Plain-language evidence map

What might be causing lupus?

There is no single cause. These are the candidates the evidence currently supports, written without jargon and sorted by how confident anyone can honestly be — including the ones that have been ruled out.

How sure is sure?See what's been tried →

How to read this page

"Linked to lupus" and "causes lupus" are very different claims

Almost everything gets called a cause somewhere online. These five rungs separate what has been proven from what is merely associated — and what has been tested and found wanting.

5
Proven — in a small group
Meets a full causal standard — the change runs in families, the mechanism is understood, and reproducing the change in an animal reproduces the disease. It explains only a small fraction of lupus.
4 on this page
4
Strong suspect
Several independent lines of evidence point the same way, and interfering with it changes disease in people. It has not been shown to start the disease.
4 on this page
3
Possible contributor
A real association in people, or proof in animals, but no human experiment has yet tested whether removing it helps.
5 on this page
2
Losing support
Once widely believed, now undercut by stronger study designs. Kept visible so the record is honest.
1 on this page
1
Ruled out
Tested directly in randomized trials and not supported. Useful mainly for correcting a common belief.
1 on this page

The other question: what does it do?

Could start itProposed to be the first domino — present before disease and capable of setting it off.

Keeps it goingProposed to sustain disease once it exists. Removing it is where a cure would most plausibly come from.

Makes it worseTurns up the volume on damage that is already happening. Blocking it can control disease without ending it.

Blocks recoveryExplains why the immune system does not simply return to normal when treatment removes the obvious drivers.

This distinction matters more than it sounds. Something that keeps the disease going is where a cure would come from. Something that only amplifies it can be blocked for years without ever ending the disease.

4 candidates

Proven — in a small group

Meets a full causal standard — the change runs in families, the mechanism is understood, and reproducing the change in an animal reproduces the disease. It explains only a small fraction of lupus.

Could start itProven — in a small group

An immune alarm stuck in the on position

Researchers call this: TLR7 gain-of-function variants

A sensor inside immune cells that normally detects viral genetic material is turned up too high, so it reacts to the body's own RNA as though it were an infection.

How much of lupus this could explain: Very rare. A handful of families worldwide with confirmed variants; well under 1 in 100 people with lupus.

Why researchers take it seriously

The variant was found in a child with severe lupus and had arisen new in her, rather than being inherited from unaffected parents.
Putting the identical change into mice gave the mice lupus — the strongest form of causal proof available.
The mechanism is understood: the sensor binds the body's own RNA more tightly and lets self-reactive B cells survive.

Why it might be wrong

It is genuinely rare, so it cannot explain the great majority of lupus.
No confirmed carrier has yet received a treatment aimed at the pathway, so it is unknown whether correcting the sensor reverses established disease.

What would settle it

Treat confirmed carriers with a drug that selectively quiets this sensor and see whether disease durably corrects — not just whether the pathway readings normalise.

Tried against it: turning the overactive alarm back down

Sources (1)

TLR7 gain-of-function genetic variation causes human lupusBrown et al., Nature (2022) · human mechanistic

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Could start itProven — in a small group

The alarm gets delivered to the wrong place

Researchers call this: UNC93B1 gain-of-function variants

A shuttle protein that escorts the RNA sensor to its correct compartment malfunctions, leaving the sensor active where it should be quiet.

How much of lupus this could explain: Very rare, and mostly identified in children with lupus or chilblain lupus.

Why researchers take it seriously

Four independent research groups reported the same cluster in 2024, which makes a chance finding unlikely.
It converges on exactly the same pathway as the stuck-on sensor, from a different direction.

Why it might be wrong

Found through childhood and severe-disease clinics, so the true frequency and severity range are unknown.
No treatment has been tested in confirmed carriers.

What would settle it

An international carrier registry that finds people without going through severe-disease clinics, to measure how often carrying the variant actually leads to lupus.

