CAR-T in lupus
A promising experimental approach with striking early case-series results—and major unanswered questions about comparison, durability, access, and safety.
Open evidence hubAn independent, source-linked map of systemic lupus erythematosus (SLE) researchCause candidates ·Fix attempts ·Methods & editorial policyEvidence cutoff: 2026-08-05 · research information, not medical advice
Evidence atlas
Move from mechanism to clinical outcome without treating association, inference, and proof as the same thing.
Illustrative pathway map—not a claim of a single linear disease process. Select a connection in the full atlas to inspect supporting and conflicting evidence.
Deep evidence hubs
Each hub combines a plain-language overview, atomic claims, exact sources, contradictory evidence, limitations, and open research questions.
A promising experimental approach with striking early case-series results—and major unanswered questions about comparison, durability, access, and safety.
Open evidence hubOne of lupus's strongest infectious associations now has sharper human mechanistic evidence, but causation, attributable risk, and prevention remain unsettled.
Open evidence hubHuman associations and animal experiments support a disease-modifier model; clinical intervention evidence remains preliminary.
Open evidence hubA replicated molecular state and validated treatment target, with important heterogeneity and unresolved biomarker-selection questions.
Open evidence hubB-cell survival biology connects autoantibody production, validated BAFF inhibition, kidney outcomes, and deeper depletion strategies—but the interventions are not interchangeable.
Open evidence hubConnection explorer
A node names a concept. The edge between concepts is the claim that needs support, grading, and challenge.
Evidence ledger
“Contradicted” cards mark claims the available evidence does not support—not proof of the opposite claim.
Current major guidelines recommend hydroxychloroquine broadly in lupus, while dose, contraindications, interactions, and eye monitoring require individualized clinical care.
The 2025 ACR guideline recommends routine hydroxychloroquine treatment unless contraindicated; EULAR likewise places antimalarial therapy at the foundation of SLE management.
Current U.S. guideline recommendation and implementation context.
Independent international recommendations support broad antimalarial use.
Review state: Citation Audited · Record ID: care-hcq-foundational
Steroids can be essential for rapid control, but current guidelines emphasize reducing ongoing exposure because cumulative treatment can cause substantial harm.
ACR and EULAR recommendations prioritize the lowest feasible glucocorticoid exposure, with earlier use of steroid-sparing conventional or biologic therapy when needed to control disease.
Recommends minimizing glucocorticoid exposure and earlier steroid-sparing treatment.
Provides an independent treat-to-target and tapering framework.
Review state: Citation Audited · Record ID: care-glucocorticoid-minimization
There is no single ranked treatment list for everyone with lupus. Kidney disease, pregnancy plans, infections, prior responses, access, and individual goals can change the safest option.
Contemporary systemic and kidney guidelines use organ- and context-specific treatment pathways and emphasize shared decision-making rather than diagnosis-wide ranking.
Organ-specific recommendations and shared decision-making framework.
Kidney-specific treatment and monitoring context.
Review state: Citation Audited · Record ID: care-organ-context
In the United States, obinutuzumab has a lupus-nephritis indication for adults receiving standard therapy. Approval does not mean it is appropriate for every person or approved the same way everywhere.
The October 2025 U.S. prescribing information added active lupus nephritis in adults receiving standard therapy to the labeled indications for obinutuzumab.
Primary U.S. label for indication, population, administration, warnings, and adverse reactions.
Review state: Citation Audited · Record ID: care-obinutuzumab-us-ln
In small groups of people with very severe lupus that had not responded to multiple treatments, many entered remission after CD19 CAR-T. This is encouraging, but it is not yet a fair comparison with standard treatment.
A five-person compassionate-use series and a later case series that included eight participants with SLE reported marked clinical improvement and DORIS remission after autologous CD19 CAR-T plus lymphodepletion. These uncontrolled observations provide a rationale for prospective trials, not an estimate of comparative efficacy.
Five-person compassionate-use SLE series reported drug-free remission during reported follow-up.
All eight SLE participants in the 15-person autoimmune series met DORIS remission in the reported follow-up.
Defines what DORIS remission does and does not mean.
Review state: Citation Audited · Record ID: cart-early-remission-refractory-sle
In early studies, B cells returned after a period of depletion and were mostly immature or naive cells. Researchers call this an immune-reset hypothesis; it has not been shown to permanently restore tolerance.
The early SLE reports describe B-cell aplasia followed by reconstitution dominated by naive B-cell phenotypes. This is consistent with, but does not prove, a durable reset of autoreactive memory or immune tolerance.
Reported B-cell depletion followed by reconstitution with naive phenotypes in the small SLE cohort.
