Study protocol for a randomized controlled trial with rituximab for psychotic disorder in adults (RCT-Rits)

Multiple sources of evidence indicate a role of inflammation and immunological mechanisms in schizophrenia [3‐8]. Autoimmune diseases and severe infections are risk factors for the later development of schizophrenia [9, 10], elevated inflammatory markers in childhood or adolescence are associated with a greater risk of schizophrenia in adulthood [11, 12], individuals with schizophrenia have increased levels of pro-inflammatory cytokines compared to healthy controls [13‐16], and autoimmune diseases are overrepresented in schizophrenia [17]. However, treatments with anti-inflammatory agents are so far of doubtful clinical relevance [18].

There is a strong genetic association between the major histocompatibility complex (MHC) locus and schizophrenia, established by genome-wide association studies (GWAS) [19]. The expression of the genetic loci linked to schizophrenia was enriched by B-lymphocytes involved in acquired immunity (CD19 and CD20 lines), providing strong genetic support for immune dysregulation, specifically implicating B-lymphocytes as a pathogenetic mechanism in schizophrenia. Further research has shown that these associations to a great extent are driven by the complement system [20‐23]. This system has a crucial role in synaptic pruning during brain development. Excessive pruning may disable important synaptic connections in the brain, resulting in a vulnerability to neuropsychiatric disorders. In a recent study, complement expression was related to cortical thinning in SSD [22]. Intriguingly, genetic links from schizophrenia to systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA) have been localised to the MHC region, despite divergent clinical manifestations [24, 25]. Interestingly, two cases of bone marrow transplantation further highlight the link between immune genetics and schizophrenia: One case was reported with complete remission from treatment-resistant schizophrenia after a bone marrow transplant [26]. Conversely, another reports the onset of treatment-resistant psychosis after receiving bone marrow from a donor with schizophrenia [27]. Accordingly, immunogenetics strongly supports the immune system’s involvement in schizophrenia pathogenesis and even suggests connections with chronic rheumatological diseases.

Increased plasma protein levels of cytokines and cytokine mRNA expression are widely reported in schizophrenia [13‐16]. Of the general proinflammatory cytokines, interleukin-6 (IL-6) is most consistently elevated and has been suggested as a trait marker of schizophrenia. Recently, changes in brain structure were genetically associated with IL-6 by Mendelian randomisation [28], indicating that IL-6 is of central aetiological relevance. Other proinflammatory cytokines, such as IL-1β, TNF-α and IFN-γ are also elevated in most studies. The anti-inflammatory cytokines IL-10 and TGF-β are also mostly elevated, indicating a dysregulated state of the immune system. Other cytokines reported as elevated or deviating in schizophrenia are IL-1 receptor antagonist (IL-1Ra), the soluble receptor of IL-2 (sIL-2R), IL-4, IL-8 and IL-17. Our research group has found similar associations [29]. We compared a mixed sample of markedly ill psychiatric patients (N = 39) with participants without psychiatric or somatic disease regarding inflammatory-related markers, using electrochemiluminescence ELISA and real-time qPCR technology. Fourteen of these patients suffered from SSD, and, in this small group, IL-1RA (p < 0.001), IL-18 (p = 0.006), IL-6 (p < 0.002) and TNF-α (p = 0.005) protein levels were statically significantly higher compared to matched controls. Also, the mRNA gene expression of CASP1 (caspase-1, part of the inflammasome complex) was elevated (p = 0.008). Other groups have found that inflammation has a closer association with negative symptoms, specifically reward processing and motivational deficits, than with positive symptoms [8]. The roles of astrocytes and regulatory T-cells (Tregs) have also been emphasised [30, 31]. Dysfunctional Tregs have been suggested as a pathophysiological mechanism that links schizophrenia to autoimmune disorders, such as RA and multiple sclerosis (MS). Flow cytometry of immune cells from 40 patients with schizophrenia [32] indicated dysfunctional Tregs and an increased percentage of B-cells (CD19 and CD20) compared to controls. In addition, the percentage of B-cells correlated positively with the severity of psychosis.

The link between an inflammatory activation of the immune system and schizophrenia is well supported, but the chain of causal mechanisms linking a genetic predisposition to the manifestations of schizophrenia is still poorly understood. Autoimmune inflammatory diseases are often characterised by antibodies directed against “self” components, so-called autoantibodies. In autoimmune encephalitides, such antibodies have a pathogenetic relevance, but this mechanism has not been supported in schizophrenia. In other CNS diseases such as MS, however, autoantibodies are not the central pathogenetic mechanism. In MS, T-cell mechanisms and antibody-independent B-cell functions are now considered important pathological processes [33, 34]. The last decade’s advancements in the treatment of this inflammatory CNS disease build on these tenets.

