Obsessive-Compulsive Disorder Associated with Autoimmunity in Youth: Clinical Course before and after Rituximab +/− Adjunctive Immunomodulation

The probability that some cases of obsessive-compulsive disorder (OCD) have an underlying autoimmune pathophysiology is supported by the association between OCD and autoimmune disorders (in patients and first-degree family members) [1, 2], as well as by multiple studies and meta-analyses showing that OCD patients have 5-fold higher rates of anti-basal ganglia antibodies than in controls [3] and exhibit neuroinflammation markers in the corticostriatal-thalamo-cortical brain regions – implicated in OCD pathophysiology [4]. Indeed, autoimmune and other inflammatory mechanisms are central to some models of OCD, including pediatric acute-onset neuropsychiatric syndrome (PANS) and pediatric autoimmune neuropsychiatric disorders associated with streptococcal infection (PANDAS).

PANS/PANDAS is thought to be caused by basal ganglia autoimmunity and is supported by the findings of: (1) autoantibodies to a subset of cholinergic interneurons in the basal ganglia that alter excitability of these neurons [5‒7], and autoantibodies that bind proteins which may reflect dopamine receptors [8, 9]; (2) imaging findings pointing toward basal ganglia pathology [10‒13]; (3) neurological soft signs pertaining to the basal ganglia (e.g., positive glabellar tap, choreiform movements) [14]; and (4) sleep studies indicating movements or atonia during rapid eye movement sleep [15‒17], Moreover, our group has noted that patients with PANS/PANDAS have a high rate of developing arthritis (28% by age 14) and other autoimmune diseases [18, 19]. Additionally, these patients commonly have findings seen in rheumatological/systemic autoimmune disease: anemia, low complement levels, circulating immune complexes, low-grade vasculopathy markers including periungual erythema and swelling (which is also seen in scleroderma and dermatomyositis), proinflammatory monocytes, etc. [18‒22].

PANS/PANDAS is characterized by the abrupt and dramatic onset of OCD and at least two other neuropsychiatric symptoms which are equally severe and abrupt, including eating restriction, tics, pronounced irritability/rage, enuresis/polyuria, sleep disruption, and behavioral and cognitive regression [23‒25]. PANS/PANDAS frequently follows a relapsing-remitting course; however, a significant minority have a primary chronic or secondary chronic course [26], in press [27‒29]. Many patients suffer not only from immense emotional distress and lability but also significant difficulty with rudimentary tasks of daily living such as eating, grooming, bowel and bladder management, handwriting, school attendance, and basic community and family participation. As a result, caregiver burden exceeds that seen for patients with dementia [30‒33].

For PANS/PANDAS, the PANS Research Consortium has established a standard approach for evaluation and treatment (immunomodulatory therapies, management of infections, and psychiatric interventions) [24, 34‒37]. PANS may be related to resistant OCD symptoms and similar treatment approaches may be helpful [38, 39]. As is the case with many uncommonly observed diseases, evidence supporting immunomodulatory therapy in presumed-neuroimmune cases of OCD derives largely from the experience of treating specialists, bolstered by systematic case collections and a few small clinical trials that have varied in scope and rigor [40, 41]. Practitioners treating patients with severe chronic PANS, PANDAS, and OCD associated with clear signs of autoimmunity have, therefore, largely based clinical decisions on the parallels between these and other autoimmune/inflammatory conditions (e.g., systemic lupus erythematosus, autoimmune encephalitis, etc.) and on the treatments required by individual patients for relevant comorbidities.

Rituximab, a monoclonal antibody (mAb) that targets the CD20 antigen, depletes B cells with long-acting effects on immune cell function, and is used as a therapeutic biologic agent in multiple sclerosis, rheumatoid arthritis, systemic lupus erythematosus, immune mediated thrombocytopenia, and other autoimmune disorders. Of interest for OCD and other neuropsychiatric disorders with potential autoimmune origins, treatment with rituximab in clinical trials for lupus nephritis has been associated with rapid and statistically significant improvement in the Mental Component Summary (MCS) score of the SF-36 – even in a study where no significant improvement was reported in the Physical Component Summary (PCS) score [42]. Additionally, significant and sustained improvements in the MCS and PCS of the SF-36 are also reported with rituximab therapy of rheumatoid arthritis [43]. Rituximab’s positive effect on the neuropsychiatric aspects of autoimmune diseases, independent of any secondary effects due to improved physical condition, is also suggested by its effect on psychiatric symptoms in the treatment of anti-NMDA receptor encephalitis [44]. A small exploratory study of rituximab in treatment-resistant OCD and schizophrenia spectrum disorder (SSD) suggests efficacy in some patients, although primarily in those with SSD [45]. There are also a few case studies reporting successful treatment in patients with PANDAS [46, 47], and refractory OCD [48].

