Introduction
Severe mental disorders (SMDs), including schizophrenia spectrum disorders (SCZ) and bipolar disorder (BD), share etiopathogenic factors and clinical characteristics such as psychotic symptoms, cognitive impairment, and mood dysregulation [Reference Dines, Kes, Ailán, Cetkovich-Bakmas, Born and Grunze1]. Their pharmacological treatment also overlaps considerably, with antipsychotic- (AP), antidepressant- (AD), and mood stabilizing agents [MS; antiepileptics (AE) and lithium] prescribed across diagnoses [Reference Baldessarini, Tondo and Vázquez2–Reference Johnsen and Kroken4]. While modulation of neurotransmitters is a major target for SMDs pharmacotherapy [Reference Ashok, Marques, Jauhar, Nour, Goodwin and Young5, Reference Howes and Kapur6], the mechanisms behind treatment effects are not completely understood. One area of growing interest is their potential immunomodulatory effects, given the increasing evidence implicating the immune system in SMDs.
Involvement of immune and inflammatory pathways in SMDs pathophysiology is indicated by a large volume of data, that is associations with infections and autoimmune diseases [Reference Benros, Nielsen, Nordentoft, Eaton, Dalton and Mortensen7, Reference Jeppesen and Benros8], alterations in peripheral and central immune factors [Reference Chen, Tan and Tian9–Reference Kroken, Sommer, Steen, Dieset and Johnsen11], and links to immune-related genetic loci [Reference Mullins, Forstner, O’Connell, Coombes, Coleman and Qiao12–Reference Ebrahimi, Teymouri, Chen, Mohiuddin, Pouget and Goncalves14]. However, the extent to which psychotropic agents impact the altered immune activity and inflammation remains unclear.
While the use of AP, AD, and MS has been linked to immunomodulatory effects [Reference Marcinowicz, Wiedlocha, Zborowska, Debowska, Podwalski and Misiak15–Reference Szalach, Lisowska, Cubala, Barbuti and Perugi32], small sample sizes and significant heterogeneity of study designs and findings indicate that additional studies are required [Reference Goldsmith, Rapaport and Miller33, Reference Bhatt, Dhar, Samanta and Suttee34]. A recent meta-analysis based on individuals with first-episode psychosis suggests anti-inflammatory properties of AP by modulating both pro- and anti-inflammatory cytokines, including interleukin (IL)-1β, IL-4, IL-6, IL-10, interferon (IFN)-γ, and tumor necrosis factor (TNF)-α [Reference Marcinowicz, Wiedlocha, Zborowska, Debowska, Podwalski and Misiak15, Reference Subbanna, Shivakumar, Venugopal, Narayanaswamy, Berk and Varambally22, Reference Tourjman, Kouassi, Koué, Rocchetti, Fortin-Fournier and Fusar-Poli24, Reference Goldsmith, Rapaport and Miller33, Reference Bhatt, Dhar, Samanta and Suttee34]. Although based on longitudinal data, the individual sample sizes were generally small with heterogenous treatment regimens. Similarly, immunomodulatory effects have been indicated for AD; however, the evidence is conflicting and primarily based on studies in major depressive disorder (MDD) [Reference Wang, Wang, Liu, Qiao, Baldwin and Hou20, Reference Baumeister, Ciufolini and Mondelli21, Reference Bhatt, Dhar, Samanta and Suttee34–Reference Szalach, Lisowska and Cubala37]. A recent review reported that associations between use of various ADs and levels of IFN-γ, C-reactive protein (CRP), IL-1β, IL-1α, IL-6, and TNF-α varied from positive to negative and non-significant in MDD [Reference Mosiołek, Pięta, Jakima, Zborowska, Mosiołek and Szulc36]. To the best of our knowledge, there are only a couple of studies of AD use and immune markers in SCZ [Reference Ding, Li and Liu38] and BD [Reference Edberg, Hoppensteadt, Walborn, Fareed, Sinacore and Halaris39], reporting negative associations between escitalopram use and levels of CRP and IL-6, and no associations between escitalopram use on CRP levels, respectively. For MS, a systematic review found that lithium use in BD was associated with increased levels of IL-4 and TNF-α, whilst no associations were found with levels of IL-1β, IL-6, and IL-8 [Reference van den Ameele, van Diermen, Staels, Coppens, Dumont and Sabbe26]. The number of studies was insufficient to draw inferences regarding AE [Reference van den Ameele, van Diermen, Staels, Coppens, Dumont and Sabbe26, Reference Szalach, Lisowska, Cubala, Barbuti and Perugi32]. A more recent study indicated reduced levels of IL-1β, MIF, and IL-6 in a BD group treated with lamotrigine compared to a control BD group treated with sodium valproate [Reference Shi, Li, Song and Wang40]. However, existing studies are often limited by small or modest sample sizes, focus on individual agents, and lack of dose or serum concentration data.