Tried against it: turning the overactive alarm back down

Sources (1)

Gain-of-function human UNC93B1 variants cause systemic lupus erythematosus and chilblain lupusDavid et al., Journal of Experimental Medicine (2024) · human mechanistic

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Could start itProven — in a small group

The body cannot clear its own cellular debris

Researchers call this: DNASE1L3 loss of function, and acquired antibodies against it

An enzyme that chops up DNA released by dying cells is missing or blocked, so debris accumulates and the immune system starts treating it as a threat.

How much of lupus this could explain: Inherited loss is rare. But antibodies that block the same enzyme are found in a large share of people with lupus kidney disease — so the acquired version may be common.

Why researchers take it seriously

Children who inherit two broken copies of the gene develop lupus at close to 100% — as direct a causal link as this field has.
In ordinary adult lupus, many people make antibodies that neutralise the same enzyme, producing the same functional defect without the genetic change.
It gives a concrete physical explanation for why the immune system starts reacting to DNA at all.

Why it might be wrong

In adults it is unclear whether the blocking antibodies appear before disease or as a consequence of it.
The first attempt to replace the enzyme in a person was stopped when they developed antibodies against the drug itself.

What would settle it

A placebo-controlled trial of enzyme replacement restricted to people who actually have the defect, measuring both debris clearance and clinical outcome.

Tried against it: replacing the missing cleanup enzyme

Sources (3)

Loss-of-function variant in DNASE1L3 causes a familial form of systemic lupus erythematosusAl-Mayouf et al., Nature Genetics (2011) · human mechanistic
Autoantibody-mediated impairment of DNASE1L3 activity in sporadic systemic lupus erythematosusHartl et al., Journal of Experimental Medicine (2021) · human mechanistic
Targeted Degradation of NETs in Lupus with DNASE1L3 DeficiencyNew England Journal of Medicine correspondence (2026) · case report

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Could start itProven — in a small group

The cleanup crew is understaffed

Researchers call this: Complement C1q/C4 deficiency and C4A copy number

Complement proteins tag dying cells for quiet disposal. Carrying fewer working copies leaves more debris around for the immune system to find.

How much of lupus this could explain: Complete deficiency is very rare. Reduced gene copy number is common and shifts risk substantially — including a large part of why lupus is far more common in women.

Why researchers take it seriously

Complete C1q deficiency is one of the strongest single risk factors known for lupus.
Copy number is the strongest common genetic effect found in lupus, resolved with unusually good genetic methods.
It fits the same theme as the debris-clearance defect — failure to dispose of dead-cell material quietly.

Why it might be wrong

Penetrance figures from case series are inflated: deficient people who never develop lupus are less likely to be found and reported.
Copy number shifts risk but is not a switch. Most people carrying the risky configuration never develop lupus.
No treatment follows directly from knowing someone's copy number.

What would settle it

Unbiased population screening rather than clinic-based case series, to measure how often deficiency actually leads to disease.

Sources (1)

Complement genes contribute sex-biased vulnerability in diverse disordersKamitaki et al., Nature (2020) · case control

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

4 candidates

Strong suspect

Several independent lines of evidence point the same way, and interfering with it changes disease in people. It has not been shown to start the disease.

Keeps it goingStrong suspect

Immune cells that attack the body survive when they should be deleted

Researchers call this: Autoreactive B-cell and plasma-cell lineage

B cells that recognise the body's own tissue normally get removed during development. In lupus they survive, multiply, and produce the antibodies that cause damage.

How much of lupus this could explain: Central to essentially all lupus, though the specific antibodies and the cells producing them vary between people.

Why researchers take it seriously

Three mechanistically different ways of removing these cells all improve disease, which is far harder to explain away than a single drug result.
Removing them deeply enough has produced drug-free remission — something no ordinary lupus drug has ever achieved.
When the cells are removed, downstream inflammation signals collapse, which places these cells upstream rather than alongside.

Why it might be wrong

Depleting B cells with older antibody drugs failed in randomized trials, so simply having B cells present is not the whole story — depth and location matter.
After 50 years, no single autoantibody has been proven to be the actual driver of tissue damage.
It is not established that these cells start the disease, only that they sustain it.