The Stanford Immune Behavioral Health Clinic has used rituximab to treat 23 severe complex patients with OCD, many who met the criteria for PANS/PANDAS at presentation and others who exhibited severe OCD (often with psychosis) in the context of systemic autoimmune conditions at presentation. Here we report on the characteristics of these patients, the reasons for selection of rituximab treatment, the specifics of the treatment, and the clinical course following administration.

The Stanford University (SU) Immune Behavioral Health (IBH) Clinic is a multi-disciplinary clinic providing specialty care in psychiatry, psychology, social work, pediatrics, rheumatology, and immunology. Since 2012, it has been evaluating/treating patients referred for suspected PANS or related conditions by regional primary care providers. Before being admitted to the IBH Clinic, the abrupt onset of OCD and/or severe eating restrictions is assessed via a screening process. The IBH Clinic admits an estimated 10% of referred patients, prioritizing those with a history suggestive of abrupt onset of symptoms, but after detailed intake, not all patients meet full criteria for PANS.

This retrospective study was approved by the Stanford Panel on Human Subjects Institutional Review Board (IRB) (approval numbers 28533 and 28922). For the patients detailed in this case review (n = 23, online suppl. Table S1; for all online suppl. material, see https://doi.org/10.1159/000544993), written informed consent was obtained from participants (or their parent/legal guardian/next of kin) as per our IRB-approved protocol 26922.

A total of 458 pre-screened cases were evaluated by our IBH clinic between September 15, 2012, and January 6, 2023, for neuropsychiatric deterioration, and of these, 225/458 (49%) met PANS criteria. Psychiatric symptoms and diagnoses were classified by child psychiatrists (M.T., K.C., Y.X., and M.S.). Rheumatological and immunological assessments were done by board-certified pediatric rheumatologists and immunologist (J.F., M.M., and J.D.H.). Per our clinic’s protocol, all are assessed for arthritis (60 joints) and enthesitis (41 entheses), including all spinous processes, sacroiliac joints, finger metacarpophalangeal (MCP) joints, toe metatarsophalangeal (MTP) joints, finger and toe proximal interphalangeal joint (PIP), and distal interphalangeal joint (DIP) joints, wrists, elbows, shoulder, hips, knees, and ankles. Our joint exam is based on the 66/68 joint count [49]. Our enthesitis exam is based on the Mander enthesitis count [50] since this is the only enthesitis exam that includes the spinous processes which are commonly involved in our patients with early-onset OCD and PANS [12, 13]. Physical exam evidence of arthritis is confirmed by imaging of all patients who are able to cooperate (which was all the case eventually). Musculoskeletal (MSK) ultrasounds were performed using a GE LOGIQ E9 ultrasound machine utilizing a 6–15 MHz linear transducer or a Siemens ACUSON Sequoia ultrasound machine utilizing an 18 MHz linear transducer. Settings were optimized for superficial MSK examinations. Both grayscale and color Doppler images of the target joints were acquired and saved on our institution’s Picture Archiving and Communication System. All imaging was read by a board-certified pediatric radiologist with fellowship training in musculoskeletal imaging (J.S.).

Pediatric rheumatologists (M.M. or J.F./B.F.) classified patients with an autoimmune disease per standard definitions which have previously been described [19]. Patients were classified as having arthritis if they had (1) pain and joint effusion (lasting >6 weeks); (2) pain and 2 or more of the following: limited or painful range of motion, tenderness, or warmth (lasting >6 weeks); or (3) joint pain or tenderness with ultrasonographic confirmation of arthritis. The first 2 categories meet the American College of Rheumatology criteria for juvenile rheumatoid arthritis [51]. Patients were further classified by the International League of Associations for Rheumatology (ILAR) [52] and/or Assessment of SpondyloArthritis International Society (ASAS) criteria [53, 54]. Although many of our patients had no or minimal joint pain at the height of their psychiatric symptoms, eventually they endorsed joint pain and/or stiffness.

