CAR T Cells Were Engineered to Resist a Tumour’s Immune Brake

Solid tumours defend themselves not only against drugs but also against immune cells. Even when a T cell recognizes a cancer cell, the tumour can surround it with signals that weaken the attack. A new study of CAR T therapy for liver cancer tried to disconnect one of those molecular brakes.

CAR T cells are a patient’s own T lymphocytes, modified in the laboratory with an artificial receptor that recognizes a selected target on cancer cells. The approach has transformed the treatment of some blood cancers, but solid tumours are harder. The cells must enter dense tumour tissue and remain active in an environment built to suppress immune responses.

Two jobs: find the tumour and resist its defence

For hepatocellular carcinoma, the researchers targeted glypican-3, a protein found on the surface of many liver cancer cells. It can act as a molecular marker that guides CAR T cells towards the tumour.

Recognition alone may not be enough. Liver tumours often contain high levels of TGFβ, a signalling molecule that restrains T-cell activity. When TGFβ binds to its receptor, it sends an inhibitory message into the immune cell.

The experimental therapy, called C-CAR031, combines two modifications. One allows the T cell to recognize glypican-3. The other adds an altered TGFβ receptor that can encounter the suppressive signal but cannot transmit it into the cell. The tumour can still press the brake, but part of the connection between the signal and the CAR T cell has been cut.

Before testing the therapy in patients, the team studied the cells in laboratory experiments and in mice. The armoured CAR T cells remained more active in the presence of TGFβ and controlled tumour growth more effectively in the preclinical models.

What happened in patients

The first-in-human trial included 36 patients with unresectable hepatocellular carcinoma that had progressed after previous systemic treatment. Their tumours expressed glypican-3, and all participants were Asian.

Selection was strict. Seventy-four potential participants failed screening, mainly because their tumours did not meet the required level of glypican-3 expression. The findings therefore apply to a specifically selected group whose cancers carried this molecular target.

This was an open-label, single-arm phase I study. There was no control group, and patients received one of several dose levels. The main purpose of this stage was to assess safety and determine how the treatment behaved in people, although tumour responses were also measured.

Measurable tumour lesions became smaller in 32 of the 36 patients. That does not mean that nearly 90% had a formal clinical response. Under standard response criteria, 44.4% of participants had enough tumour shrinkage to qualify as a partial or complete response. The median best reduction in tumour size was 41.6%.

The responses were often temporary. Their median duration was 4.4 months. In half of the patients, the disease progressed before 4.2 months; in the other half, progression occurred later.

Median overall survival was 14.2 months, meaning that half of the patients lived longer and half lived for less time. Without a comparison group, however, it is impossible to determine how much the treatment itself contributed to that result.

The therapy also produced immune toxicities associated with strong CAR T-cell activation. Cytokine release syndrome—an inflammatory reaction caused by the rapid activation of many immune cells—occurred in 34 patients. Most cases were not severe, but two reached grade 3. Nine patients experienced other serious adverse events that were not related to changes in blood-cell counts.

The tumour still found ways to escape

Analyses of tumour samples offered clues to why some patients did not respond and why responses could fade. In some tumours, the amount of glypican-3 fell. By losing the target, cancer cells became harder for the CAR T cells to recognize.

In other cases, TGFβ activity increased. The altered receptor reduced the effect of this suppressive pathway, but it did not make the cells completely resistant. Tumours can adapt: when one route of escape is blocked, selection favours cancer cells that survive through another route.

The results provide an early clinical signal that CAR T cells can be engineered not only to recognize a tumour but also to resist part of its local defence system. They do not yet show that C-CAR031 is better than existing treatment or that it extends survival. The study was small, included only Asian participants, involved a limited number of centres, used several dose levels and had no control group.

Against a solid tumour, teaching an immune cell to recognize its target is only the first task. The cell must also survive the environment designed to switch it off.