To address these limitations, we aimed to clarify associations between use of psychotropic agent classes and immune marker levels in SMDs by using a large sample, and comprehensive assessment of immune markers, including dose and serum analyses, and making extensive covariate adjustments. We adopted a novel approach by grouping patients according to the medication class used. This may yield more robust results and reflects the perspective that clinicians, in large part, relate to medication classes regarding treatment efficacy in SMDs. We hypothesized that use of AP, MS, and AD would be associated with changes in immune marker levels, indicating a reduction in immune and inflammatory dysregulation. Immune markers of healthy controls (HC) were included as a reference.
Methods
Study setting
The sample comprises patients and HC from the Thematically Organized Psychosis (TOP) study in Oslo, Norway. Patients are continuously recruited from major hospitals in Oslo if they meet the criteria for SCZ and BD according to the Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV [41]. HC are randomly selected from the population registry data within the same catchment area. General inclusion criteria are age 18–65 years and sufficient Scandinavian language skills to complete the study. Exclusion criteria include severe neurological or somatic illness affecting brain functioning, IQ < 70, or a history of severe head trauma. For HC, additional exclusion criteria are substance abuse or dependency, and SMDs in close relatives. Recruitment for this study took place between 2002 and 2018.
Sample
The sample comprises 2223 TOP study participants with available immune marker data: 777 with SCZ (schizophrenia: 443; schizophreniform disorder: 44; schizoaffective disorder: 111; and psychotic disorder not otherwise specified: 179), 438 with BD (bipolar I disorder: 271; bipolar II disorder: 143; and bipolar disorder not otherwise specified: 24), and 1008 HC. Participants with CRP levels >10.0 were excluded to minimize the influence of acute infection on immune marker levels.
Clinical assessments
Patient interviews were performed by trained clinical psychologists and medical doctors, obtaining sociodemographic, psychiatric, and somatic data. Diagnoses were established using the Structured Clinical Interview for DSM-IV Axis I Disorders, including modules for SMDs and substance use disorders [Reference Spitzer, Williams, Gibbon and First42]. Inter-rater reliability in the TOP study has shown kappa values of 0.92–0.99 [Reference Høegh, Melle, Aminoff, Olsen, Lunding and Ueland43]. Symptom severity was evaluated with the Positive and Negative Syndrome Scale (PANSS) [Reference Kay, Fiszbein and Opler44]. Somatic examinations, including height and weight for body mass index (BMI) calculations, were also performed.
Psychotropic agents
Information on psychotropic agent use and dose at inclusion was collected through interviews and medical records. Psychotropic agent classes “AP use,” “AD use,” and “MS use (AE and lithium)” were dichotomized as “use” (participants using one or more agents within the class) versus “no use.” Patient participants could be included in multiple classes. Defined daily doses (DDD; WHO, https://www.whocc.no/atc_ddd_index/) were summed within each psychotropic agent class (“AP DDD,” “AD DDD,” and “MS DDD”).