Other laboratory and physical examinations to evaluate for autoimmunity were based on the level of suspicion of a condition and the patient’s ability to cooperate with exam and lab draws. Autoantibodies, physical manifestations of inflammation (autoimmune cytopenias, non-specific signs of inflammation like CRP, ESR, anemia, thrombocytosis), signs of vasculopathy or coagulation issue, autoimmune/inflammatory rashes, signs of inflammation in cerebral spinal fluid (CSF), autoimmune/immunologic comorbidities, and first-degree family members with autoimmunity/immune deficiency are outlined in Table 3 (including parameters for abnormal labs). In 3 patients, we were unable to do a full physical examination and/or get sufficient laboratory workup due to irritability/rage and/or extreme fears.

Starting in 2017, prior to each visit, parents/patients fill out an electronic questionnaire that included the following scales: Children’s Yale-Brown Obsessive-Compulsive Scale (CY-BOCS) scores ranging from 0 to 40 [55, 56], Modified Overt Aggression Scale (MOAS) scores ranging from 0 to 100 [57], Columbia Impairment Scale (CIS) scores ranging from 0 to 52 [58], Global Impairment and % baseline scores ranging from 0 to 100% [59], Malaise Score scores ranging from 0 to 12 [60], Widespread Pain Index (WPI) scores ranging from 0 to 19 [60], Caregiver Burden Inventory [61], PANS 31-item Symptom Rating Scale which includes a rating for a variety of symptoms such as OCD, hallucinations, delusions, anxiety, irritability, pain, etc. [62]. Prior to 2017, paper forms were filled out (which included an early version of the 31-item, Global Impairment, and Caregiver Burden Inventory). Trajectories of these parent scores together with the clinician’s impression (Children’s Global Assessment Scale) informed response/recovery classification.

Data regarding neuropsychiatric symptoms, family history of autoimmune conditions, immune labs, and physical exams were captured using a combination of structured questionnaires, chart review, clinical interviews, and examinations. Some aspects of the evaluation were missed since the parents and clinicians were often overwhelmed by the severity of the child’s illness and/or not able to gain their cooperation due to severe and interfering psychiatric symptoms (e.g., rage, oppositionality, violence). Additionally, some patients were profoundly suicidal, homicidal, or had other urgent life-threatening symptoms (thus requiring urgent admission to outside hospitals), and gaining cooperation from the outside providers to order complex immune labs was difficult. Lastly, some families were hesitant to reveal symptoms of suicidality, homicidality, and violence due to fears that the child would be removed from the home and would not get adequate medical treatment. Similarly, laboratory evaluations were performed as clinically indicated and feasible; therefore, not all laboratory evaluations were available for all patients. Lastly, many patients had difficulty cooperating with blood draws and the physical examination, so immunological data in these cases are scant. A clinical course prior to rituximab was classified based on definitions in Masterson et al. [26], in press.

See Table 1 for our protocol for selecting cases in which to use rituximab. Two patients in this series did not meet these criteria. Case 4 had life-threatening eating restriction and failed multiple intensive inpatient and outpatient treatment programs including psychotropics from many classes and qualified for plasma exchange (PEX) and IV corticosteroids based on treatment recommendations for life-threatening disease [36]. This patient did not allow us to get close enough to examine him. Autoimmune labs were planned for the admission but were never obtained. Rituximab was pursued based on our experience that PEX and IV corticosteroids in PANS typically have a temporary impact, and we were aiming for a sustained response. Case 21 was treated before we solidified our Table 1 criteria and in retrospect should not have been a candidate for treatment given the lack of autoimmune signs/markers. Although he had enthesitis-related arthritis (ERA) and anemia secondary to inflammation, these conditions may not have reflected adaptive immune system involvement, especially considering data suggesting that infections and innate immune system activation may drive ERA [63].

The summaries presented (Tables 4, 5; online suppl. Table S1) reflect all OCD and PANS/PANDAS patients seen at the Stanford IBH Clinic who were treated with rituximab by the IBH team (outpatient or inpatient) and had regular IBH outpatient clinic follow-up sufficient to assess response. Based on these criteria, 23 received rituximab and were included in this retrospective review. Patients who received rituximab but were excluded (n = 7) were patients who received rituximab and/or regular follow-up from another provider/team (n = 6) or the patient was primarily psychotic without significant OCD (n = 1) and was followed by psychosis clinic. Of the excluded cases, we were informed that most (n = 4) had improvements, but there were insufficient data to classify as recovered (per definitions above). One patient had no improvement in extreme intrusive thoughts but had improved cognitive abilities and was able to start reading and going to school. Two excluded patients had notes in Epic from other providers suggesting that they did not respond (per definitions above). One excluded patient had mixed interpretations of the response.

