CAR-T therapy works by taking a patient’s own T cells, equipping them with a synthetic receptor that locks onto a protein on the target cell, and returning them to the body as a living drug. It is approved for several blood cancers. In 2026 it is being pushed in two directions at once: into autoimmune disease, where the target is the immune system’s own B cells, and into solid tumors, where it has historically failed.

Both pushes have news from the past few months. A Chinese regulator approved the first CAR-T product for a solid tumor in June, and a U.S. company is close to filing for an autoimmune indication. Here is the mechanism, the evidence and the open questions, in question-and-answer form.

What is CAR-T, mechanically?

T cells normally recognize threats through their native receptors. A chimeric antigen receptor, or CAR, is an engineered protein that fuses an antibody-like recognition piece to the internal signaling machinery of a T cell. Once the gene for it is added, the cell attacks anything displaying the matching protein, with no need for the usual presentation steps.

  1. Collection. T cells are removed from the patient’s blood.
  2. Engineering. A viral vector inserts the CAR gene.
  3. Expansion. The cells multiply in a manufacturing facility.
  4. Conditioning. The patient typically gets chemotherapy to reduce existing lymphocytes.
  5. Infusion and monitoring. The modified cells go back in, and the team watches for toxicity.

The two best-known toxicities are cytokine release syndrome, a systemic inflammatory reaction, and neurological effects. Both remain boxed warnings. In June 2025 the FDA removed the formal risk-management program for six approved CD19- and BCMA-directed products, a change described on the FDA’s own page. Labels still call for monitoring for at least two weeks and staying near a treatment center, and manufacturers must follow patients for 15 years for secondary cancers and other long-term effects.

Which products exist today?

The six products named in the FDA’s 2025 action are Abecma, Breyanzi, Carvykti, Kymriah, Tecartus and Yescarta. All are autologous, meaning each dose is made from the patient’s own cells, and all target either CD19 or BCMA, two proteins found on B-lineage cells and plasma cells. They are approved for blood cancers. Everything in the autoimmune section below is investigational. No CAR-T product is approved for an autoimmune disease in the U.S. as far as I could confirm, and the Kyverna filing, if accepted, would be the first test of that.

Why use it for autoimmune disease?

Some autoimmune diseases are driven by B cells that make harmful antibodies. Drugs can suppress them, but they often rebound. A CD19-directed CAR-T product can remove B cells deeply, including those hiding in tissue, and the hope is that the immune system then rebuilds itself without the faulty clones. That is a hypothesis supported by early data, not a settled mechanism.

Lupus

The UK CARLYSLE study uses obecabtagene autoleucel, a CD19 CAR-T product, in adults with severe lupus. An abstract from ACR Convergence 2025 described six treated patients, all with lupus nephritis and a median of five failed prior medicines. No patient had cytokine release syndrome above grade 1, and three of six reached a complete renal response, with up to eight months of follow-up. A summary on the Lupus UK site from July 2026 describes nine patients and reports that five of six on the lower dose had few or no symptoms. One participant had liver injury that fully resolved. A larger trial, LUMINA, is planned in lupus nephritis, according to that summary. UCL announced the early results in June 2026. I could not find the full journal publication, so treat these as early, small-sample reports.

Neurological autoimmunity

Kyverna Therapeutics is furthest along. On September 24, 2026 the company reported one-year topline results for its CD19 product miv-cel in two conditions. In stiff person syndrome, a single-arm registrational phase 2 trial of 26 adults showed a median improvement of 49% on a timed 25-foot walk at 12 months, with no high-grade cytokine release syndrome or neurotoxicity reported. In generalized myasthenia gravis the study was seven patients. The company says it expects to finish a rolling license application to the FDA for stiff person syndrome in the fourth quarter of 2026 (company press release via Finviz). That is a company release, and the company notes results may change after further analysis.

The same company issued a separate statement on September 1, 2026, responding to unnamed “recent third-party reports” about safety. It said more than 100 patients had been treated and that it had not seen high-grade cytokine release syndrome (BiopharmaWatch). I could not identify what those reports were, so I do not draw a conclusion from them.

What is different about solid tumors?

CAR-T cured some blood cancers because the targets, such as CD19, sit on every cell of the cancer, and the cells circulate where infused T cells can reach them. Solid tumors undo each of those advantages. A 2022 review in Frontiers in Immunology lists the recurring obstacles:

Barrier What it means in practice
Antigen variability Cells in one tumor express the target unevenly, and the target can vanish under pressure.
Poor infiltration Dense tissue and abnormal blood vessels keep T cells out.
Hostile microenvironment Low oxygen, acidity and suppressive cells blunt T-cell activity.
Exhaustion Constant antigen exposure wears the cells down.
Shared targets If healthy tissue carries the antigen too, it can be attacked.

What changed in June 2026?

China’s NMPA approved satricabtagene autoleucel, known as satri-cel, on June 22, 2026, for adults with CLDN18.2-positive, HER2-negative advanced gastric or gastroesophageal junction cancer that had progressed after at least two lines of therapy. OncLive describes it as the first CAR-T approved for a solid tumor.

The supporting evidence is a randomized, open-label phase 2 trial (NCT04581473) with 104 patients on satri-cel and 52 on physician’s-choice treatment. Median progression-free survival was 3.25 months versus 1.77 months, a hazard ratio of 0.37. The gain is real in a statistical sense, but the absolute difference is under two months, and OncLive’s summary did not give overall survival. The control arm mixed several regimens, and follow-up was far longer in the CAR-T arm. Safety was heavy: grade 3 or higher adverse events in 99% of evaluable satri-cel patients, mostly drops in blood counts, and cytokine release syndrome in 95%.

So the approval proves a principle without yet showing a transformation. It shows that a CAR-T can beat standard care on a delay-of-progression endpoint in a hard-to-treat solid tumor. It does not show cures.

Why the risk calculation differs between cancer and autoimmunity

This section is interpretation, not trial data. A person with relapsed lymphoma may accept a high chance of serious side effects because the alternative is progression. A person with lupus or myasthenia gravis often has other options, including long-term drugs with their own costs. Regulators and physicians will likely ask for a stronger safety margin before CAR-T moves into those settings. That may be why companies emphasize the absence of severe cytokine release syndrome in autoimmune cohorts, and why the long-term follow-up the FDA requires, 15 years for the approved products, will matter for any new indication. It is also why so much effort is going into gentler conditioning and, in some programs, into making the cells inside the body rather than outside it.

A brief timeline of 2025 and 2026

  • June 26, 2025: The FDA removes the risk-management program for six approved CAR-T products while keeping boxed warnings.
  • Late 2025: CARLYSLE lupus data appear as a poster at ACR Convergence.
  • June 2026: UCL publicizes early lupus results, and China approves satri-cel on June 22.
  • September 2026: Kyverna reports one-year data in stiff person syndrome and myasthenia gravis.
  • Fourth quarter of 2026: Kyverna expects to complete a rolling FDA license submission.

What stands in the way of access?

Whatever the disease, each dose is manufactured from the individual patient, and the process takes time and money. Those limits are driving two responses: off-the-shelf cells made from donors, and in vivo approaches that skip the factory altogether. The second approach, in vivo CAR-T, is the subject of a companion post on this site.

How to read the next headline

  • Check the size of the study. Seven or nine patients can inspire a larger trial but cannot establish a safety profile.
  • Check the source. Company releases are legitimate, but they are not peer review.
  • Look at the endpoint. Progression-free survival and overall survival answer different questions.
  • Look at follow-up. Autoimmune disease is lifelong, so a year is a start.

This is general information about research, not medical advice.

Sources