A novel perspective on biomarkers of neurodegeneration and neuroinflammation in first episode psychosis: a cross-sectional study Authors: Valeria Zakurazhnaya, Yana Zorkina, Olga Abramova, Daria Riabinina, Alexandra Ochneva, Valeriya Ushakova, Alexander Reznik, Georgy P. Kostyuk, Anna Morozova Published: 2026 ABSTRACT BackgroundThe biological distinction between first-episode psychosis (FEP) and chronic schizophrenia remains unclear. This study investigated a panel of non-canonical serum biomarkers related to neurodegeneration, neuroinflammation, and neurotrophic signaling across these clinical stages.MethodsIn a cross-sectional design, serum concentrations of 18 biomarkers were measured using multiplex immunoassay in 49 FEP patients, 80 patients with chronic schizophrenia (SCH), and 80 healthy controls (CON). Principal component analysis identified two major axes (neuroinflammatory-neurotrophic and synaptic), but the distributions of diagnostic groups overlapped substantially.ResultsA distinct biomarker profile was identified in FEP, characterized by significantly lower levels of YKL-40 and BDNF, elevated NCAM-1, decreased TDP-43, and a higher frequency of detectable Aβ1‑42 compared to both SCH and CON groups. In contrast, the SCH group did not differ from CON on most of these markers. Correlation analysis revealed a dense, interconnected biomarker network in SCH, centered on MIF and BDNF, while FEP exhibited a single strong correlation between TDP-43 and YKL-40. Principal component analysis identified two major axes (neuroinflammatory-neurotrophic and synaptic) but showed substantial overlap between diagnostic groups.ConclusionThe findings reveal a unique, stage-dependent biological signature in early psychosis, differentiating FEP from both health and the chronic phase of illness. This supports the view of FEP as a distinct pathophysiological state involving dysregulated neuroinflammation, synaptic plasticity, and protein homeostasis, rather than merely a prodromal stage of chronic schizophrenia.Clinical trial number Not applicable. Introduction First episode of psychosis (FEP) is the initial manifestation of psychotic disorders. This condition is characterized by heterogeneous symptoms, which can be divided into two groups: nonaffective and affective (bipolar disorder and major depressive disorder with psychotic features) [1]. By definition, FEP symptoms persist for at least one week and emerge within two years of disease onset. The main symptoms include delusions, hallucinations, disorganized speech and behaviour, catatonia, as well as negative symptoms [2]. However, not all patients experiencing FEP progress to chronic psychotic disorders. Approximately 49% of patients with FEP receive a diagnosis of schizophrenia or schizophrenia spectrum disorder within the first 5 years [3], while 30–40% experience only a single psychotic episode [4]. In clinical practice, the specific diagnosis (e.g., schizophrenia, schizoaffective disorder, acute transient psychosis) is often not yet determinable at the time of the first episode. Therefore, FEP is defined syndromatically — by the presence of psychotic symptoms — rather than by final diagnostic subtype. Our study adheres to this pragmatic, real-world definition. Emerging attention to FEP reflects its critical window for early intervention, facilitating timely treatment, rehabilitation, and social support to potentially mitigate progression to severe and chronic illness [5]. Trends towards personalized medicine in FEP include identifying people with high rehabilitation potential and preventing the development of psychotic conditions later [6]. Personalized medicine for FEP will help prevent the development of chronic mental disorders through timely and adequate intervention. To realize these opportunities, it is necessary to identify the biological markers of the disease. However, while schizophrenia biomarkers have been extensively studied, there remains a relative scarcity of data specifically addressing FEP biomarkers, partly due to challenges in patient recruitment. Comparative cross-sectional studies involving FEP and chronic schizophrenia can identify biomarkers that differ by illness stage, generating hypotheses about potential disease trajectories and underlying biological processes. Among the theories of the FEP development, there are genetic and psychosocial factors, impaired neurodevelopment, or stress experienced in childhood. Biochemical theories of FEP posit a complex pathophysiology centered on a fundamental imbalance of neuroactive substances. This model integrates dysregulation of key neurotransmitters, elevated markers of neuroinflammation and oxidative stress, impaired neuroplasticity, and a disruption of the functional integrity of the blood-brain barrier (BBB). Many studies have found a link between FEP and inflammation and abnormal immune responses. For example, a meta-analysis showed that IFN-γ, interleukin (IL)-6, and IL-12 have a large effect size in patients with FEP, while IL-17 has a moderate effect size [7]. However, the lack of specificity of inflammatory markers in psychiatric disorders complicates their diagnostic and prognostic utility. In this regard, we decided to pay attention to another link in the pathogenesis of the neuroinflammatory theory of schizophrenia, namely BBB dysfunction and impaired functioning of neuroactive peptides. Neuroactive peptides, which regulate neurodevelopment, synaptic plasticity, and neurogenesis, may play critical roles in FEP but have been underexplored in biomarker research. The neurodevelopmental hypothesis suggests that schizophrenia originates from early-life disruptions in brain development, which create a hidden vulnerability. This vulnerability becomes apparent in adulthood as a significant disruption of synaptic plasticity. Postmortem studies provide strong evidence for this, consistently showing a significant loss of dendritic spines in critical regions such as the prefrontal and temporal cortices [8]. These spines play a crucial role in regulating glutamatergic neurotransmission, and their depletion inevitably leads to maladaptive rewiring of neural circuits and a fundamental decrease in circuit plasticity [9]. This impaired synaptic plasticity is reflected in the instability of spines and synapses. The consequences of this initial deficit are thought to spread and amplify throughout development, as the refinement of neural networks progresses hierarchically. In an attempt to maintain stability, compensatory mechanisms such as the scaling of inhibitory activity may arise to counter the altered glutamatergic transmission. However, these compensations often lead to a set of suboptimally tuned cortical circuits, which can generate erroneous perceptions of stimuli and experiences. Disrupted prefrontal inputs can subsequently alter subcortical dopamine neurotransmission in late adolescence, leading to a cascade of dysfunction that culminates in psychosis [8]. The molecular basis of this process involves key neuroactive peptides that regulate synaptic plasticity and dendritic morphology. These peptides can be measured even in peripheral blood, and among them, brain-derived neurotrophic factor (BDNF) plays a crucial role [10]. Clinical evidence supports this view, as BDNF levels in the peripheral blood of individuals experiencing FEP are significantly lower than in healthy controls. This deficit appears to persist despite treatment [11], highlighting BDNF’s potential as a trait marker for underlying synaptic pathology. However, the influence of other neurotrophic factors, along with a wider range of neuroactive molecules, has not been fully explored in FEP. One of the hypotheses regarding schizophrenia pathogenesis is the connection between its pathological changes and neurodegenerative processes. This theory suggests that such degenerative processes disrupt fundamental neural circuits responsible for perception, thought, and emotion, leading to the core symptoms of the disorder, including hallucinations, delusions, cognitive disorganization, and blunted affect [12]. This perspective is supported by evidence showing that critical domains of functioning, such as cognition and white matter integrity, experience a progressive decline as the disorder progresses [13]. While neuroinflammation has been extensively studied in FEP, we deliberately shifted focus toward neurodegeneration-related pathways to explore novel, underexplored mechanisms. This choice was not arbitrary: a growing body of literature, albeit controversial, links protein aggregation, synaptic dysfunction, and neuronal injury markers to schizophrenia. The very controversy in this field highlights the need for new studies that test these hypotheses in FEP specifically. Thus, our study does not aim to replicate inflammation findings but to provide a complementary perspective that may reveal early pathophysiological events missed by conventional inflammatory panels. We selected a commercially available multiplex panel targeting 18 proteins involved in neurodegeneration, neuroinflammation, and synaptic integrity (e.g., TDP-43, Aβ1–42, BDNF, NCAM-1, YKL-40, MIF). This targeted approach allows precise quantification of low-abundance proteins that may be missed by untargeted proteomics and enables direct comparison with prior hypothesis-driven studies. We measure protein concentrations because they reflect dynamic, real-time pathophysiological states, in contrast to genomic analysis which captures static genetic risk. However, there is no clear consensus in the literature regarding which factor initiates pathogenesis or which is the primary driver. Thus, this exploratory cross-sectional study aims to identify stage-associated differences in serum biomarkers of neurodegeneration, neuroinflammation, and synaptic integrity, generating hypotheses about early pathophysiological processes in psychosis beyond conventional pathways. Materials and methods Patients The study enrolled male and female patients with mental disorders from Psychiatric Hospital no. 1 Named after N.A. Alekseev of the Department of Health of Moscow. Informed consent was obtained from all participants. This study was conducted in accordance with the recommendations of the Helsinki Declaration. The study protocols were approved by the Local Ethics Committee of Psychiatric Hospital no. 1 Named after N.A. Alexeev of the Department of Health of Moscow (Protocol No. 2/28 October 2020). Inclusion criteria: diagnosis of “schizophrenia” or “schizophrenic spectrum disorder” according to the International Classification of Diseases (F20, F21, F23, F25), age 18–65 years, ethnic origin — Caucasians, preferably ethnically Russian. Exclusion criteria: substance abuse and severe comorbid somatic or neurological disorders. Patients with somatic pathology were excluded based on medical records. In particular, patients with severe disorders of the cardiovascular, respiratory, hematopoietic, and nervous systems, accompanied by significant organ dysfunction or oncologic diseases undergoing treatment, were excluded. The control group was recruited from among volunteers who underwent preventive medical examinations at Municipal Clinical Hospital No. 121 (Moscow). The volunteers were comparable in age to the patients. Participants were excluded from the study for the following reasons: mental illness, family history (first degree relatives) of mental illness; substance abuse; severe physical illnesses. The first-episode psychosis (FEP) group included patients with first-ever psychotic symptoms (delusions, hallucinations, disorganized speech or behavior) lasting at least 7 days and manifesting within 24 months prior to study inclusion. Diagnoses were established according to ICD-10 criteria and included the following categories: schizophrenia with less than one year of observation (F20), schizotypal disorder (F21), chronic delusional disorder (F22), acute and transient psychotic disorders (F23), schizoaffective disorder (F25). Thus, the FEP group reflected a nosologically heterogeneous cohort of patients with the onset of a psychotic state, which corresponds to real-world clinical practice where the definitive diagnosis is often not yet determinable at the first episode stage. The diagnoses of the FEP patients are shown in Table 1. Table 1The diagnoses of the FEP patientsDiagnosis (ICD-10) n %F20 (schizophrenia)714.2%F21 (schizotypal disorder)1020.4%F22 (chronic delusional disorder)12%F23 (acute and transient psychotic disorders)2245%F25 (schizoaffective disorder)918.4% Table 1The diagnoses of the FEP patientsDiagnosis (ICD-10) n %F20 (schizophrenia)714.2%F21 (schizotypal disorder)1020.4%F22 (chronic delusional disorder)12%F23 (acute and transient psychotic disorders)2245%F25 (schizoaffective disorder)918.4% The diagnoses of the FEP patients The cumulative lifetime duration of antipsychotic medication intake in all FEP group patients did not exceed 4 weeks (patients were predominantly treatment-naïve or had minimal treatment experience). Prior psychiatric hospitalizations for psychosis were absent. The schizophrenia group: patients meeting ICD-10 diagnostic criteria for schizophrenia (F20), with disease duration of at least 5 years (to reliably define the chronic phase) and a history of at least two distinct psychotic episodes or continuous illness with residual symptoms. The sample of study participants was: сontrol (CON) — 80 people, schizophrenia (SCH) — 80 people, first episode of psychosis (FEP) — 49 people. Blood analysis Blood parameters were measured in blood serum. Blood samples for analysis were taken from the ulnar vein on an empty stomach, usually before 9 a.m. The serum was isolated immediately after blood collection by centrifugation at 3000 rpm for 10 min at a temperature of 4 ° C, and then stored at a temperature of − 80 ° C until analysis. All samples were thawed only once for the analysis presented in this study. The concentration of blood serum parameters was determined using multiplex analysis with a ProcartaPlex Human Neurodegeneration Panel 1 9-plex according to the manufacturer’s protocol, the concentration was determined in pg/ml. The measurements performed using a Luminex analyzer (LuminexTM 100/200TM). Thus, the following biochemical indicators were assessed in the blood during the experiment: Amyloid-beta 1‑42 (Aβ1‑42); BDNF; Ciliary Neurotrophic Factor (CNTF); Fibroblast Growth Factor 21 (FGF21); Glial Cell Line-Derived