What would settle it

Sequencing the actual cells at relapse after deep depletion — to find out whether disease returns from old surviving cells or from newly made ones. This has never been done anywhere in autoimmune disease.

Tried against it: rebuilding the immune system from scratch, a stronger antibody against b cells, the first attempt at removing b cells, cutting off the survival signal for b cells

Sources (5)

CD19 CAR-T cells for treatment-refractory autoimmune diseases: the phase 1/2 CASTLE basket trialMüller et al., Nature Medicine (2026) · nonrandomized interventional
CD19 CAR T-Cell Therapy in Autoimmune Disease — A Case Series with Follow-upMüller et al., New England Journal of Medicine (2024) · case series
Efficacy and safety of rituximab in moderately-to-severely active systemic lupus erythematosusMerrill et al., Arthritis & Rheumatism (2010) — EXPLORER · randomized controlled trial
Selective CAR T cell-mediated B cell depletion suppresses IFN signature in SLEWilhelm et al., JCI Insight (2024) · human mechanistic
CD19-CAR T-cell therapy induces deep tissue depletion of B cellsTur et al., Annals of the Rheumatic Diseases (2025) · human mechanistic

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Blocks recoveryStrong suspect

Antibody factories that hide where drugs cannot reach

Researchers call this: Long-lived plasma cells (LLPCs)

Some antibody-producing cells settle into bone marrow and tissue niches, lose the marker most treatments target, and keep producing autoantibodies for years.

How much of lupus this could explain: Probably a subset — enough to explain relapse in some people who were otherwise deeply treated.

Why researchers take it seriously

These cells persist after treatments that clear every other B cell, which has been directly observed.
When one person relapsed after deep depletion, switching to a treatment that targets these cells specifically re-induced remission.
Partial removal buys partial, temporary control — pointing to a threshold effect rather than a gradual one.

Why it might be wrong

Nobody knows how many of these cells are enough to sustain disease — there is no established threshold.
The clearest target-switching evidence comes from a single patient, and that patient had a different autoimmune disease.
These same cells hold protective immunity from past vaccines and infections, so removing them has a real cost.

What would settle it

A randomized comparison of standard depletion against depletion that also targets these cells, restricted to people shown in advance to carry a heavy reservoir.

Tried against it: going after the hidden antibody factories too, off-the-shelf drugs against the antibody factories

Sources (4)

Co-infusion of CD19-targeting and BCMA-targeting CAR-T cells for treatment-refractory systemic lupus erythematosus: a phase 1 trialFeng et al., Nature Medicine (2025) · nonrandomized interventional
BCMA CAR T cells in a patient with relapsing idiopathic inflammatory myositis after initial and repeat therapy with CD19 CAR T cellsMüller et al., Nature Medicine (2025) · case report
Teclistamab-Induced Remission in Refractory Systemic Lupus ErythematosusAlexander et al., New England Journal of Medicine (2024) · case report
Daratumumab in systemic lupus erythematosus: a single-arm phase 2 trialOstendorf et al., Nature Communications (2026) · nonrandomized interventional

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Makes it worseStrong suspect

An antiviral alarm that never switches off

Researchers call this: Type I interferon signalling

The body runs a permanent low-grade antiviral response with no virus present, which recruits more immune cells and worsens tissue damage.

How much of lupus this could explain: A raised interferon signature is present in a majority of people with lupus, though it varies over time within the same person.

Why researchers take it seriously

Blocking it works: a randomized phase 3 trial met its primary endpoint, and the approach has since been confirmed in a second route of administration.
The signature is measurable, reproducible across laboratories, and tracks with disease activity.

Why it might be wrong

In stored samples taken before diagnosis, interferon rises late — after autoantibodies have already appeared. That is the wrong order for a starting cause.
When B cells are removed, the interferon signature collapses on its own, suggesting it is downstream rather than independent.
An earlier interferon-blocking drug failed, and patients with low interferon actually responded better — the opposite of what a clean selector would predict.
Blocking it barely changes the underlying antibodies. It controls disease; it does not reset it.