Serum concentrations were measured from fasting blood samples, with a mean sampling time of 09:42 (median 09:30, range 07:30–15:15), drawn from the antecubital vein into serum vials without gel within two weeks of symptom assessment. Samples (≥ 0.5 mL) were centrifuged, stored at 4 °C, and transported to the laboratory at the Department of Clinical Pharmacology, St. Olav University Hospital, Norway, within the next working day [see [Reference Steen, Aas, Simonsen, Dieset, Tesli and Nerhus45–Reference Steen, Aas, Simonsen, Dieset, Tesli and Nerhus47]. Relative serum concentrations were calculated by dividing each serum concentration by the midpoint of its reference interval [Reference Hiemke, Bergemann, Clement, Conca, Deckert and Domschke48]; these were then summed within each class to yield total serum concentration. Details on psychotropic agent use and concentrations are presented in Table 1 and Supplementary Table 1. Information on somatic agent use, including anti-inflammatory and cardiometabolic agents, is provided in Supplementary Table 2.
Sample descriptives
Abbreviations: AD, antidepressant agent; AD, antiepileptic agent; AP, antipsychotic agent; BD, bipolar disorders; BMI, body mass index; DDD, defined daily dose; HC, healthy controls; IQR, interquartile range; M, median; MS, mood stabilizing agent; PANSS, Positive and Negative Syndrome Scale; SCZ, schizophrenia spectrum disorders; no AP/AD/MS, using no agent; one AP/AD/MS, using at least one agent; two AP/AD/MS, using at least two agents; Three AP/AD/MS, using at least three agents.
Note: Patients using psychotropic agents [median (IQR)], a) DDD and b) serum concentrations: SCZ: a) AP 1.00 (1.00), AD 1.00 (1.00), MS 0.67 (0.69), and b) AP 0.56 (0.70), AD 0.34 (0.51), MS 0.49 (0.50); BD: a) AP 0.67 (0.67), AD 1.00 (1.25), MS 1.00 (0.83), and b) AP 0.30 (0.40), AD 0.32 (0.52), MS 0.66 (0.38). Group comparisons a) AP p < 0.001 (BD < SCZ), AD p = 0.766, MS p = 0.020 (SCZ < BD), and b) AP p < 0.001 (BD < SCZ), AD p = 0.032 (BD < SCZ), MS p < 0.001 (SCZ < BD).
a AP: amisulpride, aripiprazole, dehydroaripiprazole, chlorpromazine, chlorprothixene, clozapine, flupentixol, haloperidol, levomepromazine, olanzapine, paliperidone, perphenazine, quetiapine, risperidone, ziprasidone, zuclopenthixol.
b Serum concentration AP, AD, AE, MS, lithium: the sum of relative serum concentrations [serum concentrations divided by the middle values of their reference intervals (Heimke et al. 2018)] within the psychotropic agent classes. Serum concentrations with M (IQR) given by [0.0 (0.0)], reflect that a large portion of the participants had no measurable serum concentrations of the respective agent classes. Serum concentration data and levels are based on laboratory measurements in combination with documented use or non-use of the agent.
c AD: amitriptyline, bupropion, citalopram, clomipramine, duloxetine, escitalopram, fluoxetine, mianserin, mirtazapine, paroxetine, reboxetine, sertraline, trimipramine, venlafaxine.
d AE: carbamazepine, clonazepam, gabapentin, lamotrigine, pregabalin, topiramate, valproate.
e Alimemazine, buspirone, diazepam, flunitrazepam, hydroxyzine, melatonin, nitrazepam, oxazepam, zolpidem, zopiclone.
f Atomoxetine, dexamphetamine, metylphenidate.
