B cells and plasma cells in autoimmune disease: working specification

Disease mechanisms, treatment response and patient selection

1. Objective

Determine how strongly each autoimmune indication depends on B cells, plasma cells or other mechanisms, and which observations can predict benefit from CD19-, CD20- or B-cell maturation antigen (BCMA)-directed therapy. CD19 and CD20 are cell-surface markers; BCMA is a receptor commonly expressed on antibody-secreting cells. Target expression and treatment modality must both be specified.

The initial evidence review answers this question at the indication level and identifies patient-level evidence. Rheumatoid arthritis (RA), systemic lupus erythematosus (SLE) and systemic sclerosis (SSc) receive additional discussion. Other indications provide comparisons in which antibody causality, treatment effects or biomarker interpretation are better established.

Out of scope: individual prescribing advice, an exhaustive guideline or approval inventory, and a numerical probability of response for a patient without a validated prediction model.

2. Questions

1. What shows that a B-lineage population contributes causally to disease?
2. Does the contribution involve antibodies, antigen presentation, cytokines, support of other immune cells, or several functions?
3. Does antibody production depend on continual replacement from B cells, or on established plasma cells that can survive B-cell depletion?
4. Which treatment trials establish clinical benefit, for which organ manifestation, disease stage, endpoint and background treatment?
5. Which baseline measurements predict a treatment-specific benefit?
6. What can response, relapse or nonresponse to rituximab tell us about the mechanism in an individual?
7. Which evidence supports CD19 or BCMA targeting after anti-CD20 failure, and which proposed treatment sequences remain hypotheses?

3. Evidence standards

Keep three judgments separate: mechanistic involvement, demonstrated treatment benefit and individual response prediction. A disease can have strong evidence for the first two and no validated classifier for the third.

Evidence	What it supports	Required limitation
Functional autoantibody experiments	An antibody can produce a specified effect	Confirm relevance to human disease and the population making it
Randomized clinical intervention	An incremental treatment effect in the studied population	Preserve comparator, background treatment, endpoint and time
Tissue or blood profiling	Association with a compartment or disease state	Infiltration and marker abundance do not establish causal dependence
Treatment-by-biomarker interaction	Different relative benefit by biomarker status	Require replication and prospective assessment of clinical utility
Observational response association	A candidate enrichment or monitoring marker	Confounding and prognosis can mimic treatment prediction
Uncontrolled cell-therapy series	Feasibility and a signal warranting trials	Selection, conditioning and concomitant treatment limit attribution

A marker associated with response within a treated cohort is not automatically a treatment-selection marker. The stronger test is whether the marker changes the comparative treatment effect, followed by validation in another cohort.

4. Ranking rules

The initial ranking uses ordered evidence tiers for expected clinical benefit from a specified B-lineage intervention. It is a project synthesis, not a validated score or an ordering of response percentages across diseases. Conditions within a tier are tied. An independent biological comparison records how clearly pathogenic antibodies have been established.

Split indications when mechanism or treatment evidence differs: antibody subtypes in myasthenia gravis, seropositive and seronegative RA, renal and nonrenal SLE, inflammatory activity and accumulated damage, and individual myositis syndromes. Do not label every nonresponder “plasma-cell driven.”

5. Extraction requirements

For each source, record the citation and source location, publication type, population and exclusions, randomized and analyzed counts, organ manifestation, serology, disease duration, prior therapies, target, modality, regimen, conditioning, comparator, background therapy, endpoint, assessment time, absolute result, uncertainty and relevant harms. Preserve trial amendments, missing-data handling, subgroup status and overlapping patient cohorts.

For each proposed biomarker, record whether it is diagnostic, prognostic, predictive of comparative benefit, pharmacodynamic or a relapse-monitoring marker. Record its assay, tissue, threshold, timing, external validation and what alternative explanation remains.

For treatment failure, distinguish failure to reach the relevant cells, persistence of a different population, inadequate observation time, active alternative biology, and irreversible damage. Coordinate tissue interpretation with imm-biopsy.

6. Work completed and next steps

The initial synthesis was assembled on 22 September 2026 from targeted searches of PubMed and journal sites. It includes contemporary reviews, positive and negative trials, and emerging CD19 and BCMA studies. Source access was uneven; References states exactly what was checked. This is a focused review, not a systematic review.

Next, prioritize full-methods review of patient-selection evidence in RA, SLE and SSc; extract baseline biomarkers alongside organ outcomes; and evaluate whether any classifier has prospective evidence of improving treatment choice. Resolve the outstanding source checks before building a numerical response model. Keep the ranking qualitative until comparable individual data support more detailed prediction.

Acceptance criteria for each revision:

• Every disease judgment has linked evidence and a stated population.
• Target and modality are separate; antibody and cell-therapy results are not treated as interchangeable.
• Negative trials and competing biological explanations remain visible.
• An unvalidated biomarker is labelled as such.
• Disease activity, damage, clinical benefit and antibody changes are distinct.
• Numerical effects retain their endpoint and comparator; no cross-trial response-rate league table is presented as comparative efficacy.

Document responsibilities

Document	Reader	Kind	Length
Index	Someone finding the project	How-to	Short
Specification	Someone extending the analysis	Reference	Enough to define the work and evidence standards
Evidence review	Someone comparing indications and mechanisms	Reference	Detailed enough to assess the ranking
References	Someone choosing a review or checking a claim	Reference	Annotated, with source-access limits