The main treatment of schizophrenia, antipsychotic drugs, has been shown to exert some anti-inflammatory effects. This is especially pronounced for clozapine [35, 36], the most effective antipsychotic drug hitherto [37], possibly suggesting that clozapine’s immunomodulatory effects are related to its superior efficacy. However, several trials with anti-inflammatory agents, based on the inflammatory hypothesis of schizophrenia, have been published [38‐42]. The outcome with these anti-inflammatory augmentations (celecoxib, minocycline, simvastatin, aspirin and prednisolone) has in general been unconvincing, but according to a meta-analysis [18], anti-inflammatory add-on treatment to antipsychotics did show some, albeit clinically modest, improvement in psychotic disorders.

More specific immunomodulatory treatments have recently been tested in SSD. In the first study with tocilizumab, this IL-6-antagonist was not effective [43], possibly due to low IL-6 levels in the study sample. However, in a small study with the TNF-α-antagonist adalimumab as an add-on to risperidone, a significant effect on negative and general symptoms was shown [44]. Several other trials with monoclonal antibodies for SSD are presently ongoing [45, 46], however, apart from our open pilot study [47], rituximab has not been tested in SSD.

Rituximab is a monoclonal antibody that binds to a membrane protein, CD20, which is located on the surface of pre-B cells and mature B lymphocytes. The binding of the antibody to a B cell leads to cytolysis (cell death). Rituximab is a standard treatment for primary B cell type lymphoma, including CNS lymphoma and is included in the treatment guidelines for anti-NMDA receptor encephalitis [48]. It is also a standard therapy for RA and MS. The effect of rituximab on neuroinflammation has clearly been shown by its effect in the treatment of MS [49]. The B cell fills multiple functions in the immune system, including as antigen presenter (activating T cells and other cells), secretion of immunomodulating cytokines, and production of antibodies. By using rituximab as treatment for neurological disorders, the antigen-presenting and immunomodulating functions of the B cells are thought to be largely the property that gives rise to the treatment effect, i.e. not the elimination of the antibodies [34]. The action of rituximab in CNS disorders is thought to be primarily associated with depletion of B cells in the periphery which subsequently leads to decreased B cell entry into the CNS from the bloodstream, i.e. not depending on cytolytic action against B cells within the CNS. CD20 is not expressed on hematopoietic stem cells, pro-B cells (precursor to pre-B cells and B lymphocytes), plasma cells (challenged B lymphocytes producing antibodies) or in normal tissue, which means: 1) the effect of rituximab is reversible and upon completion of treatment, the patient eventually recovers its B cells, which thus regain their function in the immune system, 2) already engineered plasma cells will continue to produce antibodies, and 3) tissues in the body that do not express CD20 are not affected. As rituximab has been used on a long-term basis by thousands of people since the nineties, its side effects in the short and long term are well-known and are usually mild. A dreaded side effect, causing progressive multifocal leukoencephalopathy (PML), has been shown to be very rare, so it does not need to be taken into consideration.

It is hypothesized that the role of the immune system in SSD is mediated by the interaction between B and T cells, glial cells and cytokines. A possible effect of rituximab treatment may be that decreased activation via B cells contributes to the normalization of T-cell-driven inflammation in the brain, possibly through enhancement of T regulatory cells (Tregs) which in turn leads to a normalization of glial activity and eventually a decrease in symptoms.

Between 2019 and 2022, we have run two open pilot trials with a single intravenous dose of rituximab 1000 mg. One study included 9 patients diagnosed with treatment-resistant SSD and the other 10 treatment-resistant patients with obsessive–compulsive disorder (OCD). Results show that seven out of nine SSD patients improved significantly at one or several time points (12, 20 or 40 weeks), but only two of the OCD patients were much improved. Rather few side-effects were reported by the participants. Symptoms such as anxiety and depression were rarely reported [47].

In the present study, we want to test if add-on treatment with rituximab can improve symptoms in patients with SSD in a placebo-controlled setting.

The primary outcome measure is change in symptoms measured as change in Positive and Negative Syndrome Scale (PANSS) score from baseline up to week 12.

All clinicians involved in the study will be trained in using PANSS and PSP to reach acceptable agreements between raters.

Levels of biomarkers related to inflammation and glial activation at the start of the study will be compared between the two groups “responders” and “non-responders” in order to identify predictors of treatment response. Profiles of various biomarkers (e.g. cytokine patterns and metabolomes), in addition to absolute values of individual substances, will be assessed to identify biochemical characteristics in patients who benefit from the treatment. Such combinations of biomarkers that can be most informative to distinguish between “responders” and “non-responders” will be analysed with, for example, PCA analysis. Changes in inflammatory markers and other measurements from baseline to week 12 and 24 (Δ values) for all relevant measurements will also be analysed in relation to clinical response after completion of the study.

