Bites of Bio

CAR-T cell therapy for solid tumors: why it struggles and what early colorectal cancer data show

Share



Credit: Unsplash

Chimeric antigen receptor T cell (CAR-T) therapy produced objective tumor responses in 39% of 18 patients with advanced colorectal cancer in an early-phase study presented at the European Society for Medical Oncology (ESMO) Congress in October 2026. The result, reported in ESMO abstract 1003O, suggests that engineered immune cells can attack some solid tumors, although small, uncontrolled trials cannot establish long-term effectiveness or survival benefit.

CAR-T works against blood cancers but faces different obstacles in solid tumors

CAR-T therapy begins by collecting a patient’s T cells and genetically engineering them to express a chimeric antigen receptor. The receptor recognizes a selected protein on the surface of a target cell. After manufacturing and expansion, the modified T cells are infused into the patient, usually following chemotherapy that reduces competing immune cells.

Approved CAR-T products have transformed treatment for some leukemias, lymphomas, and multiple myeloma. These blood cancers often display recognizable targets, such as CD19 or B cell maturation antigen, that are comparatively accessible to circulating engineered T cells. Even in these diseases, relapse, inflammatory toxicity, and manufacturing constraints remain important.

Solid cancers present a different biological problem. Their malignant cells reside within tissue surrounded by blood vessels, connective tissue, normal cells, and immunoregulatory cells. An engineered T cell must reach the tumor, cross those barriers, recognize enough cancer cells, remain functional, and avoid harming healthy tissues carrying the same target.

Finding a tumor-specific antigen remains a central safety challenge

Blood cancer targets such as CD19 can also occur on normal cells, but loss of normal B cells can often be managed medically. Many solid-tumor antigens occur on healthy organs whose injury may be much harder to tolerate. This creates an on-target, off-tumor safety problem: a receptor may recognize its intended protein while damaging normal tissue that expresses it.

Solid tumors also display antigen heterogeneity. Cells within the same mass, or between a primary cancer and its metastases, may express different amounts of the target. If treatment eliminates target-positive cells while sparing target-low or target-negative populations, the remaining cancer can regrow. Research programs are investigating dual-target receptors and engineered circuits that require combinations of molecular signals before activating a T cell.

This antigen problem also helps explain why encouraging preclinical CAR-T experiments in solid tumors do not automatically translate into durable clinical benefit. A target that appears selective in a model may behave differently across human organs and genetically diverse tumors.

The tumor microenvironment can exclude and disable T cells

Getting into a solid tumor is not equivalent to reaching every malignant cell. Abnormal blood vessels, dense extracellular matrix, and mismatched chemokine signals can limit how many infused T cells enter a lesion. Within the tumor, suppressive immune cells, inhibitory checkpoint signaling, competition for nutrients, and local metabolic stress can impair T cell activity.

Long-lasting function matters because a responding tumor can recur after the engineered cells lose potency or disappear. Research into T cell memory and exhaustion has therefore become relevant to cellular therapy design. Investigators are testing ways to alter receptor signaling, preserve less-differentiated T cell states, redirect cell trafficking, and resist suppressive signals without causing uncontrolled immune activation.

The same difficulty in maintaining antitumor T cell function is visible in laboratory work on PPARδ signaling in engineered T cells. Such experiments identify possible engineering strategies, but mouse findings and laboratory observations are not evidence of benefit in patients.

A GCC-directed CAR-T trial reported responses in colorectal cancer

At ESMO 2026, investigators presented a phase 1 trial of CAR-T cells directed against guanylyl cyclase C (GCC), a protein associated with intestinal tissue and expressed by many colorectal cancers. Eighteen patients with advanced disease received the engineered cells after lymphodepleting chemotherapy. Six received the lower cell dose and 12 received the higher dose.

The reported objective response rate (ORR), the proportion with a measurable partial or complete tumor response under the study’s assessment criteria, was 39% overall. ORR was 33% in the lower-dose cohort and 42% in the higher-dose cohort. The reported disease control rate, which also includes stable disease, was 94%. These outcomes describe a small, single-arm experiment; they do not show that CAR-T performs better than existing colorectal cancer treatments.