Immune markers
Blood samples for analysis of immune markers (Table 2, Supplementary Table 3) were drawn within two weeks of symptom severity assessment on EDTA vials. Plasma was isolated within the next working day and stored at −80 °C. For HC, the mean sampling time was 12:31 (median 11:20, range 07:20–19:00); in patients, sampling coincided with that for serum concentration analyses [Reference Morch, Dieset, Faerden, Hope, Aas and Nerhus49, Reference Hjell, Rokicki, Szabo, Holst and Tesli50]. Serum CRP was analyzed by immunoturbidimetry at the Department of Medical Biochemistry, Oslo University Hospital, Norway. The remaining immune markers were analyzed in duplicate using enzyme immunoassays (EIA) with R&D Systems antibodies (Minneapolis, MN, USA) in 384-well format, performed at the Research Institute of Internal Medicine, Oslo University Hospital. Assays were conducted using a Selma pipetting robot and a Biotek dispenser/washer. Absorbance was measured at 450 nm (with 540 nm correction) using a BIO-RAD (Hercules, CA, USA) ELISA plate reader. All EIAs had inter- and intra-assay coefficients of variation <10%. Previous publications have reported on immune marker variations and associations with psychotropic agent use in the TOP sample [e.g. [Reference Morch, Dieset, Faerden, Hope, Aas and Nerhus49, Reference Andreou, Steen, Jørgensen, Smelror, Wedervang-Resell and Nerland51–Reference Szabo, O’Connell, Ueland, Sheikh, Agartz and Andreou56]], including diurnal and postprandial variations [Reference Ormerod, Ueland, Werner, Hjell, Rodevand and Saether53, Reference Sæther, Ueland, Haatveit, Maglanoc, Szabo and Djurovic57–Reference Sæther, Szabo, Akkouh, Haatveit, Mohn and Vaskinn59]. All blood sampling was performed by a single laboratory.
List of 45 immune markers assessed
Statistical analysis
Statistical analyses were conducted using SPSS for Windows, version 29 (SPSS Inc., Chicago, IL, USA). Normality was assessed by inspecting histograms, Q–Q-plots, and Kolmogorov–Smirnov statistics. Sample descriptives and immune marker data were analyzed using chi-square tests for categorical variables and Kruskal-Wallis and Mann–Whitney U tests for continuous variables. Immune marker data were log10-transformed for use in linear regressions, and outliers were removed using the interquartile range method [Reference Jones60] (Supplementary Table 4).
Linear regressions were used to separately examine associations between pairs of immune markers (dependent variable) and AP, AD, and MS variables (‘use’ versus ‘no use’; independent variable). Main models were adjusted for age, sex, and BMI [Reference O’Connor, Bower, Cho, Creswell, Dimitrov and Hamby61], diagnosis, PANSS total score [Reference Yuan, Chen, Xia, Dai and Liu62], and plasma freezer storage time [Reference Simpson, Kaislasuo, Guller and Pal63]. A significance threshold of p < 0.001 was applied to address multiple testing concerns, to accommodate the need to minimize false positives while avoiding an excessive risk of rejecting potentially interesting associations [Reference Lydersen64, Reference Sjolander and Vansteelandt65]. Sensitivity analyses for significant associations included the following additional covariates: education (years) [Reference O’Connor, Bower, Cho, Creswell, Dimitrov and Hamby61], current smoking (yes/no) [Reference Saint-André, Charbit, Biton, Rouilly, Possémé and Bertrand66], duration of illness (years) [Reference Momtazmanesh, Zare-Shahabadi and Rezaei67], illicit substance use within the past two weeks (yes/no) [Reference Mitchell, El Jordi, Yamamoto, Aschner and Costa68], time of blood sampling, CRP (except in CRP models), and DDD and serum concentrations, respectively, of medication classes not specifically tested. Secondary linear regressions were conducted for identified pairs of psychotropic agent classes and immune markers, by substituting psychotropic agent class use with DDD or serum concentration, using the same covariates as the main analyses, and adding an adjustment for blood sampling time in the serum concentration analyses. Subgroup analyses of MS use were performed for AE without lithium and lithium without AE. Lastly, the following interactions were explored for each immune marker: APxAD, APxMS, ADxMS, and APxADxMS. Missingness in covariates or immune markers was handled by listwise exclusion.
Ethics
Participation in the TOP study required written informed consent. The study was approved by the Regional Committee for Medical and Health Research Ethics (2009/2485) and conducted in accordance with the Declaration of Helsinki.
Results
Sample characteristics
The SCZ group was younger than the BD and HC groups (p < 0.001), and the BD group had a lower proportion of male participants compared to SCZ and HC (p < 0.001). HC had a lower BMI than both patient groups (p < 0.001). PANSS total scores were higher in SCZ than BD (p < 0.001), and recent illicit substance use was more common in BD than SCZ (p = 0.021). Freezer storage time was shorter in HC than in the patient groups (p < 0.001). AP use was more frequent in SCZ than BD (p < 0.001), whereas AD use and MS use were more frequent in BD than SCZ (p = 0.008 and p < 0.001, respectively). See Table 1 for additional sample and psychotropic agent descriptives, including differences between groups.