Follow-up supports a reconstitution pattern but remains too short and small to establish permanent tolerance.
Review state: Citation Audited · Record ID: cart-reconstitution-naive-b-cells
The first published lupus cohorts mostly reported low-grade cytokine release syndrome, but CAR-T also involves chemotherapy, a period of low B cells, and infection risk. Small studies cannot reveal rare harms.
In the 15-person autoimmune case series, ten participants had grade 1 CRS; one each had grade 2 CRS, grade 1 ICANS, and pneumonia requiring hospitalization. The sample is inadequate for stable incidence estimates or comparison with oncology populations.
Primary safety observations from 15 autoimmune-disease participants.
Ongoing follow-up specifically measures adverse events for up to two years.
Review state: Citation Audited · Record ID: cart-safety-bounded
CAR-T and rituximab should not be ranked from the available studies. The CAR-T reports involve selected patients and do not directly compare the treatments.
The current source set contains no randomized or suitably controlled head-to-head SLE study of CD19 CAR-T versus rituximab. Mechanistic arguments about depth of B-cell depletion cannot substitute for comparative clinical evidence.
The report had no rituximab control arm.
The follow-up series had no comparative treatment arm.
The phase 2 study is open-label rather than a rituximab comparison.
Review state: Citation Audited · Record ID: cart-no-rituximab-superiority
Several formal studies are now following people with severe lupus who receive different CD19 CAR-T products. A trial listing tells us what researchers plan to measure; it does not show that the treatment works.
Phase 1/2 and phase 2 programs are evaluating safety, cellular kinetics, DORIS remission, and renal response in severe or treatment-refractory SLE. Registry status and enrollment are operational metadata and should be refreshed independently of scientific claims.
Registered phase 1/2 lupus-nephritis study.
Registered phase 1/2 severe refractory SLE study.
Registered phase 2 active SLE study.
Review state: Citation Audited · Record ID: cart-controlled-evidence-pending
People with lupus are more likely than controls to have some signs of previous or active EBV exposure. Because EBV is already very common and most studies look backward in time, this does not prove that the virus caused an individual's lupus.
A meta-analysis of case-control studies found higher anti-VCA IgG seroprevalence in SLE but not a statistically significant difference for anti-EBNA1 IgG; other markers also varied. The evidence supports an association while leaving temporal order, residual confounding, and publication bias unresolved.
Found significant associations for some EBV antibody markers and explicitly reported marker-specific null findings and possible publication bias.
Review state: Citation Audited · Record ID: ebv-epidemiologic-association
Researchers found a small population of EBV-infected B cells in lupus with gene activity that could help present self-antigens and activate other immune cells. It is a plausible mechanism, not yet a complete causal chain.
Single-cell and functional analyses identified transcriptionally distinct EBV-positive B cells in SLE and implicated EBNA2/RBPJ-linked antigen-presentation programs and activation of autoreactive CD4 T cells. This materially strengthens a mechanistic model but does not show that the program initiates most SLE.
Primary human single-cell and functional study of EBV-positive B cells in SLE.
Earlier epidemiology supports association but not necessity or sufficiency.
Review state: Citation Audited · Record ID: ebv-reprograms-b-cells
EBV may contribute to lupus in some people, but it is not accurate to say that the virus explains all—or most—cases from the evidence available here.
The evidence base supports epidemiologic association and a plausible human mechanism. It does not establish necessity, sufficiency, attributable fraction, or a single primary initiating mechanism across genetically and clinically heterogeneous SLE.
Reported publication bias concerns, methodological limitations, and a null association for anti-EBNA1 IgG.
Provides a mechanism but is not a prospective causal or prevention study.
Review state: Citation Audited · Record ID: ebv-not-sole-cause
An EBV vaccine is an important research idea, but no lupus prevention trial has shown that vaccination lowers lupus risk.
The mechanistic and epidemiologic evidence creates a testable prevention hypothesis. It does not justify recommending EBV vaccination as an SLE-prevention intervention outside applicable vaccine research or future approved indications.
Supplies biological rationale for prevention research, not prevention efficacy.
Association is insufficient to estimate a vaccine prevention effect.
Review state: Citation Audited · Record ID: ebv-prevention-untested
CAR-T removes many CD19-positive B cells, which can include cells carrying EBV. The studies do not show that removing EBV-infected cells is why lupus improved.
The EBV-positive B-cell model and CAR-T remission observations are biologically connectable, but no cited study measures EBV-positive clone removal as the mediator of clinical remission or compares outcomes by EBV burden.
Identifies a plausible CD19-positive EBV reservoir.
Reports remission but does not test EBV clearance as a mediator.
