Neutralizing anti-interleukin-10 autoantibodies in inflammatory bowel disease

The authors proposed a prevalence and phenotype study, and three of the works that cite this paper have now run it in adults and in children, so the useful remaining design is not another cross-sectional prevalence count but a population-based incident cohort that fixes the assay threshold, measures seropositivity before treatment starts, and tests whether the reported phenotype associations hold prospectively.

1. Has anyone done this since?

Status

addressed

Works since

• Interleukin-10 Autoantibodies and HLA-DRB1*01:03 in Inflammatory Bowel Disease (Gharahdaghi, Yeh, Ceron-Gutierrez and colleagues 2026, New England Journal of Medicine, doi:10.1056/nejmoa2513654)
  The same group screened the Oxford and United Kingdom inflammatory bowel disease (IBD) BioResource cohorts with a cellular interleukin-10 (IL-10) reporter assay and a confirmatory competitive enzyme-linked immunosorbent assay, finding neutralising anti-IL-10 in 173 of 4909 patients with IBD (3.5 percent, 95 percent confidence interval 3.0 to 4.1) and none of 1006 controls, with a strong association to HLA-DRB1*01:03 (odds ratio 50.0, 95 percent confidence interval 16.4 to 152.3 in the Oxford imputed data).
• Neutralising autoantibodies against IL-10 and HLA-DRB1*01:03 in paediatric patients with inflammatory bowel disease (Gharahdaghi, Yeh, Vadakethala and colleagues 2026, Gut, doi:10.1136/gutjnl-2026-339329)
  A cross-sectional study across paediatric IBD cohorts in four countries found anti-IL-10 in 26 of 1045 children (2.5 percent) and answered much of the phenotype half of this issue directly: more difficult-to-treat disease (23 percent versus 6 percent, p equals 0.03), more acute severe ulcerative colitis (26 percent versus 6 percent, p equals 0.038) and more colectomy (27 percent versus 6 percent, p equals 0.01) than matched controls.
• IL-10-neutralising autoantibodies in paediatric-onset inflammatory bowel disease (Yalcinkaya, Sandstrom, Degrace and colleagues 2026, Gut, doi:10.1136/gutjnl-2026-338876)
  An independent paediatric cohort (IBD n equals 239, autoimmune enteropathy n equals 37, congenital diarrhoea and enteropathies n equals 42) found five neutralising cases at 2.5 ng/mL of IL-10 and three at 5 ng/mL, a prevalence of 2.1 percent (95 percent confidence interval 0.7 to 4.8), only among patients with IBD and only among those without a known monogenic cause, and showed persistence 3 and 9 years after baseline.

Searched

The citing-works file for this paper (50 works), then Europe PMC for autoantibodies with interleukin-10 and inflammatory bowel disease, and for HLA-DRB1*01:03 with ulcerative colitis.

Still open

Prevalence is now measured in cross-section at three sites, but no study has measured it in consecutive incident cases before treatment, no study has compared the three published assays on one shared serum panel so the thresholds are not interchangeable, and the phenotype associations are retrospective and prevalence-weighted rather than prospective.

Registry query

0 of 49 records in the open node mention any of: anti-il-10 autoantibody, inflammatory bowel disease, incident cohort, il-10 reporter assay, colectomy, hla-drb1*01:03. The open node holds biology and chemistry so far, so this is the expected answer; a private node of a clinical group would be asked the same question.

2. A proposed experiment

Question

In consecutive treatment-naive patients newly diagnosed with inflammatory bowel disease, what proportion carry neutralising anti-IL-10, and does it predict colectomy at two years?

Hypothesis

Neutralising anti-IL-10 is present in 2 to 4 percent of incident inflammatory bowel disease at diagnosis and predicts more colectomy or difficult-to-treat disease at two years than seronegative cases.

Attempt type

observational-study

Design

A population-based inception cohort across several regional inflammatory bowel disease networks that already capture consecutive incident cases. Serum is taken at diagnosis before any immunosuppression, then at 6, 12 and 24 months, with disease activity, therapy escalation, hospitalisation, acute severe colitis and colectomy recorded prospectively. Every sample is run blind in the cellular interleukin-10 reporter neutralisation assay at both 2.5 and 5 ng/mL of interleukin-10, the thresholds used in the paediatric Gut study, with confirmatory competitive enzyme-linked immunosorbent assay, so that a positive call is defined before enrolment rather than after. A nested harmonisation panel of about 100 aliquots, enriched for known positives and known negatives, is distributed to the three laboratories that published the earlier assays and read independently. Primary analysis: seroprevalence at diagnosis with a confidence interval, stratified by age at onset and by disease type. Secondary analysis: Cox regression for colectomy and for advanced therapy use, with seropositivity as the exposure, adjusted for HLA-DRB1*01:03 carriage.

