher in males than in females, whereas among the HC group, CSF plexin-B3, HAGH, cadherin-3, and neutral ceramidase levels were lower in males than in females (uncorrected p < .05). Compared with medicated patients, drug-free patients had higher CSF levels of MAPT (microtubule-associated protein tau), CADM3 (cell adhesion molecule 3), NCAN (neurocan core protein), ROBO2, EPHB6 (ephrin type-B receptor 6), ADAM23, NTRK2, PDGF-R-alpha, TNFRSF21, testican-1, and tenascin-R but lower levels of carboxypeptidase A2, cadherin-3, and NAAA (uncorrected p < .05). Correlation matrices for CSF neuromarker levels in patients with schizophrenia and HC participants are presented in Tables S3 and S4, respectively. Most significant correlations among CSF neuromarker levels were positive, whereas negative correlations were uncommon (uncorrected or corrected p < .05). Discussion Lower CSF levels of cadherin 3, FcRL2, granzyme A, NAAA, plexin-B1, and sFRP-3 may represent candidate trait markers in schizophrenia. Cadherins are cell-adhesion molecules (27) that have been implicated in the synaptic dysfunction associated with schizophrenia (28), which is consistent with the current findings. FcRL2 is a member of the immunoglobulin superfamily (29). Granzyme A is a cytotoxic protease that promotes the release of proinflammatory cytokines from target cells (30). NAAA mediates anti-inflammatory biochemical reactions by regulating the signaling activity of palmitoylethanolamide (31). Taken together, these findings suggest possible alterations in both proinflammatory and anti-inflammatory pathways in schizophrenia. Plexin-B1 induces growth cone collapse and neurite retraction through interaction with semaphorin 4D (32,33); therefore, the current findings may be associated with disruption of these processes in schizophrenia. sFRP-3 acts as an antagonist in the Wnt signaling pathway (34), raising the possibility of altered Wnt signaling in schizophrenia. Several CSF neuromarkers, including BCAN, beta-NGF, carboxypeptidases A2 and M, CD200, DDR1, GDF-8, MANF, MDGA1, NAAA, neprilysin, neuropilin-2, neutral ceramidase, Nr-CAM, NTRK2, NTRK3, PDGF-R-alpha, plexin-B3, RGMA, ROBO2, testican-1, THY1, and WFIKKN1, may represent candidate state markers associated with positive symptoms in schizophrenia. Beta-NGF (35) is one of the growth factors implicated in the pathogenesis of schizophrenia (36). Nr-CAM has been linked to altered synaptic plasticity in schizophrenia (37). Pathological alterations in NTRK2 and NTRK3 signaling have been reported in neurologic and psychiatric disorders, including schizophrenia (38). PDGF-R-alpha is expressed in the hippocampus and is important for brain development and function (39). WFIKKN1 functions as a companion protein for various growth factors in the brain (40). Together, these findings suggest that CNS growth factor-related pathways may be associated with positive symptom severity in schizophrenia. BCAN is a major proteoglycan that plays an important role in cell-cell and cell-matrix interactions in the adult brain (41). DDR1 is involved in cellular adhesion, migration, matrix remodeling, and proliferation (42). MDGA1 regulates the formation of trans-synaptic bridges (43). Disturbed synaptic adhesion may have an association with positive symptoms in schizophrenia. Neuropilin-2 is a single-pass transmembrane receptor involved in axon guidance (44). Plexin-B3 is a functional receptor for semaphorin 5A, which induces growth cone collapse in retinal ganglion cells (32). RGMA contributes to impaired axonal regeneration, neuroinflammation, and neuronal degeneration (45). ROBO2 plays a crucial role in axon guidance (46). THY1 is attached to the outer surface of the plasma membrane and alters actin dynamics and neurite retraction (47). Taken together, these findings raise the possibility that abnormalities in axonal and neuritic processes are associated with positive symptom severity in schizophrenia. Carboxypeptidase A2 is a digestive enzyme, whereas carboxypeptidase M functions in the processing of peptide hormones (48). The CD200-CD200R axis has been implicated in the regulation of neuron-microglia communication in the pathogenesis of schizophrenia (49). GDF-8, also known as myostatin, is a master regulator of skeletal muscle mass (50). MANF protects neurons from endoplasmic reticulum stress–associated injury, amyloid-β–induced toxicity (51), and age-related dysfunction (52). Neprilysin is an amyloid-β–degrading enzyme (53). Neutral ceramidase hydrolyzes the N-acyl linkage between the sphingoid base and fatty acid of ceramide (54). Testican-1, a multidomain proteoglycan, carries