Prior to rituximab treatment, all patients in the series were either chronically impacted by OCD and other neuropsychiatric symptoms or had suffered a recent debilitating and life-threatening relapse and/or had organ-threatening autoimmunity (n = 1, glomerulonephritis). All had failed or reached a plateau with other treatment trials (e.g., exhaustive trials of psychotropic medications, inpatient and/or intensive outpatient therapy (or were too severe to participate). All had prior attempts with immunomodulation (e.g., intravenous immune globulin [IVIG], methylprednisolone, and other immunomodulatory treatments), but the response was incomplete or not sustained and risks of rituximab were less than the problems of current therapies (e.g., lack of response, corticosteroid side effects, inability of IVIG impact to last until the next IVIG). All were so impaired that they did not attend nor engage in school activities, engage in extracurricular activities, or leave the home for extended periods.

Our rituximab treatment protocol was two infusions of 750 mg/m² per dose (max dose 1,000 mg), separated by 2 weeks (cycle 1), then we repeated this course 6 months later if patient had clear improvements in neuropsychiatric health as determined by the managing clinical team using the psychometrics discussed above (2 cycles per year = 4 infusions per year). However, once we realized that patients had clear improvements by month 3–4 but then lost these improvements shortly thereafter, we suspected B-cell repopulation in tissues (even when peripheral CD19/CD20 counts were still low) as tissue repopulation of B cells can occur prior to evidence of repopulation in blood (observed in animal model data, per our discussions with Genentech). Thus, we adjusted our protocol and began giving rituximab quarterly (when insurance approved) as this was also the protocol at our institution for refractory nephrotic syndrome (minimal change disease) due to similar findings of break-through symptoms (proteinuria) (after a period of improvement) at the 3–4 month window after rituximab. This adjusted protocol still gave the initial two infusions (full cycle) but then only one infusion quarterly, so that the total doses per year was still four (to stay within the parameters of insurance authorizations). Since most of these cases were treated prior to rituximab being available as a biosimilar, we had tremendous difficulty getting insurance authorization in a timely manner which impacted the protocols used in some individuals.

Adjunctive immunomodulation was given to augment the impact of rituximab and/or control for infections (in the case of IVIG), and/or treat concurrent arthritis since rituximab is not indicated for the forms of arthritis seen in PANS and OCD (ERA, psoriatic arthritis [PsA], spondyloarthritis [SpA]). Additionally, based on our experience with other conditions (e.g., neuropsychiatric lupus and autoimmune encephalitis) and our early cases, we suspected that rituximab alone would not be insufficient for severe cases. For the statistical analysis assessing the impact of adjunctive immunomodulation, we only included adjunctive therapies if they were started before or during the start of rituximab treatment with the intention of ongoing use.

When IVIG was used as an adjunct to treat the presumed underlying autoimmune condition, we used 2 g/kg (maximum dose 70 g) given monthly but separated from the rituximab infusions by 1–2 weeks when possible. As the patient improved, the frequency of the IVIG was reduced and eventually given quarterly (to maintain adequate IgG levels) or shifted to subcutaneous immunoglobulin. When given quarterly, IVIG was staggered at the midpoint between rituximab infusions. When disease-modifying antirheumatic drugs (DMARDs) were used, we used the typical standard-of-care dosing for juvenile arthritis (methotrexate [MTX], leflunomide) and lupus (mycophenolate mofetil [MMF]).

All patients in this series were also treated per standard of care for infections. Each patient had a unique journey with respect to infection and infection treatment history prior to rituximab. We used standard of care for treatment of Group A streptococcal (GAS) and Mycoplasma pneumoniae infections, which were the two main bacterial infections in our clinic (online suppl. Table S1). For cases, where GAS was the recurrent trigger, we used (1) prophylactic antibiotics for GAS, with either penicillin 250 mg twice daily, amoxicillin 500 mg daily, or cephalexin 500 mg twice daily; (2) escalation of antibiotics to treatment doses when close/household contact had GAS; (3) frequent screening of patient/close contacts. Of the patients who had infections post-rituximab, all had these same types of infections prior to rituximab treatment (ENT and skin infections) (n = 3).

Patients’ responses to treatment were categorized by the clinical team using two different response classifications: one to designate the magnitude of eventual response (full recovery [FR], partial recovery [PR], no recovery), and another to designate the clinician-assessed likelihood that the response could be attributed to rituximab specifically (attribution of response).

