CAR-T Is Leaving Blood Cancer Behind
All six FDA-approved CAR-T therapies treat blood cancer. But in the pipeline, solid tumors and autoimmune disease now outnumber it more than three to one.
Every one of the six CAR-T cell therapies the FDA has approved treats a blood cancer. Leukemia, lymphoma, myeloma: the diseases the technology was built for. But that is not where the field is working anymore. In my archive’s engineered-T-cell cluster, signals about solid tumors (109) and autoimmune disease (16) now outnumber blood cancer (34) by more than three to one. The most powerful tool oncology has produced is being re-aimed at the two problems it was never designed to solve.
CAR-T earned its reputation on the diseases where it works almost miraculously: engineer a patient’s own T-cells to recognize a marker on their cancer, infuse them back, and watch treatment-resistant leukemias melt away. Six approvals later, that story is settled and, in a sense, finished. The interesting question is no longer whether CAR-T cures blood cancer (it does), but where a field goes once its founding problem is solved. My data has an answer, and it is not where the approvals are.
I watch this through convergence: a real shift shows up across independent layers that don’t coordinate, a trial registry, a grant office, a patent examiner, a journal, not in one loud place. Engineered T-cell therapy is lit across all of them right now. What the composition of that activity shows is a field that has quietly changed its target.
The frontier is aimed off blood
Start with what the work is actually about. Sorting the cluster’s signals by disease, the ratio is not close.
Figure 1: The activity has moved to the hard targets
Engineered T-cell signals in the archive by disease area. Solid tumors (109) and autoimmune disease (16) together outnumber blood cancer (34) by more than three to one. CAR-T was built and approved for blood cancers; the activity has moved to the two problems it was never designed for.
Solid tumors dominate. This matters because solid tumors are the graveyard of cell therapy: a hostile tumor microenvironment, no clean tumor-specific target, T-cells that can’t traffic in or persist once they do. CAR-T was easy in blood because leukemia cells float freely and wear a clean marker (CD19). A pancreatic or brain tumor offers neither. That the majority of the field’s effort is now pointed at exactly this wall tells you the founders’ problem is considered done and the frontier has moved to the one that isn’t.
The approvals confirm the lag by their absence. All six approved CAR-T products are hematologic (FDA, via OncLive); not one is approved for a solid tumor. The only engineered-T-cell therapy cleared for a solid tumor to date is Tecelra for synovial sarcoma (2024), and it is a TCR therapy, not CAR-T, a different mechanism. So the clinic still lives entirely in blood while the pipeline has already left it. That gap is the story.
The pipeline is enormous, and early
How committed is the field to the harder targets? Look at the trials.
Figure 2: 270 human trials, almost all still Phase 1
The 270 registered T-cell trials by clinical phase. Roughly 255 are Phase 1 or early-stage (safety and feasibility), 12 are Phase 2, and effectively none has reached Phase 3. This is a field re-starting at the bottom of the ladder on harder targets, not one wrapping up.
My archive is tracking 270 registered human trials in this cluster, and roughly 241 of them are Phase 1, safety and feasibility, against a dozen in Phase 2 and none in Phase 3. This is not a field wrapping up; it is a field re-starting from the bottom of the ladder on a new problem. The solid-tumor trials read exactly like early-stage attempts on hard targets: a Phase 1 in Ewing sarcoma and DSRCT, a Phase 1 in advanced solid tumours. Each is a small, careful shot at the microenvironment wall, and most will fail. That is what re-aiming a technology looks like: a wave of Phase 1s, years before anything reaches the clinic.
The surprise is autoimmune
The solid-tumor pivot is expected. The second one is not, and it is the more radical.
The same CD19 CAR-T built to kill B-cell cancers turns out to reset the immune system in autoimmune disease, because the B-cells driving lupus or myositis carry the same CD19 marker. In a 2024 NEJM case series from the Erlangen group, fifteen patients with severe lupus, myositis, or systemic sclerosis received a single CD19 CAR-T infusion. Every lupus patient reached drug-free remission; every myositis and sclerosis patient responded; immunosuppression was stopped entirely in all fifteen, holding at a median of fifteen months. A one-time infusion producing durable, medication-free remission in diseases usually managed for life is not an incremental result.
Get Andrii Kuzovchykov’s stories in your inbox
Join Medium for free to get updates from this writer.
The money followed. Cabaletta Bio now holds RMAT designation for its autoimmune CAR-T in myositis and lupus and is moving toward registrational trials, with Kyverna close behind. And the direction that could make this scalable arrived when AbbVie bought Capstan Therapeutics for $2.1 billion in 2025 for an in-vivo CAR-T, CAR made inside the body via a lipid nanoparticle, skipping the apheresis and chemotherapy that make today’s CAR-T a hospital ordeal, aimed first at autoimmune disease. Two years ago “CAR-T for lupus” was a curiosity. In my archive it is now its own live category.
