In vivo CAR-T therapy: what 15 months of follow-up revealed

In vivo CAR-T therapy attempts to remove one of the most cumbersome parts of cellular immunotherapy. Conventional CAR-T treatment requires collecting a patient's T cells, genetically engineering and expanding them in a manufacturing facility, and returning them after lymphodepleting chemotherapy. ESO-T01 instead delivers the CAR gene to T cells inside the patient's body through a single intravenous infusion.

The first short-term results from four people with relapsed or refractory multiple myeloma had already shown that this could work. All four developed detectable CAR-T cells and all four responded: two achieved stringent complete responses and two partial responses.

The new Nature Medicine paper follows the same patients for as long as 15 months. That longer view changes the emphasis of the story. The technology succeeded in generating functional CAR-T cells inside people, but the initial responses were durable in only one of the four patients.

CAR-T without collecting and manufacturing the cells

Multiple myeloma is a malignancy of plasma cells. Several established CAR-T products target BCMA, a protein strongly expressed by myeloma cells.

Standard CAR-T manufacturing starts with leukapheresis. T cells are removed from the patient, sent to a specialized facility, genetically modified to express a chimeric antigen receptor, expanded, tested and eventually reinfused. Patients generally receive lymphodepleting chemotherapy first.

ESO-T01 is designed to move that engineering step into the body. It is an engineered, self-inactivating lentiviral vector carrying an anti-BCMA CAR construct and designed to preferentially transduce T cells. Once those cells receive the genetic instructions, they can become CAR-T cells without first leaving the patient.

The four participants received a single ESO-T01 infusion without leukapheresis, ex vivo CAR-T manufacturing or lymphodepleting chemotherapy.

These were particularly difficult cases. The median age was 63, patients had received a median of four previous treatment lines, and all had extramedullary disease and bulky plasmacytomas larger than 5 cm.

Against that background, the initial response in all four patients provided an unusually direct proof of concept.

Four responses, but only one durable remission

The extended follow-up revealed the central problem.

One patient remained in stringent complete remission through 15 months. The other three relapsed or progressed at 5, 3 and 3 months after infusion.

Median progression-free survival was 4.0 months.

In two patients, relapse or progression coincided with CAR-T cells becoming undetectable in blood, pointing to limited cellular persistence as an important weakness of the current approach.

A fourth case showed that persistence alone is not necessarily enough. CAR-T cells remained detectable in that patient's cerebrospinal fluid, yet myeloma cells continued to increase there and the disease progressed.

The authors discuss several possible contributors, including high tumor burden, poor underlying T-cell fitness, an inhibitory tumor microenvironment and reduced antigen expression. With only four cases, those explanations remain hypotheses rather than established mechanisms.

All three patients whose disease progressed received subsequent salvage treatment and eventually died during follow-up, from infection, progressive disease and intracranial hemorrhage, respectively. The fourth patient was still in sustained complete remission at the 15-month time point.

The cost of engineering T cells inside the body

The treatment also produced substantial toxicity.

All four patients developed cytokine release syndrome, ranging from grade 1 to grade 3, as well as grade 3 or higher hematologic toxicities. One patient developed grade 1 immune effector cell-associated neurotoxicity syndrome.

Most hematologic toxicities later returned to baseline or improved. During extended follow-up, the investigators reported no important new ESO-T01-related adverse events beyond those described initially.

The team also examined risks that are particularly relevant to a lentiviral in vivo approach. No immunogenicity-related or integration-related toxicities were identified during the available follow-up. That observation is reassuring but cannot establish long-term genomic safety in four people followed for a maximum of 15 months.

The scale of the evidence is critical. This was not a new large trial but extended observation of the same four patients previously reported. The study was single-arm and open-label, and the authors did not perform formal hypothesis testing because of the extremely small sample. Enrollment was discontinued after EsoBiotec was acquired by AstraZeneca.

The 100% initial response rate therefore should not be read as an estimate of how effective a future in vivo CAR-T product will be.

What the experiment has established is more fundamental. Functional anti-cancer CAR-T cells can be generated directly inside a human body and can produce clinically meaningful tumor responses without removing the patient's T cells for individualized manufacturing.

The next challenge is now much clearer. Making CAR-T cells inside the patient proved possible. Making those cells persist and control advanced cancer for long enough may be the harder problem.