Review state: Citation Audited · Record ID: ebv-cart-mechanism-unproven
Across several studies, people with lupus had differences in gut microbial diversity and some bacterial groups compared with healthy controls. Medicines and active disease can also change the microbiome, so the direction of cause is unclear.
A meta-analysis of 11 case-control studies found lower Shannon diversity and Chao1 richness in SLE. Taxonomic signals were less uniform, and treatment exposure is an important confounder.
Pooled 11 case-control studies and reported lower diversity/richness measures in SLE.
Review state: Citation Audited · Record ID: microbiome-dysbiosis-association
One human study found more R. gnavus in lupus and stronger antibody responses to particular strains in people with active kidney disease. This is a lead to investigate, not proof that the bacterium caused nephritis.
Azzouz et al. reported approximately fivefold greater overall R. gnavus representation in SLE and associations among strain-restricted anti-R. gnavus lipoglycan antibodies, disease activity, anti-native DNA, complement, and active nephritis.
Primary human observational study with validation across cohorts.
Review state: Citation Audited · Record ID: microbiome-r-gnavus-nephritis
In a 20-person pilot, 42.12% met a trial response measure after oral donor-microbiota capsules plus their usual stable treatment. Without a control group, we cannot know how much of that change came from FMT.
The EXPLORER single-arm pilot administered encapsulated FMT weekly for three weeks to 20 people with active SLE. At week 12, 42.12% met SRI-4, with changes in disease activity, anti-dsDNA, microbial composition, metabolites, and immune measures. It was a feasibility signal, not a comparative efficacy estimate.
Primary single-arm pilot and source of the 42.12% week-12 SRI-4 figure.
Defines the SRI composite and why it is not remission.
Review state: Citation Audited · Record ID: microbiome-fmt-pilot
Moving gut microbes from lupus-prone mice into germ-free mice increased some lupus-related immune signals. That supports biological plausibility but does not tell us whether a microbiome treatment helps people.
A preclinical FMT experiment reported that microbiota from lupus-prone mice increased anti-dsDNA antibodies and altered immune-cell and gene-expression measures in germ-free recipients.
Primary germ-free mouse transfer experiment.
Review state: Citation Audited · Record ID: microbiome-preclinical-transfer
The microbiome may influence disease activity, but the evidence does not support calling it the single cause or an essential driver in every person with lupus.
Human evidence is predominantly observational, while transfer evidence is preclinical. Together these support a modifier or contributor model more strongly than a universal initiating mechanism.
Human studies establish association and identify medication effects, not necessity.
Preclinical transfer shows modification of immune features, not a universal human initiating mechanism.
Review state: Citation Audited · Record ID: microbiome-not-required
No cited study shows that FMT, probiotics, or a selected bacterial mix makes CAR-T safer or more durable in lupus.
Combining microbiome manipulation with B-cell reconstitution after CAR-T is a mechanistic hypothesis. The source set contains neither a post-CAR-T intervention study nor validated organisms, doses, timing, or safety criteria for such a combination.
FMT pilot did not involve CAR-T.
CAR-T case series did not include microbiome intervention.
Review state: Citation Audited · Record ID: microbiome-adjunct-unproven
Many—but not all—people with active lupus have a high blood-gene pattern associated with type I interferon activity. This is a pathway signal, not a complete explanation of lupus or a stand-alone diagnostic test.
Peripheral-blood transcriptomic studies identified an interferon-inducible gene-expression cluster in a subset of SLE samples. In the pooled phase 3 anifrolumab program, most enrolled participants were IFN-signature-high, while a distinct IFN-signature-low group remained, demonstrating molecular heterogeneity within clinically active SLE.
Primary human discovery study of the peripheral-blood interferon-inducible signature.
Prespecified IFN-high and IFN-low strata show that active SLE is not uniformly signature-high.
Review state: Citation Audited · Record ID: ifn-blood-signature-subgroup
One major anifrolumab trial met its main composite response endpoint; a closely related trial did not meet its different main endpoint. Together they support the pathway as clinically relevant while warning against describing the evidence as uniformly positive.
TULIP-2 reported week-52 BICLA response in 47.8% with anifrolumab versus 31.5% with placebo. TULIP-1 did not meet its week-52 SRI-4 primary endpoint: 36% versus 40%, respectively. Endpoint choice and restricted-medication rules contributed to interpretive complexity but do not erase the prespecified null result.
Positive prespecified BICLA primary endpoint in TULIP-2.
TULIP-1 did not meet its prespecified SRI-4 primary endpoint and bounds any claim of uniform phase 3 efficacy.
Review state: Citation Audited · Record ID: ifn-blockade-mixed-phase3
A high interferon signature may describe biology, but current trial data do not justify using a simple high-versus-low result to guarantee or rule out response to interferon blockade.