Controls

• reference-case Blinded aliquots of serum from patients already confirmed positive and negative in the published cohorts, inserted into every assay run.
  Rules out: That a low seroprevalence reflects a failing assay or a drifted threshold rather than a genuinely low rate in incident disease.
• background Serum from people without inflammatory bowel disease, run in the same batches; the New England Journal of Medicine 2026 study found 0 of 1006 such controls positive.
  Rules out: That the assay produces a background rate of positives in healthy people, which would make any prevalence figure in patients uninterpretable.
• blinding-check Assay operators receive samples in a scrambled order with no clinical label, and are asked at the end to guess which were patients.
  Rules out: That knowledge of diagnosis or outcome influenced the positive and negative calls.
  Vocabulary note: The kind is defined for allocation guessing in a trial; here it is applied to the laboratory reader, who should not be able to guess case status better than chance.
• permutation-null The seropositivity labels are shuffled and the colectomy and escalation analyses rerun to give the distribution of the association under no effect.
  Rules out: That the phenotype association is an artefact of a small number of positives spread across many tested outcomes.

Sensitivity

The target effect can be taken from the literature but the required cohort size cannot be derived from it without a power calculation, which this record does not supply. The two published paediatric studies found 26 positives among 1045 patients and 5 among 239, and the adult study 173 among 4909, which sets the order of magnitude of enrolment needed to see a usable number of positive cases. The effect to be detected prospectively is the reported colectomy difference of 27 percent versus 6 percent and the difficult-to-treat difference of 23 percent versus 6 percent. A formal sample size needs those rates recomputed as incidence in treatment-naive patients, which no published study reports.

Would count as

Supported → positive · opposed → negative-no-association
Inconclusive if: Fewer than about twenty seropositive incident cases accrue, or the harmonisation panel shows the three laboratories disagree on positive calls, in which case the prevalence figure stands but the phenotype comparison is inconclusive-underpowered rather than negative.

Known before

• HLA-DRB1*01:03 carriage, which is strongly associated with anti-IL-10 seropositivity, independently predicts colectomy, perianal disease and advanced therapy use across 43 762 patients with inflammatory bowel disease. (Zhang, Shakweh, Sharip and colleagues 2026, The Lancet Gastroenterology and Hepatology, doi:10.1016/s2468-1253(26)00113-5)
  So: Any phenotype association found for the autoantibody must be adjusted for HLA-DRB1*01:03, or the design cannot tell the antibody apart from the allele that predicts both.
• HLA-DRB1*01:03 is a reproducible biomarker of colectomy risk in ulcerative colitis. (Cushing-Damm, Yin, Chen and colleagues 2026, Gastroenterology, doi:10.1053/j.gastro.2026.09.017)
  So: Confirms the confounder is reproducible, so HLA typing is a required covariate rather than an optional one; no abstract was available in Europe PMC, so only the title claim is used.
• The association between HLA-DRB1*01:03 and severe ulcerative colitis was established before this paper. (Vestergaard, Nohr, Allin and colleagues 2024, JAMA, doi:10.1001/jama.2024.20429)
  So: Shows the severity signal predates the autoantibody discovery, which is the reason the design has to separate the two rather than report the antibody association alone; no abstract was available, so only the title claim is used.
• Neutralising anti-granulocyte macrophage colony stimulating factor autoantibodies appear years before diagnosis and predict complicated Crohn's disease. (Mortha, Remark, Del Valle and colleagues 2022, Gastroenterology, doi:10.1053/j.gastro.2022.05.029)
  So: Sets the precedent that an anti-cytokine autoantibody in inflammatory bowel disease can be both prevalent and prognostic, and provides the template this cohort should follow for sampling before treatment.

Searched: Europe PMC for autoantibodies with interleukin-10 and inflammatory bowel disease, for HLA-DRB1*01:03 with ulcerative colitis, and for anti-granulocyte macrophage colony stimulating factor autoantibodies with Crohn's disease and prediction; I looked for and did not find any population-based incidence study of any anti-cytokine autoantibody in inflammatory bowel disease, or any published cross-laboratory comparison of anti-IL-10 neutralisation assays.

Feasibility

A national inflammatory bowel disease inception-cohort network with central biobanking, such as the United Kingdom IBD BioResource together with paediatric centres, plus one reference immunology laboratory for the assay · 4 to 5 years, of which 2 to 3 years is enrolment and 2 years is follow-up · moderate, because the cohort infrastructure and the assays already exist and the added cost is serial sampling, central assay running and HLA typing
Blockers: No published threshold that the three laboratories agree on, which the harmonisation panel is meant to fix and which could instead reveal that the assays are not comparable; At a prevalence of a few percent, even a large incident cohort yields few positives, so the phenotype analysis may stay underpowered; Serum before any treatment is hard to obtain in severe presentations, where steroids are often started before the diagnostic sample is banked

Strongest objection

A specialist would say the question has already been answered well enough and that another cohort buys little: with a prevalence of two to four percent, a prospective incident study large enough to test phenotype associations is bigger than any single network can enrol, and the existing retrospective cohorts of 4909 adults and 1045 children already have more positives than a realistic incident cohort would. The stronger criticism is that the real gap is assay comparability, not prevalence, and that the harmonisation panel should be the whole study rather than a nested component.