What would settle it

Show that disease stays corrected after interferon blockade is withdrawn in a pre-specified subgroup. Current evidence shows only control while treatment continues.

Tried against it: switching off the false antiviral alarm

Sources (6)

Trial of Anifrolumab in Active Systemic Lupus ErythematosusMorand et al., New England Journal of Medicine (2020) · randomized controlled trial
Type I interferon inhibitor anifrolumab in active systemic lupus erythematosus (TULIP-1): a randomised, controlled, phase 3 trialFurie et al., Lancet Rheumatology (2019) · randomized controlled trial
Efficacy and Safety of Subcutaneous Anifrolumab in Systemic Lupus Erythematosus: A Randomized, Phase 3 StudyManzi et al., Arthritis & Rheumatology (2026) · randomized controlled trial
Altered type II interferon precedes autoantibody accrual and elevated type I interferon activity prior to systemic lupus erythematosus classificationMunroe et al., Annals of the Rheumatic Diseases (2016) · case control
Selective CAR T cell-mediated B cell depletion suppresses IFN signature in SLEWilhelm et al., JCI Insight (2024) · human mechanistic
Interferon-inducible gene expression signature in peripheral blood cells of patients with severe lupusBaechler et al., PNAS (2003) · human mechanistic

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Could start itStrong suspect

A common virus that rewires the wrong cells

Researchers call this: Epstein–Barr virus (EBV)

Almost everyone carries EBV for life. In lupus it appears to settle preferentially in exactly the self-reactive B cells that cause trouble, and change how they behave.

How much of lupus this could explain: Unknown. EBV infects roughly 95% of adults, so any role has to explain why so few develop lupus.

Why researchers take it seriously

The earliest autoantibodies detectable years before diagnosis often cross-react with an EBV protein — the right chronology for an initiating event.
EBV preferentially infects self-reactive B cells and reprograms them into cells that actively drive other immune cells.

Why it might be wrong

A national registry study found no increased lupus risk after symptomatic EBV infection — a genuinely negative result.
The decisive study that established EBV's role in multiple sclerosis has no lupus equivalent. That dataset simply does not exist yet.
EBV is obviously not sufficient on its own, given how common it is.
Attempts to treat autoimmune disease by targeting EBV have failed or been inert.

What would settle it

Analyse stored military serum samples for whether EBV infection precedes lupus with a large virus-specific effect — the same design that settled the question in multiple sclerosis. The samples already exist; this is the cheapest decisive study available.

Sources (4)

Epstein-Barr virus reprograms autoreactive B cells as antigen-presenting cells in systemic lupus erythematosusYounis et al., Science Translational Medicine (2025) · human mechanistic
Development of autoantibodies before the clinical onset of systemic lupus erythematosusArbuckle et al., New England Journal of Medicine (2003) · case control
Epstein-Barr virus-associated infectious mononucleosis and risk of systemic lupus erythematosusUlff-Møller et al., Rheumatology (2010) · retrospective cohort
Systematic review and meta-analysis of the sero-epidemiological association between Epstein-Barr virus and systemic lupus erythematosusHanlon et al., Arthritis Research & Therapy (2014) · systematic review

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

5 candidates

Possible contributor

A real association in people, or proof in animals, but no human experiment has yet tested whether removing it helps.

Makes it worsePossible contributor

Dying immune cells spill inflammatory material

Researchers call this: Neutrophil extracellular traps and oxidised mitochondrial DNA

Certain white blood cells fling out webs of DNA when they die. The oxidised mitochondrial DNA in those webs is unusually good at triggering the antiviral alarm.

How much of lupus this could explain: Unclear; proposed as a general amplifier rather than a subgroup feature.

Why researchers take it seriously

Blocking the process reduced disease in mouse models.
It supplies a plausible physical source for the self-DNA that the immune system reacts to.

Why it might be wrong

Human evidence is cross-sectional — it cannot separate cause from a consequence of being ill.
A closely related DNA-sensing pathway shows the opposite effect: deleting it makes several lupus models worse.