TULIP-2 reported anifrolumab-placebo BICLA differences in both IFN-high and the much smaller IFN-low strata, while TULIP-1 did not show a significant SRI-4 difference in its IFN-high subgroup. These results do not validate baseline binary IFN-gene-signature status as a stand-alone predictive biomarker.
Observed responses in the IFN-low stratum contradict a simple claim that only IFN-high participants can respond, although the low subgroup was small.
The IFN-high subgroup did not meet the prespecified SRI-4 secondary endpoint in TULIP-1.
The discovery study established association, not prospective treatment-selection performance.
Review state: Citation Audited · Record ID: ifn-signature-not-binary-selector
Anifrolumab is FDA-labeled for adults with moderate-to-severe SLE receiving standard therapy, but its label does not recommend it for severe active kidney or central-nervous-system lupus and highlights infection risks.
The April 2026 U.S. label indicates anifrolumab for adults with moderate-to-severe SLE receiving standard therapy. It states that efficacy has not been evaluated in severe active lupus nephritis or severe active CNS lupus and warns about serious infections, including increased respiratory infection and herpes-zoster risk.
Current primary U.S. source for indication, limitations of use, and warnings.
Trial report documents the higher herpes-zoster frequency and studied population underlying part of the label evidence.
Review state: Citation Audited · Record ID: ifn-label-bounded
BAFF helps B cells survive, and abnormal BAFF measures occur in many people with lupus. Levels vary between people and over time, so a blood BAFF result is not a stand-alone measure of an individual's disease activity.
A longitudinal SLE cohort found persistently or intermittently elevated serum BLyS in 50% and elevated blood BLyS mRNA in 61%, with corticosteroid-sensitive variation. A larger cohort linked prior or rising plasma BLyS with subsequent activity at the population level, while the earlier study found within-person serum changes did not track activity changes.
Demonstrates common but non-universal, treatment-sensitive BLyS dysregulation and imperfect within-person activity tracking.
Supports a longitudinal population-level relationship between plasma BLyS and later disease activity.
Clinical benefit from ligand inhibition supports pathway relevance but cannot validate circulating BAFF as an individual biomarker.
Review state: Citation Audited · Record ID: baff-expression-heterogeneous
In BLISS-52, adding belimumab to standard therapy increased the proportion meeting the SRI composite at one year. The result is an average trial effect, not proof that every person's disease is BAFF-driven.
In BLISS-52, week-52 SRI response was 51% with belimumab 1 mg/kg, 58% with 10 mg/kg, and 44% with placebo, all added to standard therapy. The study supports BAFF as a clinically actionable target in its selected population while leaving substantial nonresponse in each active-treatment arm.
Primary phase 3 source for randomized SRI response and safety results.
Defines the current U.S. indication and safety boundaries; it does not imply universal response.
Heterogeneous BAFF biology cautions against inferring that the pathway is equally dominant in all SLE.
Review state: Citation Audited · Record ID: baff-inhibition-sle-composite
In a two-year kidney trial, more participants receiving belimumab plus standard therapy met the study's kidney-response definitions than participants receiving standard therapy alone. A response definition is not the same as kidney cure.
BLISS-LN randomized 448 adults with biopsy-proven active lupus nephritis. At week 104, primary efficacy renal response was 43% with belimumab plus standard therapy versus 32% with placebo plus standard therapy; complete renal response was 30% versus 20%.
Primary randomized source for the 104-week renal response results.
Current U.S. indication includes active lupus nephritis receiving standard therapy.
Places BAFF inhibition within organ-specific, combination treatment and monitoring rather than as a stand-alone cure.
Review state: Citation Audited · Record ID: baff-ln-renal-response
Belimumab blocks soluble BAFF/BLyS, a B-cell survival signal. That differs biologically and clinically from antibody-based B-cell depletion or CAR-T, even though all can affect the B-cell system.
The belimumab label describes binding to soluble BLyS and inhibition of BLyS binding to B-cell receptors. This modulates survival and differentiation; it should not be represented as direct pan-B-cell elimination, and outcomes cannot be transferred from belimumab to CD20 depletion or CD19 CAR-T.
Primary regulatory description of belimumab's soluble-BLyS binding mechanism.
Clinical trial tests BAFF inhibition, not a B-cell-depleting product.
CAR-T directly targets CD19-positive cells after lymphodepletion, contradicting mechanistic equivalence with soluble-ligand inhibition.
Review state: Citation Audited · Record ID: baff-not-bcell-elimination
Outcome crosswalk
Trial results should be compared only after checking what each endpoint asks, when it is measured, and how missing data are handled.