Differences in levels of immune markers were found between SCZ and/or BD and HC for most immune markers (p < 0.05), except for BAFF, A2M, GFAP, VCAM-1, PSEL, IL-18RAP, CXCL16, OPG, ALCAM, MPO, vWF, PARK7, and sCD14. Specifically, BD-2, CRP, and sIL-2R levels were higher in both SCZ and BD compared to HC, and CRP levels were higher in SCZ than in BD. See Supplementary Tables 3 and 5 for details.
Associations between immune marker levels and psychotropic agent use, dose, and serum concentration
AP use was associated with higher plasma levels of BD-2 relative to no AP use (β = 0.094, p = 0.8E-4), AD use with higher serum levels of CRP (β = 0.072, p = 0.8E-3), and MS use with higher plasma levels of sIL-2R (β = 0.063, p = 0.9E-4). Table 3 shows effect estimates and p-values for various steps of adjustments. In subgroup analyses, sIL-2R was significantly associated with AE use (β = 0.061, p = 0.5E-3), and nominally with lithium use (β = 0.062, p = 0.024). Results of the sensitivity analyses showed consistent, though slightly attenuated associations (DDD and serum concentration adjustments, respectively: AP and BD-2, p = 0.002 and p = 0.005; AD and CRP, p = 0.003 and p = 0.009; MS and sIL-2R, p = 0.004 and p = 0.009, Supplementary Table 6). See Figure 1 for associations between immune marker levels and psychotropic agent use.
Association analyses between psychotropic agent class use and immune markers a
Abbreviations: AD, antidepressant agents; AE, antiepileptic agents; AP, antipsychotic agents; BD-2, beta defensin 2; BMI, body mass index; CRP, C-reactive protein; FT, freezer storage time; MS, mood stabilizing agents; PANSS, positive and negative syndrome score; sIL-2R, total score, soluble interleukin 2 receptor.
a Linear regression results reported with β (p-values) at different steps of covariate adjustments.
b Subgroup analyses: sIL-2R and AE use (β = 0.061, p = 0.5E-3) and lithium use (β = 0.062, p = 0.024).
c Model statistics: AP use and BD-2 (df = 7, F = 4.808, p < 0.001, R2 = 0.035), AD use and CRP (df = 7, F = 39.621, p < 0.001, R2 = 0.207), MS use and sIL-2R (df = 7, F = 5.353, p < 0.001, R2 = 0.051).
Associations between immune marker levels and psychotropic agent use. (A) Estimated marginal means (EMM) with 95% confidence intervals of log10-transformed immune marker levels by psychotropic agent class (AP, AD, MS), adjusted for age, sex, BMI, diagnosis, PANSS total score, and plasma freezer storage time. (B) Scatter plots showing unadjusted associations between defined daily dose (DDD) and log10-transformed immune marker levels. (C) Scatter plots showing unadjusted associations between serum concentrations and log10-transformed immune marker levels. Asterisk (*) indicates associations significant at p < 0.001. Abbreviations: Antidepressant agents (AD), Antipsychotic agents (AP), Defined daily dose (DDD), Estimated marginal means (EMM), Mood stabilizing agents (MS; antiepileptics and lithium).
The secondary analyses revealed that DDD AP was positively associated with BD-2 levels (β = 0.045, p = 2.3E-4), DDD AD was positively associated with CRP levels (β = 0.039, p = 0.001), and DDD MS was positively associated with sIL-2R levels (β = 0.048, p = 0.001). Serum concentration of AD was nominally associated with CRP (β = 0.072, p = 0.002), while serum concentrations of MS and AP were not significantly associated with sIL-2R (β = 0.038, p = 0.168) and BD-2 (β = 0.032, p = 0.088), respectively. Subgroup analyses of sIL-2R with serum concentrations of AE and lithium were not significant (β = 0.039, p = 0.245 and β = 0.033, p = 0.827, respectively). See Supplementary Table 7 for details.