Definitions of recovery [26], in press, are as follows:

• Partial recovery: unequivocally recovered from a flare/deterioration, but with residual functional deficits that are above their baseline impairment level.

• Full recovery: unequivocally recovered; symptoms and functional impairment are equivalent or near equivalent to the patient’s baseline impairment level.

Recovery status was informed by patient questionnaires/psychometrics (see methods, e.g., CY-BOCS, Columbia Impairment score, etc.), documented clinical interviews and CGAS (after 2017).

Categories used to classify attribution of response were defined as follows:

• Clear responders: consistent clinically meaningful improvement on treatment and, if applicable, deterioration on discontinuation, with minimal potential confounding factors; sufficient confidence for clinician to continue treatment (whether or not that actually happened, based on other factors)

• Likely responders: clinically meaningful improvement in treatment, but in the context of potential confounding factors. This includes patients who received concurrent induction therapy with IVIG and/or IV steroids, but who sustained response for 6+ months post-induction; sufficient confidence for the clinician to continue treatment

• Possible responders: mixed response to treatments, i.e., either mixed across treatment course (e.g., good response but then relapse) and/or mixed across symptoms, and/or significant confounding factors; insufficient confidence in response for clinician to continue treatment

• Limited/inadequate responders: limited response did not justify continuation of treatment

Exploratory analyses using one-sided Fisher’s exact tests were performed to examine the relationships between adjunctive immunomodulation and response to rituximab, with “PR” and “FR” collapsed into one “responder” category. All analyses performed are presented and Bonferroni corrections were applied in assessing statistical significance, as indicated in the table footnotes. One patient (case 15) was excluded since they only received one dose (not a full cycle) of rituximab, and 1 patient (case 4) was excluded from breakout analyses because of immune deficiency since he refused laboratory evaluation until after IVIG maintenance therapy was underway; thus, his immune deficiency status is unknown. Statistical analyses were performed using the JMP^® statistical analysis program.

Table 2 summarizes the demographics, infections preceding neuropsychiatric deterioration(s), neuropsychiatric comorbidities, neuropsychiatric symptoms for the flares (relapses) that prompted patients to be seen in our IBH clinic, and prior (pre-rituximab) immunomodulatory treatments. Males comprised 65% of the cohort. The mean age at the onset of first psychiatric deterioration was 9.5 years. The most common preceding infections (prior to deterioration that led to rituximab) were GAS infections. Nearly all (95.6%) suffered chronic psychiatric symptoms prior to rituximab treatment. The most common psychiatric symptoms were obsessive-compulsive symptoms (100%), psychotic symptoms (47.8%), eating restriction (56.5%), anxiety (95.7%), mood dysregulation (95.7%), behavioral/social symptoms (100%), aggression (65.2%), cognitive symptoms (78.3%), motor symptoms (56.5%), urinary symptoms (82.6%). All either refused or were unable to attend school. Immunomodulatory treatments that had already been trialed for most patients included IVIG (78.3%) and/or IV methylprednisolone (69.6%).

Table 3 presents the patients’ immune profiles, including autoimmune labs, physical manifestations of autoimmunity (anemia, arthritis, vasculopathy, etc.), and autoimmune/immunological conditions/diagnoses, as well as autoimmune/immunological disorders in first-degree family members. Nearly half (43.5%) exhibited low C3 and/or C4. Most (95.7%) patients had at least one physical manifestation of inflammation, including cytopenias (62.5%), high CRP and/or ESR (30.4%), low-grade arthritis (all with objective signs on imaging) (82.6%), vasculopathy markers (52.2%), rashes (21.7%). Autoimmune/immunologic comorbid diagnoses include juvenile arthritis (ERA, PsA, oligoarticular arthritis) (82.6%), Hashimoto’s thyroiditis (43.5%), immunodeficiency (30.4%), and a variety of other diagnoses present in <18% of patients. Common autoimmune/immunological conditions in first-degree family members include autoimmune arthritis (43.5%), thyroiditis (39.1%), psoriasis (21.7%), and immunodeficiency (21.7%).