Not derivable from what is published

• The sample size and power for the two-year outcome comparison, because no published study gives event rates in treatment-naive seropositive patients
• The threshold that defines a positive call across the three published assays, and whether the 2.5 and 5 ng/mL reporter thresholds correspond to any competitive enzyme-linked immunosorbent assay cut-off
• The expected incidence of seroconversion after diagnosis, which decides whether 6, 12 and 24 month sampling is frequent enough
• Whether titre, as opposed to a binary positive call, tracks disease activity, since the record reports titres only as mean fluorescence intensity in a figure

3. Send it to a lab

Send to a lab

This design is a study in people. It cannot be sent to a laboratory service; it needs a trial network, ethics approval and a sponsor.
No bench job exists: this is a prospective patient cohort with consent, clinical follow-up and surgical outcomes, which a lab-automation service cannot run. The nested assay harmonisation panel could in principle be subcontracted, but it depends on patient samples held under existing cohort consent rather than on material a requester could ship.

The authors say plainly that they cannot exclude the antibody arising after inflammation has started, and the only way to separate cause from consequence is to look for the antibody in serum banked before any symptoms, which existing prediagnostic cohorts make possible without recruiting anyone.

1. Has anyone done this since?

Status

partly addressed

Works since

• Deciphering the prodrome of inflammatory bowel disease up to 10 years before disease onset by massively parallel serology (Bourgonje, Innocenti, Livanos and colleagues 2026, Gut, doi:10.1136/gutjnl-2025-337762)
  This is the cohort the design needs: 2000 longitudinal serum samples from 200 individuals who developed Crohn's disease, 200 who developed ulcerative colitis and 100 matched controls, collected up to 10 years before diagnosis. It profiled antibody repertoires by phage-display immunoprecipitation sequencing against 357 000 peptides and reported autoantibody signatures in pre-ulcerative colitis, but a peptide display method is not the functional neutralisation assay, and the abstract reports no anti-IL-10 result, so the question is untouched in these samples.
• IL-10-neutralising autoantibodies in paediatric-onset inflammatory bowel disease (Yalcinkaya, Sandstrom, Degrace and colleagues 2026, Gut, doi:10.1136/gutjnl-2026-338876)
  Bears on the question in both directions. The autoantibody persisted 3 and 9 years after baseline under standard immunosuppression, which argues against it being a passing consequence of active inflammation, but it disappeared after haematopoietic stem-cell transplantation and the authors say it may have declined with colectomy or ileostomy, which is what an inflammation-driven epiphenomenon would do.
• Interleukin-10 Autoantibodies and HLA-DRB1*01:03 in Inflammatory Bowel Disease (Gharahdaghi, Yeh, Ceron-Gutierrez and colleagues 2026, New England Journal of Medicine, doi:10.1056/nejmoa2513654)
  An odds ratio of 50.0 for a germline class two human leukocyte antigen allele, HLA-DRB1*01:03, places an inherited risk factor upstream of the antibody, which is hard to reconcile with the antibody being purely a product of inflammation; but the same allele is itself a risk allele for severe ulcerative colitis, so the argument is suggestive rather than decisive.

Searched

The citing-works file, then Europe PMC for prediagnostic and prodromal serology in inflammatory bowel disease, for cytokine autoantibody persistence, and for anti-integrin autoantibodies antedating ulcerative colitis.

Still open

Nobody has run a functional interleukin-10 neutralisation assay on serum banked before symptom onset. Until that is done, the germline association and the persistence data remain indirect, and the possibility that the antibody follows inflammation in most seropositive patients stays open.

Registry query

1 of 49 records in the open node mention any of: anti-il-10 autoantibody, prediagnostic serum, inflammatory bowel disease, seroconversion, il-10 neutralisation assay, causality. robin-damd-candidates-r1: Robin proposes therapeutic candidates to enhance RPE phagocytosis (round 1)

2. A proposed experiment

Question

In serum banked years before inflammatory bowel disease was diagnosed, is neutralising anti-IL-10 autoantibody already present, and if so how long before diagnosis does it appear?

Hypothesis

Neutralising anti-IL-10 is detectable in a subset of prediagnostic sera at least one year before symptom onset, at a rate above that in matched controls who never developed inflammatory bowel disease.

Attempt type

secondary-analysis

Design

A reanalysis of archived serum, with no new recruitment. Use the prediagnostic cohort already assembled for the prodrome study: serial samples from individuals who later developed Crohn's disease, individuals who later developed ulcerative colitis, and matched controls who did not, spanning up to ten years before diagnosis. Run every available time point in the cellular interleukin-10 reporter neutralisation assay at 2.5 and 5 ng/mL of interleukin-10, with confirmatory competitive enzyme-linked immunosorbent assay on any positive, all read blind to case status and to collection date. Primary analysis: the proportion of future cases positive at the earliest available time point, against controls. Secondary analysis: for each positive individual, the earliest positive sample and whether the result is sustained across later samples, plotted against the date of diagnosis. Because the cohort was sampled on a fixed military schedule rather than a clinical one, time before diagnosis is an interval, not a date, and should be reported as such.