Patients will undergo MRI examinations at baseline and endpoints; these procedures are optional. Prior to scanning, patients will be asked to complete a MRI safety checklist. Patients will also be asked if they have taken benzodiazepines and/or hypnotics within 24 h of initiating MRI; those who have will be asked to come back for a later session (more than 24 h later). During scanning, patients will undergo the following protocol:

The total examination time is estimated to be approximately 60 min including surveys and preparations.

Anatomical images will be analysed using voxel-based morphometry (VBM) to see if there are any changes in brain volumes after treatment. DTI and rsfMRI will be used to study potential changes in structural and functional connectivity, respectively. The two rsfMRI acquisitions with different stimuli will be compared and analysed. Recent research suggests that the use of naturalistic film during rsfMRI might be beneficial when measuring changes in neuropsychiatric disorders [59].

Participants and informants will be asked to participate in separate qualitative interviews on their subjective experiences of the trial and changes in the participant’s symptoms, 3–5 months post-infusion. Approximately 40 participants and 40 informants will be interviewed. Interviews will be held via telephone or face-to-face. Each interview will last approximately 45–60 min.

The interview guides are semi-structured and include 15 or 19 questions respectively (Additional file 2). The interviews include questions on experience after treatment (e.g. if there are changes in psychotic symptoms, behaviour, motivation and emotions). In addition, four open questions on how informants have experienced the research project will be administered.

Informed consent for this auxiliary study is obtained at baseline. The qualitative data will be analysed through thematic analysis [60]. Participants, informants, and researchers are blinded to treatment allocation.

Different studies report vast variations in placebo response in randomized placebo-controlled trials on schizophrenia, which makes sample size calculation difficult to estimate. The placebo response has increased considerably in more recent studies [61, 62].

An important factor that may affect the placebo response is the selection of patients. If inclusion criteria demand patients to be in relapse or in an acute exacerbation, the natural course of the disease predisposes to regression to the mean, which comes out as a placebo response [61]. We will mostly include chronic stable patients with schizophrenia, which should imply a lower risk for placebo response.

Drop-out rates are also expected to be low, as we use few assessments, only administer one single dose of the research drug and offer a reimbursement (700:- SEK) at both endpoints.

A mean baseline PANSS score of approximately 85 (SD = 17) is anticipated, based on the proposed selection of participants with mostly stable chronic course of their disease.

We expect 33% in the rituximab-treated group to be much or very much improved (i.e., score 1 or 2 on CGI-I), compared to an expected 10% in the placebo group. To reach a power of 80% with a significance level of 5% (two-sided), we need to include 47 participants in each group. To allow for a 10% drop-out, we need 52 (47/0.9) participants per group.

Response criterion is defined as much or very much improved according to CGI-I.

We aim to include a total of 120 participants to account for a loss of degrees of freedom when statistically adjusting for prognostic factors in the model.

Rituximab is associated with an increased risk of infections. In our pilot study, all rituximab patients displayed a clear reduction of circulating B-cells following treatment, from a mean percentage of 3.92 (± 1.31) at baseline to 0.016 (± 0.01) at week 12. However, at 40 weeks after treatment, all patients except two showed signs of repopulation of circulating B-cells.

Venepuncture may be associated with mild discomfort but will be performed by a nurse, trained in venepuncture.

Lumbar puncture, which is optional in this study is associated with discomfort in most patients. Lumbar punctures are performed by a physician trained in the procedure. To prevent post-dural puncture headaches, atraumatic, small-bore needles will be used.

Participants will be provided with a panic button. They can communicate with the researcher and scan operator throughout the MRI scan. Participants are screened prior to examination to ensure that no metal is present on or within the body.

No risks are anticipated for the staff, but staff will adhere to normal safety procedures.

Participants will be selected based on the inclusion and exclusion criteria. Additionally, participants will be tested for Tuberculosis (TB), HIV, and Hepatitis B and C prior to the infusion. Participants are offered vaccination against varicella zoster and COVID, and if accepted vaccinations are given at least 4 weeks prior to infusion. Female participants of childbearing age or who do not practice sexual abstinence will be given a pregnancy test, which must be negative. In addition, they are informed to avoid getting pregnant for 1 year post-infusion and must adhere to using effective contraception during this period, if sexually active.

If a participant show sign of any infection on the day of the planned infusion, the treatment will be postponed.

Infusions will be given at a rheumatology or neurology ward under the supervision of a study nurse. The infusion bag containing rituximab or saline will be covered with a black plastic bag to keep the patient and study nurse blinded. Temperature, cardiac rate and blood pressure will be measured during the infusion, in line with the use of rituximab for treating patients with somatic disorders.

After infusion, participants will be advised to seek help if they feel unwell. They are provided with a card with a telephone number to the responsible clinician. If necessary, we will unblind the participant and inform their health professional whether they received rituximab or normal saline. Post-trial care, if needed, is available within the regular health care system.