Table 4 summarizes the patients’ responses to rituximab treatment, classified by recovery status as described in the Methods, along with the adjunct therapies applied, adverse events experienced, and where applicable, reasons for discontinuation of rituximab treatment. As described more fully in the detailed case descriptions (see online suppl. Table S1), most patients had signs of improvement within 3–4 months of initiating rituximab. After 1–5 years, 16 out of the 23 (70%) met our definition of recovery (see methods). For half the cases, the attribution of response to rituximab was “clear” or “likely.” For the other half, there were confounding treatments that contributed to recovery (e.g., concurrent psychotherapy, psychotropics, and/or adjunct immunomodulation) such that we could not confidently attribute improvement to rituximab (see online suppl. Table S1 for details). There were no serious adverse events. Eleven (47.8%) patients experienced a transient increase in psychiatric symptoms and/or a transient increase in joint pains particularly with the initial infusion. Fourteen (60.8%) experienced hypogammaglobulinemia. Only 5 of the 23 patients remain on rituximab therapy as of this writing; reasons for discontinuation among the other 18 were evenly divided between the positive (resolution of symptoms, ability to “step down” in therapy) and the negative (insufficient risk/benefit ratio). Three patients experienced recurrent infections before and during rituximab, which was at least one of the reasons for discontinuing rituximab.

Table 5 presents analyses relevant to the potential impact of adjunctive immunomodulatory therapy. Patients who received IVIG and/or DMARDs had a higher likelihood of achieving recovery compared with those who did not receive these adjunct therapies (Fisher’s exact Test, one-sided, p < 0.0001). Of those who received any one of these adjunct therapies (at initiation of rituximab), 100% (14/14) achieved at least PR, including 21% with FRs. Among those who did not receive at least one of these adjunctive immunomodulators, in contrast, only 1 patient out of 8 (13%) achieved recovery; notably, this patient received PEX induction plus 6 weekly IV methylprednisolone pulses, a significantly more aggressive induction than all the other patients in our series.

Although sample sizes are small, once patients were segmented by immune deficiency history, it is notable that no patient with suspected immune deficiency achieved recovery without concurrent monthly IVIG therapy (compared with those that received IVIG, p = 0.07) (Table 5). Additionally, no patient with the diagnosis of immune deficiency achieved an FR. Immunocompetent patients receiving any adjunctive immunomodulation (IVIG, DMARDs) appeared to achieve the highest rate of FR (compared to immunocompetent patients not receiving adjunctive therapy, p = 0.007) (Table 5).

During the early years of the Stanford IBH Clinic (prior to our current screening process of seeing new-onset cases), our patients often presented in the chronic phase of illness (primary or secondary chronic) and often had signs of autoimmunity/inflammation. Starting in 2015, due to a small clinic staff, we began to prioritize new-onset cases which typically respond to less aggressive treatment approaches including: clearing infection, applying standard of care for psychiatric symptoms, and detecting and addressing inflammation/autoimmunity at an early stage of presentation (i.e., by systematically screening for arthritis using ultrasound and assessing for common autoimmune conditions like thyroiditis, celiac disease, etc.). For many such patients, treatment with traditional, non-biologic DMARDs has been sufficient. However, a small group has had a refractory disease course; thus, we have been employing targeted biologic DMARDs (e.g., cytokine inhibitors) per standard of care for the patient’s comorbid autoimmune conditions; we are in the process of assembling data on this subset.

When these targeted and less aggressive approaches have failed and symptoms remain severe, broader-range and/or more intensive immunomodulatory therapies such as IVIG, IV corticosteroids, and rituximab have been tried and, as reported here, may be beneficial to some. Out of the 458 patients we followed in the IBH clinic, only 23 patients had a clinical immune profile and disease severity warranting a trial of rituximab.

The data presented here support the use of rituximab for a small subset of presumed-neuroimmune OCD, while strongly underscoring the importance of adjunct therapy. In our patients, use of rituximab in conjunction with adjunct IVIG, DMARDs, and/or MMF (applied at initiation of rituximab) was associated with a 100% rate of response (unequivocal recovery of function, either FR or PR). On the other hand, recovery was achieved by only 1 patient (out of 6) who did not receive such adjunct therapy, and this patient had received particularly aggressive induction (PEX + IV methylprednisolone ×6) compared to our typical protocol. The highest rates of FR were seen with adjunct DMARDs among patients without immune deficiencies and 50% (3 out of 6) returned to baseline functioning (FR) after rituximab despite severe psychiatric symptoms, disability, and numerous prior treatment failures.