No other significant associations were found between psychotropic agent classes and immune markers (Supplementary Table 8). Diagnosis-stratified results are presented in Supplementary Table 9.
Exploratory analyses identified an interaction between AD and MS use for ALCAM levels (β = −0.056, p = 0.001). None of the other tested interactions between psychotropic agent classes were significant across the 44 remaining immune markers.
Discussion
Our main findings were that use of AP, AD, and MS in individuals with SMDs was associated with elevated levels of BD-2, CRP, and sIL-2R, respectively. These associations were observed both when comparing users versus non-users of psychotropic agents, in relation to dose, and partially in relation to serum concentrations. No other significant associations were found between these psychotropic agent classes and the broad immune marker panel, suggesting minor and specific immunomodulatory associations of psychotropic agent use.
Beta-defensins, including BD-2, have extensive innate immune effects such as antimicrobial activity and immunomodulatory functions [Reference Williams, Castellani, Weinberg, Perry and Smith69]. BD-2 is thought to influence central nervous system immune processes through effects on dendritic cells, and reduced BD-2 expression has been associated with neurodegenerative conditions involving immune dysregulation and inflammation [Reference Williams, Castellani, Weinberg, Perry and Smith69]. While SMDs are primarily considered to have a neurodevelopmental origin, neurodegenerative features are also indicated [Reference Gupta, Advani, Yadav, Ambasta and Kumar70], potentially linked to positive symptoms [Reference Gupta and Kulhara71]. Notably, APs are suggested to have neuroprotective properties, including dampening of microglial activation [Reference Chen and Nasrallah72]. Moreover, evidence indicates peripheral immune homeostatic and anti-inflammatory effects of beta-defensins by e.g., cytokine signaling regulation [Reference Kim, Cho and Jang73]. AP may modulate immune activity through various mechanisms, including interactions with dopamine, serotonin, and acetylcholine receptors expressed on immune cells [Reference de Bartolomeis, Barone, Begni and Riva74]. However, there is limited research on the association between beta-defensins and SMDs or their treatment, and to the best of our knowledge, this is the first study to demonstrate an association with AP. Our finding is strengthened by the significant associations with dose, and one might speculate that AP use improves regulatory functions of BD-2 including mediating immune homeostatic, anti-inflammatory, and neuroprotective effects. Clinically, such effects could be relevant for symptom control and neuroprotection in subgroups of patients, particularly those with immune dysregulation or treatment resistance [Reference Pisanu, Severino, Minelli, Dierssen, Potier and Fabbri75].
CRP is an acute-phase protein induced by pro-inflammatory cytokines such as IL-1β and IL-6, reflecting non-specific systemic inflammation [Reference Fernandes, Steiner, Bernstein, Dodd, Pasco and Dean76]. Prior studies of AD use and CRP in SCZ and BD are limited. Ding et al. [Reference Ding, Li and Liu38] reported reduced CRP levels after 8 weeks of escitalopram add-on treatment in treatment-resistant SCZ, while Edberg et al. [Reference Edberg, Hoppensteadt, Walborn, Fareed, Sinacore and Halaris39] reported reductions in CRP between weeks 4 and 8 in BD with treatment-resistant depression, but no change between baseline and week 8. Discrepancies with the current findings may reflect different sample characteristics, such as treatment resistance, and differences in age and duration of illness. A meta-analysis of MDD found no consistent association between AD treatment and CRP [Reference Więdłocha, Marcinowicz, Krupa, Janoska-Jaździk, Janus and Dębowska77]. There are, however, some individual studies that support our findings, such as those by Chang et al. [Reference Chang, Lee, Gean, Lee, Chi and Yang78] and Carboni et al. [Reference Carboni, McCarthy, Delafont, Filosi, Ivanchenko and Ratti79], which found elevated levels of CRP with AD treatment. Associations between AD use and pro-inflammatory markers in MDD mainly include decreased levels of TNF-α and IL-6 following AD treatment [Reference Liu, Wei, Strawbridge, Bao, Chang and Shi80, Reference Köhler, Freitas, Stubbs, Maes, Solmi