Controls

• reference-case Blinded aliquots from patients known to be anti-IL-10 positive and known to be negative, distributed through the assay runs.
  Rules out: That an absence of prediagnostic positives is assay failure or loss of antibody activity in long-stored serum rather than true absence.
• background The matched controls in the same cohort who never developed inflammatory bowel disease, run in the same batches.
  Rules out: That any prediagnostic positives simply reflect the background rate of this autoantibody in the general population.
• instrument-noise-floor Repeat testing of split aliquots of the same serum across runs and across storage times, to establish the assay's own repeatability on archived material.
  Rules out: That a scattering of single positive time points is assay drift rather than genuine intermittent seropositivity.
• permutation-null Case and control labels, and collection dates, are shuffled and the time-to-diagnosis analysis rerun.
  Rules out: That an apparent rise in seropositivity towards diagnosis is an artefact of unequal sampling density between cases and controls.

Sensitivity

Partly derivable. The New England Journal of Medicine 2026 study found 0 of 1006 non-inflammatory bowel disease controls positive, so the expected background is close to zero and even a small number of prediagnostic positives would be informative. What cannot be derived is the negative side: with roughly 200 future Crohn's disease and 200 future ulcerative colitis individuals and a prevalence in established disease of about 2 to 3.5 percent, the expected number of positives is small, so finding none would not establish absence. A real negative needs either a much larger prediagnostic set or a prediagnostic cohort enriched for HLA-DRB1*01:03 carriers, and the number required cannot be set without a power calculation on the prevalence confidence intervals, which the published studies do not provide in a form this design can reuse.

Would count as

Supported → positive · opposed → inconclusive-underpowered
Inconclusive if: No prediagnostic sample is positive while the reference-case positives also read weakly, or serum stored for a decade loses neutralising activity, in which case the result says nothing about when the antibody appeared.

Known before

• Neutralising anti-granulocyte macrophage colony stimulating factor autoantibodies were detectable years before the diagnosis of Crohn's disease and predicted complicated disease. (Mortha, Remark, Del Valle and colleagues 2022, Gastroenterology, doi:10.1053/j.gastro.2022.05.029)
  So: This is the direct methodological precedent: an anti-cytokine autoantibody measured in prediagnostic serum from a military cohort, which shows the design is feasible and sets the expectation that the antibody can precede disease.
• Anti-integrin alpha-v beta-6 autoantibodies antedate the development of ulcerative colitis. (Livanos, Dunn, Fischer and colleagues 2023, Gastroenterology, doi:10.1053/j.gastro.2022.12.042)
  So: A second precedent for autoantibodies preceding ulcerative colitis in the same style of cohort, and it is cited by this paper itself, so the comparison is one the authors already accept.
• Near-complete blockade of interleukin-10 signalling is required to induce intestinal inflammation, and even complete biallelic loss of function produces a spectrum of disease. (This paper's Discussion, summarising existing genetic and animal studies, doi:10.1056/NEJMoa2312302)
  So: If only near-complete blockade matters, then a prediagnostic positive at low titre does not by itself establish a causal role, so the design should report titre or neutralising capacity at each time point rather than a bare positive call.

Searched: Europe PMC for prodromal and prediagnostic serology in inflammatory bowel disease, for anti-cytokine autoantibodies preceding diagnosis, and for anti-integrin alpha-v beta-6 antedating ulcerative colitis; the integrin paper was located through this paper's own reference list rather than by search, because my search string for it returned nothing.

Feasibility

The holders of an existing prediagnostic serum cohort, together with one of the laboratories that runs the interleukin-10 reporter neutralisation assay · 9 to 18 months, almost all of it sample access approval and assay running · low, because no recruitment or follow-up is needed and the assay is already established
Blockers: Access to the prediagnostic cohort depends on the holders agreeing to a new assay on a limited volume of irreplaceable archived serum; Serum volume per archived time point may be too small for a reporter assay plus a confirmatory enzyme-linked immunosorbent assay; Unknown stability of neutralising activity in serum stored for up to a decade, which the instrument-noise-floor control can characterise but not repair

Strongest objection

A specialist would object that a prediagnostic positive would not settle causality either way. Subclinical gut inflammation is known to precede diagnosis by years, so an antibody found eight years before diagnosis could still be a response to inflammation that had already begun, and the design has no way to date the onset of inflammation independently. The honest version of this objection is that the study can only move the antibody earlier in the sequence, not out of it, and that a mechanistic test in a model system or a therapeutic reversal in seropositive patients carries more causal weight than any serology.

Not derivable from what is published

• The number of prediagnostic individuals needed to make a null result interpretable
• Whether neutralising activity survives a decade of frozen storage, which no published study reports for this antibody
• The serum volume available per archived time point, and so how many assays each sample can support
• An independent marker of when subclinical inflammation began in each individual, without which an early positive cannot be called pre-inflammatory

3. Send it to a lab

Send to a lab

This design is a reanalysis of existing data. There is nothing for a bench to run; it needs the data holder.
This is a reanalysis of an existing archived serum collection under its current consent, so there is no new bench job to commission; the assay work is tied to samples the cohort holder controls and cannot be fulfilled from material a requester could supply.

The authors offer a trigger hypothesis in one sentence and cite the existence of virus-encoded interleukin-10 homologues as its basis; nobody has tested whether patient antibody actually binds a viral interleukin-10, which is a bench question that can be answered on serum that already exists.