Neurotrophic Factor (GDNF); Glial Fibrillary Acidic Protein (GFAP); Kallikrein-6 (KLK6); Macrophage Migration Inhibitory Factor (MIF); Neural Cell Adhesion Molecule 1 (NCAM-1); Neurofilament Heavy chain (NF-H); Nerve Growth Factor beta (NGF‑β); Phosphorylated Tau at threonine 181 (p‑tau181); S100 Calcium-Binding Protein B (S100B); TAR DNA-binding protein 43 (TDP-43); Total Tau protein (total Tau); Ubiquitin C-terminal Hydrolase L1 (UCH-L1); Chitinase-3-like protein 1 (YKL-40). Statistical analysis Statistical analysis of differences between groups was performed using Jamovi (version 2.3.13). Visualization was performed with RStudio soft (version 2025.09.2). The normality of the data distribution was assessed using the Shapiro-Wilk test. Due to the non-normal distribution of most parameters, group comparisons were performed using the Kruskal-Wallis test for continuous variables (Dwass-Steel-Critchlow-Fligner Pairwise Comparisons (DSCF)) and the Chi-squared test for categorical variables. p < 0.05 was considered significant. For the obtained blood analysis data, concentration values that were below the minimum standard were replaced with extrapolated values. The imputation method for values that fell above or below the detection range and could not be extrapolated was as follows: values of LLOD/2 (lower limit of detection, i.e., the concentration of the maximum diluted standard divided by 2) were used to replace missing values below the detection limit, and values of ULOD*2 were used to replace missing values above the detection limit. Parameters for which more than 50% of the results were represented by values below the minimum threshold were treated as categorical variables. The remaining parameters were quantitative [14]. P-values were adjusted for biomarkers multiple comparisons using the Benjamini-Hochberg false discovery rate (FDR) procedure, applied separately within each family of tests (quantitative markers analysed with Kruskal-Wallis, and categorical markers analysed with chi-square). p < 0.05 was considered significant. Principal component analysis (PCA) was performed using JASP statistical software (version 0.18.3). Visualization was performed with RStudio soft (version 2025.09.2). Prior to analysis, all biomarker variables were standardized (z-scored) to ensure comparability and mitigate scale-related bias. Component loadings ≥ 0.40 were considered practically significant for interpretation. All statistical tests were two-tailed, with p < 0.05 considered statistically significant. Pearson correlation coefficients were calculated to assess pairwise relationships between biomarkers within each group. The threshold for statistical significance was set at p < 0.01. The coefficient of determination (r²) was computed to estimate the proportion of variance explained by each correlation. Correlation analyses were exploratory and not adjusted for multiple comparisons; results should be interpreted as hypothesis-generating. Analyses were performed using Jamovi (version 2.3.13). Results The sociodemographic characteristics of the participants are presented in Table 2. Table 2Sociodemographic characteristics of the study participantsCharacteristicFEP (n = 49)SCH (n = 80)CON (n = 80)P-valueAge, years 0.001 Median (Q1; Q3)26.0 [22.0, 33.0]35.0 [31.0, 41.5]35.0 [32.0, 41.5]Range18–4818–5522–59Gender, n (%)0.109Male21 (42.9%)23 (28.6%)41 (51.2%)Female28 (57.1%)57 (71.4%)39 (48.8%)BMI, kg/m2 < 0.001 Median (Q1; Q3)21.8 [19.8, 24.0]24.2 [21.3, 27.7]24.7 [22.3, 27.0]Chlorpromazine equivalents, mg/day 0.215 Median (Q1; Q3)350 [200, 450]300 [225, 450]-Range8–70015–1000-Type of antipsychotic, n (%) 0.015 Atypical32 (65.3%)54 (67.5%)-Typical6 (12.2%)16 (20.0%)-Mixed therapy11 (22.5%)10 (12.5%)-* Group codes: CON (controls), FEP (first episode psychosis), SCH (schizophrenia). p < 0.05 was considered significant Table 2Sociodemographic characteristics of the study participantsCharacteristicFEP (n = 49)SCH (n = 80)CON (n = 80)P-valueAge, years 0.001 Median (Q1; Q3)26.0 [22.0, 33.0]35.0 [31.0, 41.5]35.0 [32.0, 41.5]Range18–4818–5522–59Gender, n (%)0.109Male21 (42.9%)23 (28.6%)41 (51.2%)Female28 (57.1%)57 (71.4%)39 (48.8%)BMI, kg/m2 < 0.001 Median (Q1; Q3)21.8 [19.8, 24.0]24.2 [21.3, 27.7]24.7 [22.3, 27.0]Chlorpromazine equivalents, mg/day 0.215 Median (Q1; Q3)350 [200, 450]300 [225, 450]-Range8–70015–1000-Type of antipsychotic, n (%) 0.015 Atypical32 (65.3%)54 (67.5%)-Typical6 (12.2%)16 (20.0%)-Mixed therapy11 (22.5%)10 (12.5%)-* Group codes: CON (controls), FEP (first episode psychosis), SCH (schizophrenia). p < 0.05 was considered significant Sociodemographic characteristics of the study participants * Group codes: CON (controls), FEP (first episode psychosis), SCH (schizophrenia). p < 0.05 was considered significant This confirms the validity of our study design, which aimed to compare different stages of the psychotic disorder. At the same time, despite the significantly shorter duration of therapy in the FEP group, the median daily doses of antipsychotics in terms of chlorpromazine equivalents between the groups had no statistically significant differences (p = 0.215), indicating a comparable intensity of pharmacological load at the time of examination. At the same time, significant differences in the structure of prescribed therapy were revealed (p = 0.015): in the FEP group, atypical antipsychotics or combination therapy were significantly more often used, while in the SCH group, the proportion of patients receiving typical antipsychotics was higher. In the chronic schizophrenia (SCH) group, the disease duration was a median of 8.0 years (interquartile range: 4.0–14.0 years). Comparison of clinical severity between the FEP and SCH groups showed significant differences in total Positive and Negative Syndrome Scale (PANSS) score (mean ± SD: SCH = 94.74 ± 19.08; FEP = 85.51 ± 25.67; t = 2.333, p = 0.021). However, significant differences were found on the Clinical Global Impression (CGI) -Severity scale: the FEP group had higher scores (median = 6.00; 25th–75th percentiles: 5.00–6.00) compared to the SCH group (median = 5.00; 25th–75th percentiles: 4.00–5.00; U = 946.500, p < 0.001). On the CGI-Improvement scale, the differences did not reach statistical significance (p = 0.098). At the initial stage of analysis, we identified differences in the measured blood parameters between the study groups. Table 3; Fig. 1 show the studied blood biomarkers and their statistical differences between the three groups of patients: FEP, schizophrenia and control. Table 3Quantitative and categorical serum biomarker differences between the study groupsThe serum parametersFEPSCHCONχ² p FDR-adjusted p-value Quantitative parameters KLK61489 (1151; 1747)1341 (938; 1666)1238 (832; 1834)1.770.410.41YKL-403509 (3063; 4180)4730 (3269; 6036)4248 (3520; 5159)16.32 0.00028 0.002 NCAM-187,409 (72726; 105199)72,086 (58437; 94998)79,645 (64799; 108714)9.54 0.0085 0.03 BDNF402 (255; 581)446 (202; 905)613 (320; 983)7.46 0.024 0.04 MIF40.8 (29.3; 49.2)41.4 (28.5; 64.8)46.1 (35.4; 60.9)4.340.110.15TDP-43198 (48; 412)218 (48; 751)409 (157; 743)8.54 0.014 0.03 Neurogranin3294 (1695; 5877)3563 (1758; 7621)3050 (1271; 4942)3.150.210.24 Quality parameters UCH-L13/49 (6.1%)7/80 (8.7%)5/80 (6.2%)0.480.791.0CNTF6/49 (12.2%)11/80 (13.7%)14/80 (17.5%)0.780.681.0Aβ1‑4231/49 (63.3%)29/80 (36.2%)27/80 (33.7%)12.4 0.00199 0.02 total Tau1/49 (2%)2/80 (2.5%)2/80 (2.5%)0.030.981.0NGF‑β1/49 (2%)0/80 (0%)0/80 (0%)3.280.190.70S100B0/49 (0)2/80 (2.5%)1/80 (1.2%)1.370.501.0GFAP3/49 (6.1%)3/80 (3.7%)3/80 (3.7%)0.510.771.0NF-H5/49 (10.2%)5/80 (6.2%)4/80 (5%)1.360.511.0GDNF9/49 (18.4%)7/80 (8.7%)7/80 (8.7%)3.540.170.93p‑tau1811/49 (2%)2/80 (2.5%)2/80 (2.5%)0.030.980.98FGF217/49 (14.3%)9/80 (11.2%)10/80 (12.5%)0.260.881.0*Continuous data are reported as Med (Q1; Q3) and analyzed with the Kruskal–Wallis test; categorical data are shown as count/total (percentage) and assessed with the chi-square test. Group codes: CON (controls), FEP (first-episode psychosis), SCH (schizophrenia). p-values were adjusted for multiple testing using the false discovery rate (FDR) procedure. p < 0.05 was considered significant; Aβ1‑42 – Amyloid-beta 1‑42; BDNF – Brain-Derived Neurotrophic Factor; CNTF – Ciliary Neurotrophic Factor; FGF21 – Fibroblast Growth Factor 21; GDNF – Glial Cell Line-Derived Neurotrophic Factor; GFAP – Glial Fibrillary Acidic Protein; KLK6 – Kallikrein-6; MIF – Macrophage Migration Inhibitory Factor; NCAM-1 – Neural Cell Adhesion Molecule 1; NF-H – Neurofilament Heavy chain; NGF‑β – Nerve Growth Factor beta; p‑tau181 – Phosphorylated Tau at threonine 181; S100B – S100 Calcium-Binding Protein B; TDP-43 – TAR DNA-binding protein 43; total Tau – Total Tau protein; UCH-L1 – Ubiquitin C-terminal Hydrolase L1; YKL-40 – Chitinase-3-like protein 1 Table 3Quantitative and categorical serum biomarker differences between the study groupsThe serum parametersFEPSCHCONχ² p FDR-adjusted p-value Quantitative parameters KLK61489 (1151; 1747)1341 (938; 1666)1238 (832; 1834)1.770.410.41YKL-403509 (3063; 4180)4730 (3269; 6036)4248 (3520; 5159)16.32 0.00028 0.002 NCAM-187,409 (72726; 105199)72,086 (58437; 94998)79,645 (64799; 108714)9.54 0.0085 0.03 BDNF402 (255; 581)446 (202; 905)613 (320; 983)7.46 0.024 0.04 MIF40.8 (29.3; 49.2)41.4 (28.5; 64.8)46.1 (35.4; 60.9)4.340.110.15TDP-43198 (48; 412)218 (48; 751)409 (157; 743)8.54 0.014 0.03 Neurogranin3294 (1695; 5877)3563 (1758; 7621)3050 (1271; 4942)3.150.210.24 Quality parameters UCH-L13/49 (6.1%)7/80 (8.7%)5/80 (6.2%)0.480.791.0CNTF6/49 (12.2%)11/80 (13.7%)14/80 (17.5%)0.780.681.0Aβ1‑4231/49 (63.3%)29/80 (36.2%)27/80 (33.7%)12.4 0.00199 0.02 total Tau1/49 (2%)2/80 (2.5%)2/80 (2.5%)0.030.981.0NGF‑β1/49 (2%)0/80 (0%)0/80 (0%)3.280.190.70S100B0/49 (0)2/80 (2.5%)1/80 (1.2%)1.370.501.0GFAP3/49 (6.1%)3/80 (3.7%)3/80 (3.7%)0.510.771.0NF-H5/49 (10.2%)5/80 (6.2%)4/80 (5%)1.360.511.0GDNF9/49 (18.4%)7/80 (8.7%)7/80 (8.7%)3.540.170.93p‑tau1811/49 (2%)2/80 (2.5%)2/80 (2.5%)0.030.980.98FGF217/49 (14.3%)9/80 (11.2%)10/80 (12.5%)0.260.881.0*Continuous data are reported as Med (Q1; Q3) and analyzed with the Kruskal–Wallis test; categorical data are shown as count/total (percentage) and assessed with the chi-square test. Group codes: CON (controls), FEP (first-episode psychosis), SCH (schizophrenia). p-values were adjusted for multiple testing using the false discovery rate (FDR) procedure. p < 0.05 was considered significant; Aβ1‑42 – Amyloid-beta 1‑42; BDNF – Brain-Derived Neurotrophic Factor; CNTF – Ciliary Neurotrophic Factor; FGF21 – Fibroblast Growth Factor 21; GDNF – Glial Cell Line-Derived Neurotrophic Factor; GFAP – Glial Fibrillary Acidic Protein; KLK6 – Kallikrein-6; MIF – Macrophage Migration Inhibitory Factor; NCAM-1 – Neural Cell Adhesion Molecule 1; NF-H – Neurofilament Heavy chain; NGF‑β – Nerve Growth Factor beta; p‑tau181 – Phosphorylated Tau at threonine 181; S100B – S100 Calcium-Binding Protein B; TDP-43 – TAR DNA-binding protein 43; total Tau – Total Tau protein; UCH-L1 – Ubiquitin C-terminal Hydrolase L1; YKL-40 – Chitinase-3-like protein 1 Quantitative and categorical serum biomarker differences between the study groups *Continuous data are reported as Med (Q1; Q3) and analyzed with the Kruskal–Wallis test; categorical data are shown as count/total (percentage) and assessed with the chi-square test. Group codes: CON (controls), FEP (first-episode psychosis), SCH (schizophrenia). p-values were adjusted for multiple testing using the false discovery rate (FDR) procedure. p < 0.05 was considered significant; Aβ1‑42 – Amyloid-beta 1‑42; BDNF – Brain-Derived Neurotrophic Factor; CNTF – Ciliary Neurotrophic Factor; FGF21 – Fibroblast Growth Factor 21; GDNF – Glial Cell Line-Derived Neurotrophic Factor; GFAP – Glial Fibrillary Acidic Protein; KLK6 – Kallikrein-6; MIF – Macrophage Migration Inhibitory Factor; NCAM-1 – Neural Cell Adhesion Molecule 1; NF-H – Neurofilament Heavy chain; NGF‑β – Nerve Growth Factor beta; p‑tau181 – Phosphorylated Tau at threonine 181; S100B – S100 Calcium-Binding Protein B; TDP-43 – TAR DNA-binding protein 43; total Tau – Total Tau protein; UCH-L1 – Ubiquitin C-terminal Hydrolase L1; YKL-40 – Chitinase-3-like protein 1 Fig. 1Statistically significant changes in the concentration of biomarkers between groups of patients with the first episode of psychosis (FEP), schizophrenia (SCH) and control (CON); (A) Serum YKL-40 concentration in participants; (B) Serum NCAM-1 concentration in participants; (C) Serum BDNF concentration in participants; (D) Serum TDP-43 concentration in participants; (E) serum Aβ1‑42 concentration in participants; A-D: Data are expressed as Med (Q1; Q3) in boxplots, Black dots – outliers, Colored yellow dots - single data points; E: Data are expressed as count plot, Red boxes - number of patients with increase amyloid, Blue boxes - number of patients with low amyloid; * - p < 0.05; ** - p < 0.01 Statistically significant changes in the concentration of biomarkers between groups of patients with the first episode of psychosis (FEP), schizophrenia (SCH) and control (CON); (A) Serum YKL-40 concentration in participants; (B) Serum NCAM-1 concentration in participants; (C) Serum BDNF concentration in participants; (D) Serum TDP-43 concentration in participants; (E) serum Aβ1‑42 concentration in participants; A-D: Data are expressed as Med (Q1; Q3) in boxplots, Black dots – outliers, Colored yellow dots - single data points; E: Data are expressed as count plot, Red boxes - number of patients with increase amyloid, Blue boxes - number of patients with low amyloid; * - p < 0.05; ** - p < 0.01 The Kruskal–Wallis test and chi‑square tests revealed significant differences between‑group only for the following blood parameters: YKL‑40, NCAM‑1, BDNF, TDP‑43, and Aβ1‑42 (Table 3). Post‑hoc analysis indicated that serum YKL‑40 levels were significantly lower in the FEP group compared to both the CON (p = 0.003) and SCH (p = 0.001) groups (Fig. 1A). NCAM‑1 concentration differed significantly between the SCH and FEP groups, with higher levels observed in FEP patients (p = 0.01) (Fig. 1B). Furthermore, BDNF and TDP‑43 concentrations were significantly lower in the FEP group relative to controls (p = 0.02 and p = 0.01, respectively) (Fig. 1C, D). Significant differences were also observed for the categorical parameter Aβ1‑42. The analysis revealed an increased proportion of detectable serum Aβ1‑42 in the FEP group, whereas the frequency of elevated amyloid levels was lower