chondroitin sulfate and other oligosaccharides (55). Although their interactions remain unclear, these functions may be associated with positive symptom severity in schizophrenia. CSF protogenin may represent a candidate negative state marker associated with negative and general psychopathology symptoms in schizophrenia. Protogenin is a transmembrane protein that mediates signaling involved in cell adhesion, neural differentiation, and neurogenesis in the developing nervous system (56,57). Although the underlying mechanisms remain unclear, CSF protogenin may be inversely associated with negative and general psychopathology symptoms in schizophrenia. This study has several limitations. First, this proteomic study could present only exploratory associations, while biologically meaningful conclusions remain to be drawn about the overall candidate neuromarkers presented here. Second, one might consider the fact that the study recruited through advertisements as a source of bias; however, unless CSF is routinely obtained through care, this is probably unavoidable. Third, HC participants were not age matched, although age was included as a covariate in the trait marker analyses. However, given the effects of aging on brain physiology and generally positive correlations between age and CSF neuromarker levels in our sample, higher age in the HC group might have distorted some data from this cohort. Fourth, psychotropic medication use was associated with 14 CSF proteins; however, this factor was included as a covariate in the trait and state marker analyses. Fifth, additional CSF neuromarkers may exist beyond the 92 proteins targeted in this study. Sixth, the identified trait and state markers were only nominally significant, which increases the possibility of type I error. Finally, because of the cross-sectional design, the causal and temporal relationships between CSF neuromarkers and the pathogenesis of schizophrenia could not be determined. Given that schizophrenia usually implies a chronic disorder and long-term treatment with medication, the relevance of a single cross-sectional sample at one point toward a longitudinal neurodevelopmental disorder needs to be carefully interpreted. To reach real trait and state markers of schizophrenia, scope for longitudinal studies including medication-naïve first-episode psychosis integrated into care is warranted in the future, referring to our cross-sectional data. Conclusions In this primarily exploratory and hypothesis-generating study, we identified several candidate CSF trait and state markers associated with schizophrenia and symptom severity. Further preclinical and longitudinal clinical studies are needed to validate these candidate neuromarkers and clarify their potential clinical relevance in schizophrenia. Acknowledgments and Disclosures This work was supported by the SENSHIN Medical Research Foundation (to SH), the Teikyo University Research Encouragement (Grant No. 2400002 [to SH]), the Japan Society for the Promotion of Science KAKENHI (Grant No. 26K10476 [to SH]), and the Japan Agency for Medical Research and Development GAPFREE4 (Grant No. 21ak0101151h0002 [to SH, KH, and HK]). These funding sources had no role in the study design, data collection, data analysis, manuscript preparation, or decision to submit the manuscript for publication. SH, KH, and TE established the diagnoses, assessed the symptoms, and performed the lumbar punctures for CSF (collection). KH and MT prepared the CSF samples. RM created and maintained the database. HK contributed to the intellectual content of the study. SH performed the statistical analyses and drafted the manuscript, which all authors revised and approved. We thank Ms. Chiori Maeda, Mr. Ikki Ishida, Ms. Junko Matsuo, Ms. Moeko Hiraishi, Mr. Takahiro Tomizawa, and Ms. Tomoko Kurashimo for their assistance with participant recruitment, clinical assessments, and lumbar punctures. We also thank Ms. Ayumi Fujisawa, Ms. Fuyuko Yoshida, Ms. Izumi Sato, Ms. Haruna Kaneko, Ms. Mayumi Kando, and Ms. Misao Nakano for their assistance with sample management. This manuscript was proofread by a scientific English editor from Editage (Tokyo, Japan). Written permission was obtained from all individuals named in this section. The authors report no biomedical financial interests or potential conflicts of interest. Footnotes Supplementary material cited in this article is available online at https://doi.org/10.1016/j.bpsgos.2026.100804. Supplementary Material Figure S1 Tables S1–S4 Key Resources Table References Associated Data Supplementary Materials Figure S1 Tables S1–S4 Key Resources Table