1. Has anyone done this since?

Status

partly addressed

Works since

• Deciphering the prodrome of inflammatory bowel disease up to 10 years before disease onset by massively parallel serology (Bourgonje, Innocenti, Livanos and colleagues 2026, Gut, doi:10.1136/gutjnl-2025-337762)
  Reports elevated herpesvirus-directed antibody responses, Epstein-Barr virus in particular, up to ten years before a diagnosis of Crohn's disease. That supports the exposure half of the hypothesis, that herpesvirus responses are raised early, but it tested peptide binding across a large library and did not test cross-reactivity with human interleukin-10 or any neutralising function.
• Interleukin-10 Autoantibodies and HLA-DRB1*01:03 in Inflammatory Bowel Disease (Gharahdaghi, Yeh, Ceron-Gutierrez and colleagues 2026, New England Journal of Medicine, doi:10.1056/nejmoa2513654)
  Identifies a class two human leukocyte antigen allele as the dominant risk factor for seropositivity, which supplies the susceptibility half of the authors' sentence and makes a trigger study worth doing in carriers specifically, but says nothing about which antigen primes the response.

Searched

The citing-works file, then Europe PMC for viral interleukin-10 with Epstein-Barr virus and cross-reactive autoantibodies, which returned no hits, and for molecular mimicry with cytokine autoantibodies and epitope mapping, which returned work on other antigens only.

Still open

Everything specific to this hypothesis. No published work tests whether anti-IL-10 autoantibody from a patient binds a virus-encoded interleukin-10 homologue, nor whether seropositivity tracks Epstein-Barr virus or cytomegalovirus serostatus within a carrier population.

Registry query

0 of 49 records in the open node mention any of: anti-il-10 autoantibody, viral il-10 homologue, epstein-barr virus, molecular mimicry, epitope mapping, cross-reactivity. The open node holds biology and chemistry so far, so this is the expected answer; a private node of a clinical group would be asked the same question.

2. A proposed experiment

Question

Does anti-IL-10 autoantibody from seropositive patients cross-react with virus-encoded IL-10 homologues, and does pre-adsorption on them remove its neutralising activity?

Hypothesis

Immunoglobulin G from anti-IL-10 positive patients binds viral IL-10 homologues of Epstein-Barr virus and cytomegalovirus, and adsorption on them removes most neutralising activity against IL-10.

Attempt type

wet-lab

Design

Immunoglobulin G is purified from serum of anti-IL-10 positive patients, from Epstein-Barr virus and cytomegalovirus seropositive patients with inflammatory bowel disease who are anti-IL-10 negative, and from healthy donors. Each preparation is tested in parallel by enzyme-linked immunosorbent assay against recombinant human interleukin-10 and against recombinant virus-encoded homologues, the Epstein-Barr virus BCRF1 product and the human cytomegalovirus UL111A product, with an unrelated recombinant cytokine of similar size as the irrelevant antigen. Binding is then mapped by overlapping peptide scan across human interleukin-10 and the viral homologues to locate shared and unshared epitopes. The functional arm is a depletion experiment: each immunoglobulin G preparation is pre-adsorbed on immobilised viral homologue, on immobilised human interleukin-10, or on an irrelevant protein, and the flow-through is tested in the cellular interleukin-10 reporter neutralisation assay used in the original paper. Read-out is the loss of neutralising activity after each adsorption, with three independent immunoglobulin G preparations per group and triplicate assay wells.

Controls

• designated A commercial neutralising anti-human interleukin-10 monoclonal antibody of known epitope, run through the same binding and depletion steps.
  Rules out: That a failure to deplete reflects a broken adsorption step rather than a genuine absence of cross-reactivity, since this antibody must be depleted by immobilised human interleukin-10.
• isotype Immunoglobulin G at matched concentration from healthy donors and from an irrelevant specificity, in every binding and neutralisation plate.
  Rules out: That the observed binding or inhibition is a non-specific effect of adding purified immunoglobulin G at high concentration.
• background An unrelated recombinant protein of similar size used both as a coating antigen and as the irrelevant adsorption matrix.
  Rules out: That binding reflects the recombinant expression system, a tag, or sticking to the plate, rather than recognition of interleukin-10 sequence.
• within-subject For each patient preparation, the same immunoglobulin G is split across the three adsorption conditions and read in the same assay run.
  Rules out: That differences in depletion come from donor-to-donor variation in antibody concentration rather than from the antigen used to adsorb.
• active-comparator Immunoglobulin G from patients with inflammatory bowel disease who are Epstein-Barr virus or cytomegalovirus seropositive but anti-IL-10 negative.
  Rules out: That any viral-homologue binding simply follows from past viral exposure, which most adults have, rather than tracking the autoantibody.
  Vocabulary note: The comparator is a patient group rather than a treatment: virus-seropositive, anti-IL-10 negative patients with inflammatory bowel disease.

Sensitivity

Cannot be set from the published record. Neither the original paper nor the follow-up studies report an anti-IL-10 titre in mass units, a binding affinity, or the fraction of total immunoglobulin G that is anti-IL-10, so there is no basis for stating the smallest depletion this design could detect. What is needed is a pilot: the half-maximal inhibitory concentration of patient immunoglobulin G in the reporter assay, the well-to-well variation of that assay across three runs, and the capacity of the adsorption matrix. With those three numbers, the detectable loss of neutralising activity follows; without them, the design can state only that complete depletion against human interleukin-10 must be demonstrable in the designated control before any partial depletion against a viral homologue is interpreted.