and Veronese81]. However, pro-inflammatory effects have been suggested through decreased levels of the anti-inflammatory IL-10 [Reference Köhler, Freitas, Stubbs, Maes, Solmi and Veronese81]. The mechanisms remain unclear, but might involve metabolic disturbances of AD or effects on immune cell serotonin receptors [Reference Szalach, Lisowska and Cubala37, Reference Vojvodic, Mihajlovic, Vojvodic, Radomirovic, Vojvodic and Vlaskovic-Jovicevic82, Reference Hamer, Batty, Marmot, Singh-Manoux and Kivimäki83]. Importantly, to the best of our knowledge, most single studies in the literature are smaller and not as well-adjusted as the current one. Our findings are further supported by the associations with dose and, suggestively, serum concentration, and one might speculate that pro-inflammatory effects of AD use in SCZ and BD lead to a less favorable clinical course in subgroups of patients. Notably, prior evidence links elevated CRP to poorer clinical outcomes in SMDs [Reference Fernandes, Steiner, Bernstein, Dodd, Pasco and Dean76, Reference Chamberlain, de Boer, Mondelli, Jones and Drevets84]. For instance, the proposed mania and rapid cycling inducing effects of AD in BD might be relevant to study in this context, given the inflammatory activity associated with these conditions [Reference Munkholm, Weikop, Kessing and Vinberg85, Reference Gottlieb and Young86]. The finding supports further investigation of CRP or related markers as potential indicators for monitoring or stratifying patients during AD treatment, and clinicians should keep in mind the potential for subtle negative clinical effects of AD treatment in the SMD population.
Elevated sIL-2R is generally considered to reflect T-cell activation by triggering of IL-2R production and shedding through elevated expression of IL-2 [Reference Dik and Heron87]. Several studies have demonstrated altered levels of IL-2 components in SMDs compared to HC [Reference Kroken, Sommer, Steen, Dieset and Johnsen11, Reference Goldsmith, Rapaport and Miller33, Reference Tsai, Yang, Kuo, Chen and Leu88]. Importantly, sIL-2R may act as a decoy receptor and dampen IL-2 signaling [Reference Dik and Heron87, Reference Russell, Moore, Fallon and Walsh89], potentially promoting anti-inflammatory effects [Reference Dik and Heron87]. Thus, previous findings of elevated sIL-2R in depressive, euthymic, and manic phases of BD, in addition to a positive correlation between sIL-2R and symptom severity in mania [Reference Tsai, Yang, Kuo, Chen and Leu88, Reference Breunis, Kupka, Nolen, Suppes, Denicoff and Leverich90], might indicate an inflammatory state with enhanced immune activation, but possibly also a pro-homeostatic compensatory mechanism. However, as IL-2 has pleiotropic immune functions, with both stimulatory and suppressive effects [Reference Abbas91], the interpretation of these findings remains complex. The limited literature on sIL-2R and psychotropic treatment in SMDs is inconclusive, including reduced sIL-2R levels following lithium treatment [Reference Rapaport, Guylai and Whybrow92], a non-significant lower level of sIL-2R after valproate treatment [Reference Maes, Bosmans, Calabrese, Smith and Meltzer93], and no clear association between MS use and sIL-2R in a recent review [Reference van den Ameele, van Diermen, Staels, Coppens, Dumont and Sabbe26]. The current sample is, to the best of our knowledge, by far the largest single sample in which the association between MS use and sIL-2R is examined, and included extensive covariate adjustment. The association was further supported by a correlation with dose and, nominally with serum concentration. While the positive association might indicate an inflammatory effect of MS use, potential anti-inflammatory effects of MS through increased shedding of membrane-bound IL-2R and increased levels of IL-2 decoy receptors are not implausible. Immunosuppressive effects of certain MS have been suggested through inhibition of protein synthesis and suppression of lymphocyte activity [Reference Beghi and Shorvon29]. Interestingly, levels of IL-2 have been reported to be reduced in vitro after treatment with MS [Reference Himmerich, Bartsch, Hamer, Mergl, Schönherr and Petersein28]. The clinical relevance of the current finding is unclear, but one might speculate that an immune homeostatic process may be contributing to the mood-stabilizing effect of MS.