One might contend that it is not the rituximab that is making a significant impact – rather it is the IVIG and DMARDs that are resolving inflammation/neuroimmune problems and/or psychiatric treatments. This is countered by the fact that in 4 patients (cases 2, 7, 8, 9), attempts to space or wean rituximab within 3 years of initiating failed. Additionally, most patients had improvements after rituximab, but then deteriorated in the anticipated timeframe that B-cell repopulation may occur in tissues. Confirmation was evident when therapists were making headway with regard to CBT, ERP and in one case (case 8), dysarthria and loss of progress/escalation of symptoms were documented in the therapy notes (in three cases, therapists were blinded to B-cell repopulation that occurred in the blood) (see case descriptions in online suppl. Table S1). With regard to the adjunctive immunomodulation therapies, we do not know whether these additional immunomodulators prevented immunogenicity against rituximab or whether they were synergistic treatments in that they addressed other activated immune pathways. For example, proinflammatory monocytes have been implicated in OCD and PANS [21, 22], and we suspect that MTX, leflunomide, and MMF are effective in inhibiting the functions of proinflammatory monocytes and shifting their profile to repairing/regulatory profiles [64‒67]. Based on reactions in 3 patients receiving additional cycles of rituximab we suspect that at least three cases (cases 14, 19, 20) may have had anti-rituximab antibodies, and possibly more since most had very early B-cell repopulation and/or loss of gains.

Rituximab is known to be immunogenic as it is a chimeric (murine-human) mAb, fusing variable regions of a mouse antibody with the constant regions of a human. Background therapy with MTX and other anti-proliferative agents is associated with a lower incidence of anti-drug antibodies (ADAs) to rituximab and a longer efficacy of treatment [68]. These anti-proliferative agents include MTX, leflunomide, MMF, and azathioprine. They must be given with the first dose of the mAb to be effective and should be continued throughout the therapy to prevent ADAs [68]. Other immunosuppressive agents and glucocorticoids do not have this effect. The immunogenicity of rituximab often goes unrecognized as it is administered intermittently and loss of benefit due to the development of ADAs may not be as apparent until infusion reactions are evident.

For better or for worse, rituximab can have a profound long-lasting impact on the immune system. In patients treated with rituximab, activation of the immune system to deplete B cells, together with other immunosuppressive properties (including T cell effects) and the immunogenicity of rituximab itself, can result in adverse effects including infusion reactions, reduced immunoglobulin levels, and increased risk of certain infections. In our sample, the rate of hypogammaglobulinemia (61%) was higher than the rate reported for children in the literature (11%–50%) [69]. This likely reflects three conditions: (1) the high rate of immunodeficiency seen in our population [25], (2) the aggressive rituximab protocol and prolonged rituximab use in many cases and (3) adjunctive immunomodulation. Given this potential safety risk, it is important that immunocompetence be assessed prior to initiating treatment, and it may be prudent to consider rituximab only for patients with normal IgG levels or who are receiving (or able to receive) monthly IVIG or subcutaneous Ig. In addition to this safety consideration, as discussed previously, patients with confirmed or suspected immune deficiencies only achieved a response to rituximab when receiving concurrent IVIG therapy and even in this context, none achieved more than a partial response.

One could argue that the risk of rituximab is not worth a PR (half of our recovered cases). However, all the patients in PR were able to go back to school, enjoy extracurricular activities, resolve suicidal and homicidal thinking, and they and their families reported improved health-related quality of life that outweighed the need for IVIG to treat hypogammaglobulinemia.

Additionally, a report suggests that even a single infusion of 1,000 mg rituximab has been shown to significantly delay arthritis development in some at-risk patients (pre-arthritis stage of autoantibody-positive rheumatoid arthritis) [70], suggesting its potential to delay or prevent the additional autoimmune conditions that so commonly develop in patients with PANS [19], and routine OCD [1]. Among the greater population of Stanford IBH patients (beyond this rituximab cohort), the estimated cumulative incidence of arthritis is nearly 30% by age 14, and ∼20% for another autoimmune disease (thyroiditis, psoriasis, celiac, Behcet’s, inflammatory bowel disease, atypical systemic lupus erythematosus, myasthenia gravis, etc.) by age 18 [19]. Even among patients not (yet) meeting the criteria for an arthritis diagnosis, 47% demonstrate inflammatory signs on joint ultrasound, and many also present with indirect signs of vasculopathy including: periungual redness/swelling (13%), livedo reticularis (33%), prominent onychodermal bands (33%) [19]. These findings are supported by published data from other cohorts. In a Swedish clinical cohort, 24% of 45 PANS patients were reported to have comorbid autoimmune or inflammatory diagnoses including arthritis, celiac disease, autoimmune thyroiditis, and Henoch-Schonlein’s purpura, and upon examination, 61% had skin abnormalities (including livedo) [70]. Similarly, in a large (N = 698) US PANS survey cohort, joint conditions were reported in ∼46%, and of those with persistent (>50% of days) PANS symptoms, ∼38% reported muscle conditions/pain, ∼7% reported celiac disease, and ∼42% reported “unexplained” dermatologic conditions; several other autoimmune/inflammatory conditions were also more common than expected [25, 33]. A genetic propensity toward systemic autoimmune conditions in patients with PANS is further suggested by the findings (in multiple cohorts) of very high rates of autoimmune disease among first-degree family members [25, 27, 71‒75].