What would settle it

A human trial that measurably reduces these traps and then shows a clinical difference — not just a change in the laboratory readings.

Tried against it: replacing the missing cleanup enzyme

Sources (1)

Neutrophil extracellular traps enriched in oxidized mitochondrial DNA are interferogenic and contribute to lupus-like diseaseLood et al., Nature Medicine (2016) · animal model

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Makes it worsePossible contributor

Gut bacteria that get somewhere they should not be

Researchers call this: Pathobiont translocation and gut dysbiosis

Particular gut species cross the intestinal wall and reach liver and lymph tissue, where they appear to provoke autoimmunity; another species blooms during kidney flares.

How much of lupus this could explain: Unclear. Blooms have been observed around flares in some people with kidney involvement.

Why researchers take it seriously

In mice, both antibiotics and a targeted vaccine against the organism reduced disease — an unusually strong preclinical result.
The organism has been detected in human liver tissue, and matching antibodies are found in a subset of patients.
Blooms of a second species track with kidney flares within the same person over time.

Why it might be wrong

The causal evidence is entirely in mice. No human eradication experiment has ever been run.
Cross-cohort replication of the human associations has been inconsistent.
Being ill and taking immunosuppressants both change the gut, so the direction of causation is unresolved.

What would settle it

A pre-registered prospective cohort sampling monthly through flares, which would either turn the association into a usable prediction or kill it.

Sources (4)

Translocation of a gut pathobiont drives autoimmunity in mice and humansManfredo Vieira et al., Science (2018) · animal model
Lupus nephritis is linked to disease-activity associated expansions and immunity to a gut commensalAzzouz et al., Annals of the Rheumatic Diseases (2019) · case control
Association between systemic lupus erythematosus and disruption of gut microbiota: a meta-analysisXiang et al., Lupus Science & Medicine (2022) · systematic review
Gut microbiota promote the inflammatory response in the pathogenesis of systemic lupus erythematosusMa et al., Molecular Medicine (2019) · animal model

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Makes it worsePossible contributor

Two X chromosomes mean a double dose of the alarm gene

Researchers call this: TLR7 escape from X-chromosome inactivation

The gene for the RNA sensor sits on the X chromosome, and in some immune cells it stays active on both copies — a plausible reason lupus affects women roughly nine times as often.

How much of lupus this could explain: Proposed to explain the sex distribution of lupus generally, not to identify a treatable subgroup.

Why researchers take it seriously

Directly observed in human immune cells rather than inferred.
Men with an extra X chromosome have markedly higher lupus risk, which fits a dosage explanation.
It connects the sex bias to the same sensor implicated by the rare genetic causes.

Why it might be wrong

It has not been shown that this escape starts disease rather than accompanying it.
Later work raises the possibility that the escape is partly caused by the disease.
No treatment follows from it — you cannot act on someone's chromosome count.

What would settle it

Longitudinal measurement showing the escape is present before disease onset rather than appearing alongside it.

Tried against it: turning the overactive alarm back down

Sources (1)

TLR7 escapes X chromosome inactivation in immune cellsSouyris et al., Science Immunology (2018) · human mechanistic

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Blocks recoveryPossible contributor

The immune system's brakes are worn

Researchers call this: Regulatory T-cell insufficiency

Regulatory T cells normally restrain immune responses. In lupus there appear to be too few working ones, so responses that should be damped keep running.

How much of lupus this could explain: Measurable in many people with active disease; the size of the deficit varies.

Why researchers take it seriously

A phase IIb randomized trial found that restoring these cells with low-dose interleukin-2 improved disease.
The effect was dose-dependent, which is what a real biological relationship looks like.

Why it might be wrong

An earlier randomized trial of the same approach missed its primary endpoint.
The effect fades when dosing stops — it restores balance rather than teaching tolerance.
It is not established whether the shortfall precedes disease or results from it.

What would settle it

Show that a time-limited course produces lasting benefit after it is stopped. Everything so far reverses on withdrawal.