A fully-lit funnel, not a hype cycle
The reason I trust this signal is where it comes from.
Figure 3: Every layer of the pipeline is lit
The cluster spans 270 trials, 192 PubMed papers, 39 OpenAlex papers, 35 NIH grants and 23 patents across 13 independent sources. The full research-to-clinic-to-IP stack is lit, including the public research budget — the signature of a real shift, not a press-release cycle.
The cluster is 270 trials, 192 PubMed papers, 35 NIH grants, and 23 patents, across thirteen independent sources, the full research-to-clinic-to-IP stack, including the public research budget and the companies protecting the engineering. You can see the whole thing, with links to every underlying signal, on the trend’s public evidence page. This is the expensive, slow corner of the field, grant offices and trial registrars and patent examiners, not a wave of press releases. When the costly layers move together, the shift is real rather than a cycle of attention.
Two normal explanations and one harder one
Blood-cancer CAR-T is mature, so the marginal paper naturally goes elsewhere. Weight ~30%. Once a problem is solved, research moves to unsolved ones; the pivot could be simple gradient-following. True, and it doesn’t weaken the read, where the gradient points is the signal.
The autoimmune and solid-tumor results are early and will mostly disappoint. Weight ~25%. Fifteen patients and a wall of Phase 1s are not approvals; solid tumors have broken better ideas than this. Correct, and I say so below. But the breadth of independent commitment, grants, IP, registrational-track trials, a $2.1B acquisition, is more than early hype usually attracts.
The field is repurposing its best tool from a niche it conquered into two mass diseases. Weight ~45%. This is the read I hold. Blood cancers are rare; solid tumors and autoimmune disease are enormous. A technology that reached drug-free remission in lupus and is being aimed at the microenvironment wall is trying to move from a few thousand patients to millions. If it half-works, cell therapy stops being an oncology sub-specialty and becomes a platform. That is a different scale of thing than the approvals suggest.
What I am specifically not claiming
This is not a stock call, and nothing here names a ticker. The pivot is in the pipeline, not the clinic: all six approvals remain blood cancers, and the solid-tumor and autoimmune work is overwhelmingly Phase 1, early, small, and failure-prone. Fifteen patients is a signal, not a verdict. Tecelra is a TCR therapy, not CAR-T, and I have kept that distinction. My counts are collection, not a census, they reflect what I gather and how densely, which is why the load-bearing claim is the ratio (activity aimed off blood), not any single number. Market sizes I have seen for cell therapy range from roughly $6B in 2025 to wildly different 2030 projections; I treat them as a range, not a fact. And CAR-T carries a real risk: the FDA required a class-wide boxed warning in 2024 over secondary T-cell cancers. Re-aiming a powerful, risky therapy at healthier autoimmune patients raises the bar, not lowers it.
What to watch
- The first real solid-tumor efficacy signal. Not a Phase 1 safety readout, a durable response in a common solid tumor. That is the milestone that would move this from “trying” to “working.”
- Cabaletta and Kyverna registrational autoimmune data (2026). Registrational trials turning the fifteen-patient result into a controlled one is the line between a case series and a category.
- In-vivo CAR-T clinical results. If CAR-T made inside the body works, it removes the manufacturing and hospitalization that cap who can ever receive it, the difference between thousands and millions of patients.
- A boxed-warning update. More secondary-malignancy data would reset the risk-benefit for the healthier autoimmune population fastest.
Reproducibility note
The cluster figures come from the live signal store on 1 August 2026, trend #118: 586 signals across 13 sources, 270 CLINICAL_TRIALS, 192 PUBMED, 35 NIH_REPORTER grants, 23 EPO_PATENTS. The disease split (109 solid-tumor / 34 blood / 16 autoimmune signals) and the phase split (≈241 Phase 1, 12 Phase 2, 0 Phase 3 of the 270 trials) are keyword scans over the cluster's signal titles; every trial, paper, and patent links to its clinicaltrials.gov, PubMed, or Espacenet record, browsable at signals.sovenyr.com/t/118. Archive counts are collection, not a census of world research, so the load-bearing claim is the composition, activity aimed off blood, which is robust to collection density. External facts (six FDA CAR-T approvals all hematologic; the NEJM autoimmune series; Cabaletta's RMAT; the AbbVie–Capstan acquisition; the FDA boxed warning; Tecelra as a TCR therapy) are from published third-party sources, cited inline and dated. Where I say "the field is repurposing its best tool," that reading is mine.
Numbers are from the archive as dated. Reading judgments (what the composition means) are flagged as such., Andrii
Originally published at https://sovenyr.substack.com on August 1, 2026.