Would count as

Supported → positive · opposed → negative-no-effect
Inconclusive if: The designated monoclonal control is not itself depleted by immobilised human interleukin-10, or patient immunoglobulin G loses neutralising activity through the purification and adsorption steps regardless of the antigen used, in which case the result is inconclusive-controls-failed rather than evidence against cross-reactivity.

Known before

• Several virus families independently acquired a cellular interleukin-10 gene, so virus-encoded interleukin-10 homologues exist and have diverged within viral genomes. (Ouyang, Rakus, van Beurden and colleagues 2014, Journal of General Virology, doi:10.1099/vir.0.058966-0)
  So: This is the reference the authors cite for the trigger hypothesis, and it is what makes the design possible: it establishes that viral interleukin-10 homologues are real targets to express and test, rather than a hypothetical antigen.
• Neutralising autoantibodies against interferon gamma, interleukin-23, type one interferon, granulocyte macrophage colony stimulating factor and interleukin-6 each produce phenotypes resembling loss of function of the matching receptor. (This paper's Introduction, citing Tangye 2022, Ku 2020 and Cheng and Holland 2023, doi:10.1038/s41577-023-00933-2)
  So: Anti-cytokine autoantibody phenocopies are a recognised class, so a trigger mechanism that applies to one cytokine should be tested against the others; if the viral mimicry route holds for interleukin-10 it predicts similar homologue cross-reactivity elsewhere, which is a cheap extension of this design.
• Non-neutralising autoantibodies against interleukin-10 signalling components occur in patients with inflammatory bowel disease. (Frede, Glocker, Wanders and colleagues 2014, BMC Immunology, doi:10.1186/1471-2172-15-10)
  So: Any epitope map must compare neutralising with non-neutralising patient antibody, because the interesting question is whether the viral homologue shares the neutralising epitope specifically rather than any interleukin-10 epitope.

Searched: Europe PMC for viral interleukin-10 with Epstein-Barr virus and cross-reactive autoantibodies, which returned zero hits, for molecular mimicry with cytokine autoantibody and epitope mapping, which returned only unrelated antigens, and the paper's own reference list for the virus-encoded interleukin-10 literature; I looked for and did not find any published epitope map of human anti-IL-10 autoantibody.

Feasibility

A clinical immunology laboratory with access to anti-IL-10 positive serum from the published cohorts, working with a protein production facility or a contract laboratory for the recombinant viral homologues and peptide scans · 6 to 12 months · moderate, because recombinant expression of the viral homologues and overlapping peptide scans are the main costs and the serum already exists
Blockers: Access to serum from anti-IL-10 positive patients, who number a few dozen worldwide in the published cohorts, and in volumes large enough for immunoglobulin G purification; Correctly folded recombinant viral interleukin-10, since a conformational epitope will not be detected by a misfolded protein or by linear peptides alone; The record does not say whether the autoantibody recognises a linear or a conformational epitope, which decides whether the peptide scan is informative at all

Strongest objection

A specialist would say the premise may be backwards. Virus-encoded interleukin-10 homologues are immunosuppressive agonists that the virus benefits from, so a host response to them is plausible, but the sequence divergence between viral and human interleukin-10 may be exactly where the neutralising epitope sits, in which case no cross-reactivity is expected and a negative result tells you nothing about a trigger. The more likely trigger, they would add, is a germline-encoded human leukocyte antigen restricted response to human interleukin-10 itself, given the odds ratio of 50 for HLA-DRB1*01:03, which would make the viral mimicry test a side issue rather than the main one.

Not derivable from what is published

• Which viral homologues to prioritise, since the record names Epstein-Barr virus and cytomegalovirus only as examples and gives no sequence or structural basis for the choice
• The sequence identity between human and viral interleukin-10 at the neutralising epitope, which decides whether cross-reactivity is even plausible
• Whether the patient autoantibody epitope is linear or conformational, and so whether a peptide scan can map it
• The pilot variance and half-maximal inhibitory concentration needed to state a detectable depletion
• Epstein-Barr virus and cytomegalovirus serostatus of the two index patients, which the paper does not report

3. Send it to a lab

The authors treated one child with rituximab and raise humorally targeted therapy as a possibility; the follow-up work has now defined the group who would be offered it, patients with anti-IL-10 and difficult-to-treat disease, but nobody has run a controlled trial, so the design has to supply the comparator, the biomarker endpoint and the stopping rule that a single case cannot.