in both the SCH and CON groups (Fig. 1E). To assess the potential effect of age on the studied biomarkers, Spearman correlation analysis was performed between patient age and serum concentrations of all measured proteins in the overall sample. No statistically significant correlations were found for any of the biomarkers (all p > 0.05). The highest correlation coefficients were observed for GDNF (ρ = − 0.219, p = 0.051) and Neurogranin (ρ = 0.211, p = 0.060), but neither reached the conventional level of statistical significance (p < 0.05). Notably, for the Aβ1‑42 marker, which according to literature data may be associated with age, no significant correlation was found (ρ = 0.125, p = 0.270). Thus, age was not a substantial confounding factor for the analyzed biomarkers in this study. In the SCH group, no statistically significant differences between males and females were found for any of the biomarkers (Mann–Whitney U test, all p > 0.05). The closest approach to the significance threshold was observed for NCAM-1 (p = 0.050), but after correction for multiple comparisons this result did not reach statistical significance. Analysis for GFAP and pTau181 was not performed due to zero variability of these indicators. In the CON group, a statistically significant difference between males and females was found only for NCAM-1: marker levels were higher in males (M = 99.6; 75.9; 135.6) compared to females (M = 71.7; 62.4; 105.5); U = 997.0, p = 0.006. For the remaining biomarkers, no significant sex differences were found (all p > 0.05). Analysis for GFAP and pTau181 was not performed due to zero variability. Spearman correlation analysis in the control group revealed only one significant association: a negative correlation between age and total tau protein levels (ρ = − 0.467, p = 0.012). For the remaining biomarkers, including Aβ1–42, BDNF, YKL-40, and NCAM-1, no significant age-related changes were found (all p > 0.05). In the FEP group, Spearman correlation analysis revealed a significant negative association between age and NCAM-1 (ρ = − 0.450, p = 0.001) and a positive association with Neurogranin (ρ = 0.306, p = 0.034). For the remaining biomarkers, including YKL-40, BDNF, TDP-43, Aβ1–42, and GFAP, no significant correlations with age were found (all p > 0.05). A statistically significant difference between sexes was found for only one biomarker — NCAM-1. NCAM-1 levels were significantly lower in males compared to females (U = 148.000, p = 0.003). For all other studied biomarkers, including KLK6, UCHL1, NF-H, YKL-40, CNTF, Aβ1–42, GDNF, total tau, NGF-β, BDNF, MIF, TDP-43, Neurogranin, and FGF21, no statistically significant differences between males and females were found (all p > 0.05). GFAP and pTau181 variables were excluded from the analysis due to lack of variability in the group. To assess the potential effect of illness duration on the studied biomarkers, Spearman correlation analysis was performed between disease duration and serum concentrations of all measured proteins. No statistically significant correlations were found for any of the biomarkers (all p > 0.05). The highest correlation coefficient was observed for UCH-L1 (ρ = 0.229, p = 0.042), but this result did not reach the conventional level of statistical significance (p < 0.05). Thus, illness duration in the analyzed sample was not a significant confounding factor for the studied biomarkers. A principal component analysis (PCA) was conducted on seven protein biomarkers (BDNF, KLK6, MIF, Ncam, Neurogranin, TDP43, YKL40) to identify underlying patterns and reduce data dimensionality. The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy was 0.621, indicating middling but acceptable factorability. Parallel analysis based on simulated data indicated that only the first two components had eigenvalues greater than those from random data. Therefore, two components were retained for rotation. Together they accounted for 46.9% of the total variance (Component 1: 27.2%, Component 2: 19.7%) (Fig. 2A). Varimax rotation was applied to improve interpretability. The rotated component loadings are presented in (Fig. 2C). Component 1 showed strong positive loadings from MIF (0.745), BDNF (0.671), TDP43 (0.589), and YKL40 (0.585). Component 2 was predominantly loaded by Neurogranin (0.731), KLK6 (0.726), and NCAM-1 (0.619). The uniqueness values (variance not explained by the components) ranged from 0.442 to 0.641, indicating moderate to high specificity of each variable beyond the two common factors. The biplot of the first two principal components revealed substantial overlap among the three diagnostic groups (Fig. 2B). While Component 1 (27.2% of variance) captured a trend where the CON group tended to have slightly higher scores (positive values) compared to the patient groups, the distributions of all groups overlapped considerably along this axis. Component 2 (19.7% of variance) did not show any visually discernible separation between the FEP and SCH groups. The extensive overlap suggests that, their combined values do not provide clear, categorical separation between the diagnostic groups in this sample. Fig. 2PCA of seven protein biomarkers (BDNF, KLK6, MIF, Ncam, Neurogranin, TDP43, YKL40) across three diagnostic groups: healthy controls (CON), first‑episode psychosis (FEP), and schizophrenia (SCH). (A) Scree plot showing the percentage of variance explained by each principal component. The first two components (PC1 and PC2) were retained for further analysis, accounting for 27.2% and 19.7% of the total variance, respectively. (B) PCA biplot of individuals (points) in the space defined by PC1 (Dim1) and PC2 (Dim2). (C) Heatmap of varimax‑rotated component loadings for PC1 and PC2 PCA of seven protein biomarkers (BDNF, KLK6, MIF, Ncam, Neurogranin, TDP43, YKL40) across three diagnostic groups: healthy controls (CON), first‑episode psychosis (FEP), and schizophrenia (SCH). (A) Scree plot showing the percentage of variance explained by each principal component. The first two components (PC1 and PC2) were retained for further analysis, accounting for 27.2% and 19.7% of the total variance, respectively. (B) PCA biplot of individuals (points) in the space defined by PC1 (Dim1) and PC2 (Dim2). (C) Heatmap of varimax‑rotated component loadings for PC1 and PC2 To account for the effects of age, sex, BMI, disease duration, and antipsychotic class, multivariable linear regression with logarithmic transformation of biomarkers was performed. Comparison was made between the FEP and SCH groups (reference – SCH). After adjustment, significant differences remained for the following indicators: S100B – higher in FEP (B = 0.524, p < 0.001); UCHL1 – higher in FEP (B = 0.500, p = 0.002), also higher with typical antipsychotic use (B = 0.437, p = 0.010); Aβ1‑42 – higher in FEP (B = 1.217, p = 0.006), higher with typical antipsychotic use (B = 1.356, p = 0.004), and lower with increasing BMI (B = − 0.074, p = 0.040); MIF – lower in FEP (B = − 0.439, p = 0.025); NCAM-1 – the model was significant (p = 0.007), but the group difference did not reach the threshold (B = 0.163, p = 0.197); a sex effect was found (lower in females, p = 0.024); Neurogranin – the model was significant (p = 