How to find CAR-T targets and trials for any gene
Look up GPC2 in almost any drug database and it comes back empty — no approved drug, no clinical candidate. Yet there are two active CAR-T trials against it right now, in relapsed neuroblastoma and medulloblastoma. Look up GD2, the most established CAR-T target in neuroblastoma, and you will not find it at all, because GD2 does not have a gene. This is a structural blind spot in how drug data is organised, and it hides an entire treatment modality.
Why gene databases miss CAR-T targets
Resources like the Open Targets Platform are keyed on drug–target records: a molecule, the protein it binds, the diseases it is tested in. That model works well for small molecules and antibodies. It breaks for cell therapy in two specific ways.
1. The antigen has no gene
GD2 is a disialoganglioside — a glycolipid on the cell surface, not a protein. It is built by enzymes (B4GALNT1, ST8SIA1), but GD2 itself is not a gene product you can key a database on. The same applies to other glycan and glycolipid antigens such as Lewis-Y. A gene-indexed database has nothing to attach these trials to, so they simply do not appear.
2. The target has no conventional drug
Many cell-therapy antigens were never pursued as small-molecule or antibody targets, so they carry no drug records at all. GPC2 and L1CAM look completely undrugged in a gene–drug view, while both have CAR-T programmes in the clinic. The absence of a drug record is read as "nothing here", when the real answer is "nothing of that kind here".
The fix: search the trials, not the gene index
Clinical trials describe what they are testing in plain text: "Anti-GD2 CAR T Cells in Pediatric Patients", "B7H3 CAR T Cell Immunotherapy for Recurrent/Refractory Solid Tumors", "GPC2 CAR T Cells for Relapsed or Refractory Neuroblastoma". That text is searchable, and it is where the antigen actually lives.
BioTransfer's Drug Target Explorer now queries ClinicalTrials.gov alongside Open Targets. Enter a gene and you get, in one view: every conventional drug acting on it with its indications and clinical stage, and every CAR-T or engineered cell-therapy trial naming it as the target.
Genes are not what trials call the antigen
One detail decides whether this works. Trials write the antigen name, not the gene symbol. Search CD276 literally and you miss nearly every trial, because they all say B7-H3. The tool carries an alias map:
| Gene symbol | What trials call it | Typical setting |
|---|---|---|
| TNFRSF17 | BCMA | Multiple myeloma |
| CD276 | B7-H3 | Paediatric & adult solid tumours |
| B4GALNT1 | GD2 (glycolipid) | Neuroblastoma, sarcoma, glioma |
| L1CAM | CD171 | Neuroblastoma |
| MS4A1 | CD20 | B-cell lymphoma |
| FOLH1 | PSMA | Prostate cancer |
| MSLN | Mesothelin | Mesothelioma, pancreatic, ovarian |
| CLDN18 | Claudin 18.2 | Gastric, pancreatic |
| IL3RA | CD123 | AML, BPDCN |
What the CAR-T landscape actually looks like
Across roughly 2,500 CAR trials registered on ClinicalTrials.gov, the distribution is heavily concentrated. Haematological targets dominate by volume; solid-tumour programmes are numerous but almost entirely early phase.
| Target | Scale | Where |
|---|---|---|
| CD19 | The largest programme by a wide margin | B-ALL, DLBCL, CLL, follicular lymphoma |
| BCMA (TNFRSF17) | Second largest; several approved products | Multiple myeloma, AL amyloidosis |
| CD22, CD20, CD30, CD123, CD7 | Established, mostly Phase 1–2 | Haematological malignancies |
| GD2 | Leading solid-tumour antigen; ~30 trials, most active | Neuroblastoma, osteosarcoma, glioma |
| B7-H3 (CD276) | Fast-growing; ~30–60 trials | Paediatric solid tumours, glioblastoma |
| Mesothelin, GPC3, claudin 18.2, HER2 | Active early phase | Pancreatic, HCC, gastric, sarcoma |
| GPC2, L1CAM | Emerging, single-digit trials | Neuroblastoma, medulloblastoma |
What makes something a viable CAR-T target
Two hard requirements, and they eliminate most genes immediately.
- It must be on the cell surface. A chimeric antigen receptor binds antigen directly, without MHC presentation. This is why MYCN — the defining driver of high-risk neuroblastoma — returns zero CAR-T trials and always will: it is an intracellular transcription factor. The same holds for TP53 and most oncogenic drivers.
- Tumour expression must exceed essential normal tissue. CAR-T cells cannot tell tumour from healthy cells carrying the same antigen. Expression on heart, lung, nerve or gut is what produces on-target off-tumour toxicity, historically the field's most serious safety failures. This is why a promising expression difference in tumour data is only half an argument — the normal-tissue side decides it.
How to use it
- One gene: enter the symbol — drugs and CAR-T trials load together.
- A gene set: paste up to 25 symbols. Drug data loads for all of them; CAR-T trials load per gene on demand, so a long list stays fast.
- Non-protein antigens: for GD2, search the synthase B4GALNT1 — the alias map routes it to GD2 trials.
- Every row links back to its NCT record on ClinicalTrials.gov, so you can read eligibility, sites and sponsors at the source.
Search any gene, gene set, disease or pathway for its drugs, indications, clinical stage and CAR-T trials.
Open the tool →Frequently asked questions
How do I find CAR-T trials for a specific gene?
Enter the gene symbol in the Drug Target Explorer. It searches ClinicalTrials.gov for CAR-T and engineered cell-therapy trials naming that gene's antigen and lists each with its phase, recruitment status, condition and a direct NCT link. Because trials name the antigen rather than the gene, the tool maps symbols to antigen names automatically.
Why does GD2 not appear in gene-based drug databases?
GD2 is a disialoganglioside — a glycolipid, not a protein. It is synthesised by enzymes such as B4GALNT1 and ST8SIA1, but has no gene of its own, so a database keyed on gene-to-drug records has nothing to attach GD2 trials to. Searching trial text by antigen name recovers them.
Which CAR-T targets are used in neuroblastoma?
GD2 is the most established and has the largest number of trials. B7-H3 (CD276) is being tested across paediatric solid tumours. GPC2 and L1CAM (CD171) are newer neuroblastoma-directed targets in early-phase trials. MYCN cannot be a CAR-T target despite driving high-risk disease, because it is intracellular.
Can any gene be a CAR-T target?
No. The antigen must be expressed on the cell surface, since a CAR binds it directly without MHC presentation, and it must show a favourable expression difference between tumour and essential normal tissue to avoid on-target off-tumour toxicity.
Is the trial data free to use?
Yes. It comes from ClinicalTrials.gov (US NIH / National Library of Medicine). As a US Government work it is public domain, served through a documented public API with no key or registration. BioTransfer passes metadata through and links every row back to its NCT record.
Related tools
- Drug Target Explorer — drugs, indications, clinical stage and CAR-T trials for any gene, disease or pathway.
- GEO Dataset Finder — find expression datasets to check where your candidate antigen is expressed.
- qPCR Primer Designer — validate that expression at the bench.