1. Has anyone done this since?

Status

partly addressed

Works since

• Neutralising autoantibodies against IL-10 and HLA-DRB1*01:03 in paediatric patients with inflammatory bowel disease (Gharahdaghi, Yeh, Vadakethala and colleagues 2026, Gut, doi:10.1136/gutjnl-2026-339329)
  Defines the trial population rather than treating it: of 1045 paediatric patients with inflammatory bowel disease, 26 were anti-IL-10 positive and had more difficult-to-treat disease (23 percent versus 6 percent), more acute severe ulcerative colitis (26 percent versus 6 percent) and more colectomy (27 percent versus 6 percent) than matched controls. That is the unmet need a trial would address, and 80 percent of the positives with human leukocyte antigen data carried HLA-DRB1*01:03.
• IL-10-neutralising autoantibodies in paediatric-onset inflammatory bowel disease (Yalcinkaya, Sandstrom, Degrace and colleagues 2026, Gut, doi:10.1136/gutjnl-2026-338876)
  The closest thing to a therapeutic result: the autoantibody persisted under standard immunosuppression in two patients 3 and 9 years after baseline, but disappeared in the one patient who underwent haematopoietic stem-cell transplantation. That tells you conventional immunosuppression does not clear the antibody, which is the rationale for a B cell directed trial, but it is one transplanted patient and not a test of B cell depletion.
• Engineered TYK2 dependence enables myeloid-selective IL-10 receptor agonism (Canaria, Nguyen, Huang and colleagues 2026, Nature, doi:10.1038/s41586-026-11176-1)
  Offers a competing therapeutic route that does not depend on removing the antibody: a tetravalent nanobody-based myeloid-selective interleukin-10 receptor agonist that reduced inflammation in humanised mouse models of colitis and gave durable myeloid suppression for up to 30 days in non-human primates. A receptor agonist that the autoantibody cannot bind would bypass the problem rather than clear it.

Searched

The citing-works file, then Europe PMC for rituximab in very early onset inflammatory bowel disease, for anti-interferon gamma autoantibody treated with rituximab, and for plasma cell depletion in cytokine autoantibody disease.

Still open

No controlled study of B cell depletion in anti-IL-10 positive inflammatory bowel disease exists, in children or adults. Nothing establishes whether clearing the antibody changes disease course once inflammation is established, which patients should be offered it, or how it compares with bypassing the antibody using a receptor agonist.

Registry query

0 of 49 records in the open node mention any of: anti-il-10 autoantibody, rituximab, b cell depletion, inflammatory bowel disease, steroid-free remission, autoantibody clearance. The open node holds biology and chemistry so far, so this is the expected answer; a private node of a clinical group would be asked the same question.

2. A proposed experiment

Question

In anti-IL-10 seropositive patients with active disease despite standard therapy, does adding B cell depletion clear the autoantibody and improve clinical outcome?

Hypothesis

Adding rituximab to standard therapy in anti-IL-10 seropositive patients clears the autoantibody by week 24 and raises steroid-free clinical remission at week 52 above standard therapy alone.

Attempt type

clinical-trial

Design

A multicentre randomised double-blind add-on trial restricted by biomarker. Eligibility: anti-IL-10 seropositive on the central reporter neutralisation assay, with active inflammatory bowel disease despite an advanced therapy, identified by screening within the networks that already run the assay. Participants continue their existing therapy and are randomised to rituximab or matched placebo infusions on the schedule used in the index case, which was 1.5 grams per square metre in divided doses, three-monthly. Two primary endpoints, hierarchically ordered: first, the biomarker, meaning the proportion whose neutralising activity falls below the assay threshold at week 24; second, the clinical endpoint, steroid-free clinical remission at week 52. Peripheral B cell count by CD19 confirms target engagement in every participant at every visit. A pre-specified interim futility analysis at half the planned enrolment stops the trial if the biomarker endpoint is not separating. Follow-up continues to week 104 for relapse after the last infusion and for hypogammaglobulinaemia, which the index patient developed and still has at almost 10 years.

Controls

• placebo Matched placebo infusions added to the participant's existing therapy, which continues unchanged in both arms.
  Rules out: That improvement reflects the natural waxing and waning of colitis, intensified monitoring, or the effect of the background therapy rather than B cell depletion.
• internal-positive Peripheral CD19 positive B cell count measured at every visit, which must fall to the expected depleted range in the active arm.
  Rules out: That a null clinical result reflects a drug that never engaged its target, which is what separates a true negative from an uninformative one.
• blinding-check Participants and assessing clinicians guess allocation at week 24; the guess should be no better than chance. B cell counts are withheld from the assessing clinician.
  Rules out: That unblinding through the B cell count or through infusion reactions drove the clinical assessment.
• reference-case Blinded known-positive and known-negative sera in every batch of the central antibody assay.
  Rules out: That an apparent fall in the autoantibody is assay drift over the long run-in of a multiyear trial.

Sensitivity

Only one half is derivable. For the biomarker endpoint, the index case gives a plausible target: anti-IL-10 became undetectable in the treated patient, and the follow-up cohort shows the antibody persisting for 3 and 9 years under standard immunosuppression, so the placebo arm's expected clearance rate is close to zero and even a modest clearance rate in the active arm would separate. For the clinical endpoint, the sensitivity cannot be set: no published study reports the rate of steroid-free remission in anti-IL-10 seropositive patients on standard therapy, and the available numbers, 23 percent versus 6 percent difficult-to-treat disease and 27 percent versus 6 percent colectomy, are prevalence comparisons in observational cohorts, not event rates for a trial endpoint. What is needed is a cohort of seropositive patients followed on standard therapy with the proposed endpoint measured, which the biomarker-screening phase of this trial could itself provide before randomisation opens.