Tried against it: restoring the immune system's brakes

Sources (1)

Low dose IL-2 therapy restores regulatory T cells in patients with systemic lupus erythematosus in a dose-dependent manner: a phase IIb trialNature Communications (2026) · randomized controlled trial

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Keeps it goingPossible contributor

Immune cells running their engines too hot

Researchers call this: mTOR activation and T-cell metabolic dysfunction

T cells in lupus show an altered metabolism that keeps them in an activated state, and drugs that dial that metabolism down appear to help.

How much of lupus this could explain: Unclear; no validated way to identify who has this pattern.

Why researchers take it seriously

A placebo-controlled trial of an mTOR inhibitor reported a substantial improvement over placebo.
An earlier randomized trial of a different metabolic agent was also positive.

Why it might be wrong

The strongest recent result has been presented at a conference but not yet published in full — it cannot be treated as established.
The earlier positive trial was never independently replicated.
Nothing shows this metabolic state exists before disease starts rather than resulting from chronic activation.

What would settle it

Peer-reviewed publication of the full trial result, followed by independent replication in a different population.

Tried against it: cooling down overactive immune cell metabolism

No published source yet

The strongest evidence for this candidate has been presented at a conference but not published in full. This site does not cite unpublished results as sources, so the record stays here without one until that changes.

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

1 candidate

Losing support

Once widely believed, now undercut by stronger study designs. Kept visible so the record is honest.

Makes it worseLosing support

Low vitamin D

Researchers call this: Vitamin D insufficiency

People with lupus often have low vitamin D, which led to a widespread belief that the deficiency helps cause the disease. Genetic evidence does not support that.

How much of lupus this could explain: Low levels are common — but so is being indoors, unwell, and avoiding sun on medical advice.

Why researchers take it seriously

A large randomized prevention trial found supplementation lowered a combined autoimmune-disease outcome.
Vitamin D has real effects on immune cells in laboratory work.

Why it might be wrong

Genetic studies that are resistant to confounding find no effect of vitamin D level on lupus risk.
The relationship is better explained in reverse: being ill with lupus leads to sun avoidance and lower levels.
The positive prevention trial used a combined outcome across many diseases and included very few lupus cases.

What would settle it

A prevention trial powered for lupus specifically. The existing evidence cannot answer a lupus-specific question.

Sources (2)

Vitamin D level and risk of systemic lupus erythematosus and rheumatoid arthritis: a Mendelian randomizationBae & Lee, Clinical Rheumatology (2018) · case control
Vitamin D and marine omega 3 fatty acid supplementation and incident autoimmune disease: VITAL randomized controlled trialHahn et al., BMJ (2022) · randomized controlled trial

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

1 candidate

Ruled out

Tested directly in randomized trials and not supported. Useful mainly for correcting a common belief.

Makes it worseRuled out

Estrogen and the contraceptive pill trigger flares

Researchers call this: Exogenous estrogen as a flare driver

Because lupus overwhelmingly affects women, estrogen was long assumed to drive flares — and people were routinely advised against combined contraceptives. Randomized trials did not bear this out.

How much of lupus this could explain: Relevant to a great many women with lupus who were given restrictive advice on this basis.

Why researchers take it seriously

The striking female predominance of lupus made a hormonal explanation reasonable to test.

Why it might be wrong

Two randomized trials published on the same day both found no increase in severe flares among women with stable lupus.
The sex bias is better explained by X-chromosome gene dosage and complement genetics than by hormone levels.

What would settle it

Nothing further is needed for stable disease. The remaining open questions concern active disease and women with antiphospholipid antibodies, who were excluded from those trials.

Sources (1)

Combined oral contraceptives in women with systemic lupus erythematosusPetri et al., New England Journal of Medicine (2005) — SELENA · randomized controlled trial

Last reviewed 2026-08-05 · agent-assisted draft, not domain-reviewed

Before you take this anywhere

None of this tells you why you have lupus.

These are population-level research candidates. No test on this page can be applied to an individual, and nothing here should change anyone's treatment. The rare proven causes are proven precisely because they are rare — they were found in families, not in clinics.

What's been tried against these How we grade evidence