As this is a retrospective series, this study lacks rigor, especially with regard to systematic data collection, small patient numbers, and lack of placebo control. Although, most of these rituximab-treated patients went from extreme levels of OCD and other psychiatric symptoms which prevented them from typical childhood routines (e.g., school, sports), and after rituximab, they had a return to a normal childhood routine; but without a placebo arm there is a potential for rater bias in how parents and clinicians interpret recovery and assign attribution. Nonetheless, these data may inform future clinical trials, particularly with regard to the need for adjunctive immunomodulation.

While most patients with OCD recover with less aggressive treatments, the cases presented in this series suggest that, for primary or secondary chronic cases with underlying autoimmunity and severe neuropsychiatric symptoms or organ-threatening autoimmunity (e.g., glomerulonephritis), rituximab may offer a treatment option with a positive ratio of benefits to risks when used in combination with properly selected adjunct therapies (IVIG for immunodeficient, DMARDs for immunocompetent). Given the profoundly negative consequences of disease activity in such cases, not only with respect to current functioning but also with respect to the potential for permanent injury caused by inflammation and the tendency for progression to additional autoimmune diagnoses, consideration of rituximab may be clinically justifiable and should be a priority for clinical trials. However, evidence for the use of rituximab in the PANS population should currently be considered provisional until proven to be helpful in clinical trials. If it is being considered, clinicians should use rituximab in patients who pass screening criteria (Table 1) and only when standard of care for psychiatric symptoms and comorbid autoimmunity has failed.

We would like to thank the Neuroimmune Foundation and Sam Pleasure, MD, PhD (University of California, San Francisco), for organizing/facilitating this special edition following the 2024 Inflammatory Brain Disorders Conference. We are grateful to our patients and families who understand treatment limitations and continue to lend their time and cooperation to research participation. We would also like to thank the Stanford IBH Clinic team, current and former members of our research staff, collaborating physicians, and all of our students and volunteers who have assisted with chart reviews and data collection.

This study was approved by the Stanford Panel on Human Subjects Institutional Review Board. Written informed consent was obtained from all adult patients and from the parent/legal guardians of patients less than 18 years. This study protocol was reviewed and approved by the Stanford Panel on Human Subjects Institutional Review Board, Approval Nos. 26922 and 28533.

The authors have no conflicts of interest to declare except for Kiki Chang, who is a consultant for AbbVie Inc. and Human Health.

Funding for the infrastructure of our program and all clinical and basic science research conducted by the Stanford Immune Behavioral Health Clinic and collaborators came from (1) the Lucile Packard Foundation for Children’s Health; (2) the Stanford Maternal and Child Health Research Institute (MCHRI); (3) Stanford SPARK; (4) the Neuroimmune Foundation; (5) the Dollinger PANS Biomarker Discovery Core; (6) PANDAS Physician Network (PPN) and Global Lyme Alliance who provided funding for the healthy controls; (7) Caudwell Children’s Foundation, (8) O’Sullivan Foundation; (9) the Brain Foundation; (10) Oxnard Foundation; (11) the Louisa Adelynn Johnson Fund for Complex Disease; (12) Brain Inflammation Collaborative; (13) Susan Swedo and the National Institute of Mental Health – Pediatrics and Developmental Neuroscience Branch which supported the initial creation of the Stanford PANS Program.

J.F. and D.C. made substantial contributions to the conception and design of the work as well as the acquisition, analysis, and interpretation of data for the work; drafted the work; and reviewed it critically for intellectual content. M.M., O.K., K.M., C.M., J.D.H., J.S., B.F., Y.X., M.S., J.M., V.S., K.C., and M.T. made substantial contributions to the acquisition, analysis, or interpretation of data and reviewed it critically for intellectual content. All authors approve the final version of this manuscript and agree to be accountable for all aspects of the work.

Jennifer Frankovich and Denise Calaprice are co-first authors, Kiki Chang and Margo Thienemann are co-last authors.

The data that support the findings of this study are available from the corresponding authors upon reasonable request. All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding authors.