Would count as

Supported → positive · opposed → inconclusive-underpowered
Inconclusive if: The trial cannot enrol enough seropositive patients, which is the likely outcome at a prevalence of two to four percent, so it reports the biomarker endpoint and leaves the clinical endpoint undecided; or B cell counts fail to fall, in which case the result is inconclusive-controls-failed.

Known before

• In the index patient, rituximab at 1.5 grams per square metre three-monthly with intravenous immunoglobulin 2 grams per kilogram fortnightly and corticosteroid produced a prompt clinical and histological response, the autoantibody became undetectable from 31 months of age, all immunomodulatory therapy was withdrawn at 72 months, and the child remained well at almost 10 years with fecal calprotectin of 37 micrograms per gram. (This paper, Case Reports and Discussion, doi:10.1056/NEJMoa2312302)
  So: Supplies the dose, the schedule and the biomarker trajectory the trial should use, and also the strongest confound: the child received immunoglobulin, steroid and anti-tumour necrosis factor therapy at the same time, which is why the trial has to be an add-on with a placebo arm.
• The second patient in this paper, with the same autoantibody but milder disease, was managed without B cell depletion and was mostly in remission over two years on infliximab, with the titre reduced but not gone. (This paper, Case Reports, Clinical Presentation and Evaluation of Patient 2, doi:10.1056/NEJMoa2312302)
  So: Shows seropositivity alone does not justify B cell depletion, which is why eligibility must require active disease despite an advanced therapy rather than the biomarker alone.
• HLA-DRB1*01:03 carriage is associated with colectomy in ulcerative colitis, perianal disease and advanced therapy use across 43 762 patients. (Zhang, Shakweh, Sharip and colleagues 2026, The Lancet Gastroenterology and Hepatology, doi:10.1016/s2468-1253(26)00113-5)
  So: Most seropositive patients carry this allele, so a severe course may be driven by the allele as well as by the antibody, and allocation should be stratified by carriage so the two are not confounded in a small trial.
• Neutralising autoantibodies against other cytokines, including interferon gamma, define recognised anti-cytokine autoantibody syndromes in which B cell directed therapy is used. (Cheng and Holland 2023, Nature Reviews Immunology, doi:10.1038/s41577-023-00933-2)
  So: Gives the precedent class for the intervention and a comparison literature for safety and for how often antibody clearance tracks clinical response, which is the mechanism this trial assumes.

Searched: Europe PMC for rituximab with very early onset inflammatory bowel disease, which returned transplantation and case-report literature but no trial, for anti-interferon gamma autoantibodies treated with rituximab, and for plasma cell depletion in cytokine autoantibody disease, which returned nothing usable; I looked for and did not find any registered or published trial of B cell depletion selected by an anti-cytokine autoantibody in inflammatory bowel disease.

Feasibility

A paediatric and adult inflammatory bowel disease trial network with central antibody screening, such as the consortium behind the four-country paediatric cohort together with the United Kingdom IBD BioResource · 5 to 7 years, dominated by screening to find enough seropositive patients · high, because a biomarker-screened randomised trial in a group comprising two to four percent of patients requires screening thousands to randomise dozens
Blockers: Rarity: at two to four percent seroprevalence, the screening burden to randomise even 40 patients is large and may exceed what a network will fund; Equipoise: clinicians who have read the index case may be unwilling to randomise a severely affected seropositive child to placebo; Most seropositive patients are children, so a paediatric trial needs an adult safety precedent that does not yet exist for this indication; A receptor agonist that bypasses the antibody may reach trials first and remove the rationale for depleting B cells

Strongest objection

A specialist would object that the trial tests the wrong thing at the wrong time. Removing the antibody after years of established colitis may not reverse disease even if the antibody caused it, because the index patient was treated in infancy and still needed anti-tumour necrosis factor therapy for years afterwards, and the authors themselves note that even complete genetic loss of interleukin-10 signalling gives a spectrum of disease shaped by genetics and the microbiome. They would also argue the comparator is wrong: the real question is B cell depletion against other escalation options in difficult-to-treat disease, not against placebo, and that at this prevalence the honest design is an international registry with a protocolised treatment pathway rather than a randomised trial that will never finish.

Not derivable from what is published

• The rate of steroid-free clinical remission on standard therapy in anti-IL-10 seropositive patients, without which the clinical endpoint cannot be powered
• The sample size, since no published study reports the needed event rates
• The assay threshold that defines clearance of the autoantibody, and whether falling below it is the same thing as loss of neutralising capacity in vivo
• The right rituximab schedule for adults, since the only dose in the record is weight and surface-area based paediatric dosing in one child
• Whether intravenous immunoglobulin, which the index patient received throughout, is part of the mechanism or incidental

3. Send it to a lab

Send to a lab

This design is a study in people. It cannot be sent to a laboratory service; it needs a trial network, ethics approval and a sponsor.
No bench job exists: this is a randomised interventional trial in patients, with infusions, blinding, consent and multiyear clinical follow-up, none of which a lab-automation service can run. The central antibody assay and the CD19 counts are routine clinical laboratory work tied to the trial's own participants.