0.002), with no group difference (p = 0.483); significant effects were found for age (B = 0.045, p = 0.002), sex (lower in females, p = 0.023), and mixed therapy (lower, p = 0.046). For the remaining biomarkers (KLK6, GFAP, NF-H, YKL-40, CNTF, GDNF, total tau, pTau181, NGF-β, BDNF, FGF21, TDP-43), no statistically significant differences between FEP and SCH were found after adjustment (all p > 0.05). Age and BMI acted as independent predictors for some biomarkers (e.g., for KLK6: age p = 0.007, BMI p < 0.001). Thus, the observed differences between FEP and SCH for S100B, UCHL1, Aβ1‑42, and MIF are not explained by covariates and may reflect stage-dependent pathological processes. Additionally, we conducted a correlation analysis between quantitative blood biomarkers in each group separately. After applying the strict significance threshold (p < 0.01), the following significant correlations were identified (Table 4). Table 4Summary of significant biomarker correlations across study groupsProtein 1Protein 2 r p-valuer² Control Group KLK6Ncam0.3080.0050.095MIFYKL-400.2890.0090.084NeurograninKLK60.3470.0020.120 Schizophrenia Group BDNFYKL-400.4923.61e-060.242MIFYKL-400.2990.0070.089MIFNcam0.3020.0060.091MIFBDNF0.3990.00020.159TDP-43BDNF0.3340.0020.112NeurograninKLK60.3780.00050.143NeurograninNcam0.3350.0020.112 FEP Group TDP43YKL400.6291.28e-060.396Note: Only correlations with a p-value < 0.01 are reported (Pearson correlation). The coefficient of determination (r²) is provided as a measure of shared variance between each biomarker pair. Abbreviations: BDNF, Brain-derived neurotrophic factor; FEP, First-episode psychosis; KLK6, Kallikrein-6; MIF, Macrophage migration inhibitory factor; Ncam, Neural cell adhesion molecule; TDP-43, TAR DNA-binding protein 43; YKL-40, Chitinase-3-like protein 1 Table 4Summary of significant biomarker correlations across study groupsProtein 1Protein 2 r p-valuer² Control Group KLK6Ncam0.3080.0050.095MIFYKL-400.2890.0090.084NeurograninKLK60.3470.0020.120 Schizophrenia Group BDNFYKL-400.4923.61e-060.242MIFYKL-400.2990.0070.089MIFNcam0.3020.0060.091MIFBDNF0.3990.00020.159TDP-43BDNF0.3340.0020.112NeurograninKLK60.3780.00050.143NeurograninNcam0.3350.0020.112 FEP Group TDP43YKL400.6291.28e-060.396Note: Only correlations with a p-value < 0.01 are reported (Pearson correlation). The coefficient of determination (r²) is provided as a measure of shared variance between each biomarker pair. Abbreviations: BDNF, Brain-derived neurotrophic factor; FEP, First-episode psychosis; KLK6, Kallikrein-6; MIF, Macrophage migration inhibitory factor; Ncam, Neural cell adhesion molecule; TDP-43, TAR DNA-binding protein 43; YKL-40, Chitinase-3-like protein 1 Summary of significant biomarker correlations across study groups Note: Only correlations with a p-value < 0.01 are reported (Pearson correlation). The coefficient of determination (r²) is provided as a measure of shared variance between each biomarker pair. Abbreviations: BDNF, Brain-derived neurotrophic factor; FEP, First-episode psychosis; KLK6, Kallikrein-6; MIF, Macrophage migration inhibitory factor; Ncam, Neural cell adhesion molecule; TDP-43, TAR DNA-binding protein 43; YKL-40, Chitinase-3-like protein 1 The correlation analysis at a stringent threshold (p < 0.01) revealed distinct biomarker interaction patterns across groups. Healthy controls showed only three modest correlations, the strongest being between Neurogranin and KLK6 (r = 0.347). In contrast, chronic schizophrenia patients exhibited the most extensive network, with seven significant correlations, most notably a moderate positive association between BDNF and YKL40 (r = 0.492) and multiple links involving MIF. The FEP group displayed a single, but strong, positive correlation between TDP43 and YKL40 (r = 0.629), which was the most robust association observed in the entire study. To statistically test the differences in correlation patterns between groups, Fisher’s z-transformation was performed. The correlation between TDP-43 and YKL-40 was significantly higher in the FEP group compared to the SCH group (z = 2.93, p = 0.003). In contrast, the correlations of MIF with BDNF (z = 2.42, p = 0.016) and BDNF with YKL-40 (z = 2.07, p = 0.038) were significantly higher in the SCH group compared to FEP. These results confirm that the observed differences in correlation networks between groups are not an artifact of visual perception or different sample sizes, but reflect statistically significant changes in intermolecular interactions at different stages of the disease. Discussion This cross-sectional study provides a comparative assessment of non-canonical serum biomarkers in FEP relative to chronic schizophrenia and healthy controls. Our findings suggest that FEP is characterized by a distinct biomarker profile that differs from both chronic schizophrenia and healthy states. However, due to the cross-sectional design, we cannot determine whether this profile represents a transient early phenomenon or a stable trait. The results should be interpreted as hypothesis-generating. Multivariable regression analysis adjusting for age, sex, BMI, and antipsychotic treatment revealed that several initially observed differences between FEP and SCH were no longer statistically significant. Specifically, lower YKL-40 and BDNF levels in FEP did not survive correction for covariates, suggesting that these differences may be partly explained by demographic or treatment-related factors rather than illness stage per se. In contrast, elevated S100B, UCHL1, and Aβ1‑42, as well as decreased MIF, remained significantly different between FEP and SCH after adjustment. This pattern suggests that these four biomarkers may be more robustly associated with illness stage and warrant further investigation as potential stage-dependent markers. The most pronounced distinction in the FEP group was a significant reduction in serum protein YKL-40 (CHI3L1) levels compared to both healthy controls and patients with chronic schizophrenia. YKL-40, a glycoprotein secreted by macrophages, chondrocytes, and other cell types, is implicated in inflammation, tissue remodeling, and fibrosis, though its precise biological function and receptor remain unclear [15]. It is considered a non-specific biomarker of systemic inflammation and remodeling, with levels increasing with age and in various inflammatory conditions [16]. Genetic association studies link CHI3L1 variants to schizophrenia [17, 18], and altered YKL-40 concentrations have been reported in psychiatric disorders. Notably, cerebrospinal fluid (CSF) YKL-40 levels were found to be lower in schizophrenia patients, particularly in those receiving psychotropic medication [19]. In contrast, most peripheral blood studies report elevated YKL-40 in psychotic states. Serum YKL-40 is increased in both chronic and first-episode schizophrenia compared to controls [20, 21], and this elevation correlates with cognitive impairment [20]. The sole previous study in FEP reported increased plasma YKL-40, unrelated to medication, alongside monocytosis and a non-significant CSF increase [22]. Our finding of decreased serum YKL-40 in FEP contrasts with some existing literature. This discrepancy may indicate cohort-specific