The study has several strengths, including a large, well-characterized sample, a broad immune marker panel, and detailed psychotropic medication exposure data, allowing investigation of associations between psychotropic agent use and immune markers in SMDs with extensive adjustments. The immune markers reflect relevant pathways, although sgp130 and sTNF-R1 are less investigated components of the IL-6 [Reference Rose-John, Jenkins, Garbers, Moll and Scheller94] and TNF pathways [Reference Paccalet, Crola Da Silva, Mechtouff, Amaz, Varillon and de Bourguignon95]. Moreover, the selection includes markers linked with neuroinflammation, such as SERPINA3, BAFF, and YKL-40; however, peripheral measurements clearly limit the interpretation relative to central processes. Nevertheless, future studies could benefit from investigating broader panels linked to neuroinflammation and microglial activation. Information on psychotropic dose and serum concentrations enabled us to capture various aspects of exposure-immune associations, supporting the indicated associations with use. However, the study’s cross-sectional, non-randomized design limits causal inference; the larger TOP study also includes follow-up examinations of patients at varying intervals of years, but is not designed for rigorous longitudinal analysis of medication effects. Thus, although no follow-up is currently planned, longitudinal studies are needed to clarify temporal and causal relationships. Particularly, reverse causality cannot be excluded, despite the demonstrated associations across use, dose, and serum concentration; i.e., immune abnormalities may have influenced clinical presentation and thereby affected treatment decisions. Moreover, the lack of longitudinal data hinders examinations of associations related to changes in illness severity. Lastly, The sensitivity analyses showed attenuations in significance levels for all identified associations; however, there were relatively minor reductions in effect sizes. As the sensitivity analyses involved more than a doubling of number of covariates and some loss of participants due to missingness in covariate data, the results seem robust, and attenuations are likely related to the size and complexity as well as sample reductions. Nevertheless, the significant findings should be interpreted with caution given the large proportion of negative results.
In the current study, we demonstrated associations between elevated levels of BD-2, CRP, and sIL-2R and AP, AD, and MS use, respectively, in a large and well-characterized sample of individuals with SMDs. These findings, supported by associations dose, and, in part, serum concentrations, underscore the need to account for medication effects when interpreting immune alterations in SMDs. The associations suggest that psychotropic agents may modestly influence specific immune pathways of relevance for immune homeostasis and inflammatory regulation in SMDs. Importantly, our findings support the translational potential of immune-informed stratification in SMDs, aligning with emerging evidence that inflammatory biomarkers can help identify subgroups with differential treatment response [Reference Ortega, Fraile-Martinez, Garcia-Montero, Diaz-Pedrero, Lopez-Gonzalez and Monserrat96–Reference Wessa, Janssens, Coppens, El Abdellati, Vergaelen and van den Ameele99]. Nevertheless, the lack of associations for most immune markers indicates that psychotropic agents exert limited effects on broader immune signaling, and further longitudinal studies are warranted to clarify directionality and clinical implications.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1192/j.eurpsy.2025.10104.
Data availability statement
Due to ethical restrictions and the need to protect participant anonymity, the datasets generated and/or analyzed during the present study are not publicly available.
Acknowledgements
We thank the participants for their invaluable contributions, and all colleagues, particularly Isabel Viola Kreis, PhD, for her assistance with the matching procedure.
Financial support
This study was funded by the South-Eastern Norway Regional Health Authority (grant numbers 2019–108, 2017–112) and the Research Council of Norway (grant numbers 223273, 283798, 283799).
Competing interests
MBEGO, MAS, TU, GH, LR, LSS, SHL, ITJ, DA, TU, TVL, IM, SD, and NES declare that they have no conflicts of interest. OAA is a consultant to Cortechs.ai and Precision Health AS and has received Speaker’s honoraria from Lundbeck, Sunovion, Janssen, and Otsuka.
Open access
Comments
No Comments have been published for this article.