characteristics or a distinct biological process in the acute phase of psychosis. While elevated YKL-40 is typically associated with active neuroinflammation, its reduction in FEP may suggest an alternative scenario: a specific dysregulation or insufficient activation of innate immune cells (e.g., macrophages/microglia) at disease onset. This could represent a transient feature of the untreated acute phase, which later shifts toward elevation with illness chronicity or treatment initiation, consistent with findings of lower CSF YKL-40 in medicated patients [19]. The contradiction with Orhan et al. (2018) may also point to biological heterogeneity within FEP, implying different pathophysiological subtypes [22]. However, after adjusting for covariates in multivariable regression, this difference was no longer statistically significant. Thus, the observed effect may be partially confounded by demographic or treatment variables, and the independent association of YKL-40 with FEP remains to be confirmed in larger, well-controlled studies. We observed a significant elevation of serum NCAM-1 levels in the FEP group compared to the SCH cohort. NCAM-1, an immunoglobulin superfamily glycoprotein expressed on neurons and glia, is a key regulator of synaptic plasticity, neurite outgrowth, and neuronal migration [23, 24]. Dysregulation of NCAM-1 signaling is implicated in the pathophysiology of schizophrenia [23]. Our finding of elevated NCAM-1 in FEP contrasts with most prior studies, which predominantly report decreased peripheral NCAM-1 levels in both chronic schizophrenia and FEP compared to healthy controls [24–26]. One notable exception is a study linking higher serum NCAM-1 to better cognitive performance in schizophrenia [27]. A critical confounding factor is the suppressive effect of antipsychotic treatment on NCAM-1 levels, which has been demonstrated previously. Reductions have been specifically associated with second-generation antipsychotics [28], and lower NCAM-1 levels in CSF and blood were found only in medicated FEP patients [26]. Therefore, the elevated NCAM-1 in our FEP cohort may reflect a pharmacologically unmodulated state, capturing a distinct early-disease biological signature. This elevation could be interpreted in two ways: as an indicator of compensatory synaptic repair and adaptive neuroplasticity or, conversely, as a marker of aberrant synaptic remodeling inherent to the pathology. The discrepancy with the broader literature may also underscore the biological heterogeneity of psychotic disorders. In summary, NCAM-1 emerges as a dynamic biomarker sensitive to both disease stage and antipsychotic exposure. Our results highlight the necessity of rigorously accounting for medication status in future studies of synaptic biomarkers in psychosis. The observed reduction in BDNF in the FEP group in unadjusted analyses is consistent with previous literature [29–33]. However, after multivariable adjustment, this difference was no longer statistically significant. This suggests that lower BDNF in FEP may be influenced by factors such as BMI, sex, or medication status, which have been previously shown to affect BDNF levels. Nevertheless, the direction of the effect is consistent with the hypothesis of insufficient neurotrophic support in early psychosis, and future studies should carefully control for potential confounders. We identified an intriguing biomarker profile in FEP, characterized by divergent serum concentrations of molecules involved in overlapping functional pathways. Specifically, the concurrent elevation of serum NCAM-1 and reduction in serum BDNF suggests a state of dysregulated neuroplasticity in early psychosis. This pattern may indicate active, yet maladaptive, synaptic remodeling (reflected by high NCAM-1) occurring in the context of deficient neurotrophic support for circuit stabilization and maturation (reflected by low BDNF). Importantly, these are peripheral measurements; their precise relationship to central nervous system processes in the brain requires direct investigation. Nonetheless, this serum biomarker profile supports the concept of imbalanced plasticity mechanisms at disease onset. Our analysis identified significant differences in two proteins typically linked to neurodegenerative processes: TDP-43 and Aβ1–42. While proteostasis failure and pathological protein aggregation (e.g., amyloid, tau, α-synuclein) are defining features of neurodegenerative disorders, the role of specific proteinopathies in psychiatric illnesses remains unclear. However, emerging evidence suggests that aberrant protein accumulation may contribute to the pathophysiology of mental disorders, including schizophrenia [34]. We observed a decrease in serum TDP-43 concentration in the FEP group. TDP-43 is a multifunctional nuclear protein critical for RNA splicing and chromatin structure [35]. Pathological TDP-43 aggregation into ubiquitin-positive cytoplasmic inclusions is characteristic of amyotrophic lateral sclerosis and behavioral variant frontotemporal dementia [36, 37]. Importantly, both excessive and insufficient TDP-43 levels impair neuronal function and survival. To our knowledge, this study provides the first evidence of altered peripheral TDP-43 in early psychosis. A second key finding was a significantly higher frequency of detectable plasma Aβ1–42 in the FEP cohort, suggesting dysregulated amyloid precursor protein (APP) metabolism or impaired amyloid clearance during acute psychosis [38]. This indicates a potential molecular overlap with neurodegenerative pathways. These alterations occurred in young FEP patients in the absence of changes in classical neurodegenerative markers (tau, GFAP). Therefore, they likely reflect states of acute neuronal stress, synaptic dysfunction, or compensatory responses rather than progressive neuronal loss. This shifts the pathogenic focus from degenerative models toward frameworks emphasizing synaptic and cellular dysregulation in early psychosis. Furthermore, the unchanged levels of astrogliosis markers (GFAP, S100B), axonal injury markers (NF-H), and tau pathology markers (total Tau, p-tau181) across all groups consistently argue against a model of progressive neuronal death in schizophrenia. Collectively, these findings shift the pathogenic focus away from neurodegeneration and towards hypotheses centered on neurodevelopmental origins and synaptic dysfunction. Notably, the SCH group did not differ from healthy controls in the levels of the key identified biomarkers (YKL-40, NCAM-1, BDNF, TDP-43, Aβ1–42). This apparent normalization in the chronic phase may be attributable to several factors, including long-term adaptation of immune and neurotrophic systems to the disease state, the modulating effects of prolonged antipsychotic treatment, and the substantial clinical and biological heterogeneity within the chronic cohort, which encompasses patients with varying illness severity and treatment responses. Correlation patterns differed across groups. Healthy controls showed few correlations. In chronic schizophrenia, a denser network was observed, particularly involving MIF and BDNF. In FEP, the strongest correlation was between TDP-43 and YKL-40. These patterns suggest that biomarker interactions may change with illness stage, but longitudinal studies are needed to confirm this interpretation. The PCA revealed a two-dimensional structure underlying the seven examined protein biomarkers. The first component, loaded by MIF, BDNF, TDP43, and YKL40, the second component, defined by Neurogranin, KLK6, and NCAM. These components together account for nearly half of the total variance in the data. The PCA biplot indicates that these components do not provide clear categorical separation between the diagnostic groups. First, the substantial overlap between diagnostic groups in the PC1-PC2 space suggests that these seven biomarkers, while informative, are insufficient to reliably distinguish schizophrenia patients from healthy controls or FEP patients. This finding aligns with the clinical heterogeneity of schizophrenia and supports the notion that current biomarker panels may lack the specificity needed for diagnostic applications. Second, the observation that PC1 and PC2 together explain less than 50% of the total variance indicates considerable residual heterogeneity not captured by the first two components. The observed overlap aligns with the known clinical and biological heterogeneity of psychotic disorders. Rather than reflecting a clear boundary between health and illness, the biomarker profiles appear to exist along a continuum, with considerable individual variability within each group. These results indicate that while PCA can reveal underlying data structure, it does not support diagnostic classification of individual patients using this biomarker panel. Our cross-sectional data suggest that biomarker profiles differ between FEP and chronic schizophrenia, which is compatible with stage-dependent pathophysiology. However, definitive conclusions about stage dependence would require longitudinal studies tracking biomarker changes within individuals over time. Based on our findings, the identified biomarker signature shows potential clinical utility for differentiating FEP from both healthy controls and chronic disease states, which could aid in early patient stratification. However, this requires validation in larger, independent cohorts. Furthermore, the results point to novel targets for early intervention, such as modulation of the YKL-40/CHI3L1 pathway and strategies to enhance neurotrophic support. A key question for future research is the longitudinal dynamics of these changes. It is essential to determine whether the unique FEP profile represents a transient reaction to acute psychosis or serves as a prognostic biological signature that predicts transition to chronic illness and defines the subsequent clinical trajectory. We did not find significant correlations between biomarker levels and illness duration within the schizophrenia group. This absence of association may reflect the clinical and biological heterogeneity of chronic schizophrenia, where long-term illness course is influenced by multiple factors including treatment history, comorbid conditions, and individual variability. Alternatively, it may suggest that the biomarker alterations we identified occur early and then stabilize, rather than progressing linearly with time. Longitudinal studies are needed to distinguish between these possibilities. This cross-sectional study is limited by its inability to establish causality or track biomarker dynamics over time. The modest sample size, particularly in the FEP group, and its restriction to an ethnically homogeneous population may affect statistical power and generalizability. Pharmacological confounders are significant, as differences in antipsychotic type and exposure between the FEP and chronic schizophrenia groups could influence biomarker levels independently of disease stage. Finally, the reliance on peripheral blood measurements limits direct conclusions about central nervous system processes, as the correlation between serum and brain levels of the studied biomarkers remains uncertain. Several additional limitations should be acknowledged. First, we did not measure APOE4 genotype or Aβ1‑40, which could have provided additional context for interpreting the Aβ1‑42 findings. Second, correlation analyses were exploratory and not adjusted for multiple comparisons; therefore, these results should be interpreted as hypothesis-generating. Third, the lack of significant correlations between biomarkers and illness duration may be due to limited statistical power or the heterogeneity of the chronic schizophrenia group. This study is intended as a hypothesis-generating, proof-of-concept investigation. Rather than claiming definitive answers, we aim to draw attention to an underexplored set of biomarkers in FEP — those related to neurodegeneration, proteostasis, and synaptic integrity. The panel we used, while commercially available, is biologically grounded in prior literature linking these proteins to psychosis. We acknowledge that this targeted approach has limitations. However, we believe that the primary contribution of this work is not to provide clinical-grade biomarkers but to demonstrate that these pathways are altered in FEP and warrant systematic investigation using larger, independent cohorts and more comprehensive (e.g., untargeted) methodologies. Conclusion This cross-sectional study compared serum levels of 18 non-canonical biomarkers in patients with first-episode psychosis (FEP), chronic schizophrenia (SCH), and healthy controls (CON). After adjusting for age, sex, BMI, and antipsychotic treatment, the most robust differences between FEP and SCH were elevated S100B, UCHL1, and Aβ1‑42, as well as decreased MIF, in FEP. In contrast, unadjusted differences in YKL-40 and BDNF did not survive correction for covariates, suggesting that these may be influenced by demographic or treatment-related factors. Correlation analysis revealed distinct patterns across groups: FEP showed a strong correlation between TDP-43 and YKL-40, while SCH exhibited a dense interconnected network involving MIF and BDNF. Principal component analysis identified two biologically plausible axes (neuroinflammatory-neurotrophic and synaptic) but showed substantial overlap between diagnostic groups, indicating that this biomarker panel does not provide clear categorical separation. Collectively, these cross-sectional findings suggest that FEP has a distinct biomarker profile compared to chronic schizophrenia, challenging the view of FEP as merely a prodromal stage. However, due to the cross-sectional design, causal inferences cannot be made, and the results should be interpreted as hypothesis-generating rather than definitive. The absence of significant correlations between biomarkers and illness duration within the SCH group, as well as the exploratory nature of correlation analyses without multiple testing correction, are additional limitations. Future longitudinal studies are essential to determine whether the observed differences represent transient early phenomena or stable trait markers, and to establish whether the FEP biomarker profile predicts transition to chronic illness. Untargeted proteomic approaches in larger, independent cohorts are needed to discover novel biomarkers beyond the limited panel used here.