Toxicity was substantial. All 18 patients developed cytokine release syndrome (CRS), an inflammatory reaction associated with immune-cell activation. Most cases were grade 1 or 2, but one patient in the higher-dose group had grade 3 CRS. Grade 3 diarrhea occurred in 10 of the 18 patients, and the presentation described additional mucositis, rash, and infections. Gastrointestinal injury is particularly relevant when the antigen is present in intestinal tissue.

No treatment-related deaths were reported in this 18-patient presentation. That finding should not be taken as proof of safety: the sample was small, follow-up and dosing were still being evaluated, and uncommon serious events may emerge in larger studies. The higher-dose cohort also had more heavily pretreated patients, making simple between-dose comparisons unreliable.

A separate PRAME study used TCR-engineered cells, not CAR-T

Another ESMO presentation involved IMA203CD8, an investigational T cell receptor (TCR)-engineered therapy directed against preferentially expressed antigen in melanoma (PRAME). Unlike conventional CAR-T receptors, which bind proteins on the cell surface, engineered TCRs recognize short protein fragments presented by human leukocyte antigen molecules. TCR therapy can therefore target certain intracellular proteins, although it requires the appropriate antigen-presentation machinery and patient eligibility.

In this phase 1 study, 98 patients with advanced PRAME-positive solid tumors were enrolled, and 90 were included in the efficacy-evaluable population. Investigators reported an unconfirmed ORR of 51% (46 of 90). The confirmed ORR was 41% (36 of 87) among patients evaluable for response confirmation. These denominators should not be treated as interchangeable.

The cohort included patients with melanoma, synovial sarcoma, ovarian cancer, uterine cancer, and other tumors. Seven complete responses were reported in ESMO’s summary, and some responses had persisted for years. Those observations are encouraging, but they came from a nonrandomized study with varying tumor types, doses, and follow-up periods.

CRS occurred frequently with IMA203CD8 as well. The ESMO summary reported grade 3 CRS in 8% and grade 4 CRS in 1%, along with less frequent neurotoxicity. Because the PRAME trial used TCR-engineered cells, its 41% confirmed response rate must not be presented as a CAR-T result or compared directly with the 39% GCC CAR-T response rate.

Early responses do not yet establish durable benefit or routine clinical use

Both studies were early-phase and lacked randomized control groups. Their chief purposes included evaluating safety, dose selection, and initial antitumor activity. Objective response rates measure tumor shrinkage under defined criteria, but they are not equivalent to prolonged survival, improved quality of life, or a cure.

The GCC trial involved only 18 people and included a clinically important burden of intestinal toxicity. The PRAME data involved a different therapy platform and a selected antigen-positive population. Larger studies will need to establish which patients benefit, how long responses last, and whether toxicity can be reduced enough for treatment outside specialized centers.

These studies also need to be distinguished from established therapies using tumor-infiltrating lymphocytes, another form of cellular immunotherapy that is not CAR-T. Describing every engineered or infused T cell treatment as CAR-T obscures crucial differences in receptor biology, eligibility, and evidence.

New engineering approaches aim to improve access, recognition, and persistence

Current strategies target several different failure points. Locally delivered cells or altered chemokine receptors may improve tumor entry. Multispecific receptors and logic-gated recognition systems are intended to reduce antigen escape or restrict activation to tumor cells. Gene editing and altered intracellular signaling may help cells resist immune suppression or maintain useful activity longer.

Each strategy introduces tradeoffs. Stronger activation can increase inflammation; broader antigen recognition may increase normal-tissue injury; and complex cell manufacturing can limit accessibility. A credible path forward requires simultaneous evidence of tumor control, manageable safety, and reproducible production.

ESMO 2026 adds early human evidence that engineered T cell therapies can generate measurable responses in selected solid tumors. It does not erase the biological obstacles that have limited CAR-T outside blood cancers. The next decisive evidence will come from larger trials testing durability, comparative benefit, and adverse events rather than response rates alone.

References

  1. Chang L, et al. Phase I study of GCC CAR-T therapy in patients with advanced colorectal cancer. ESMO Congress 2026. Abstract 1003O. ESMO congress summary.
  2. Busse A, et al. Targeting PRAME-positive tumors with IMA203CD8, a PRAME-directed TCR T cell therapy: results from a phase 1 study. ESMO Congress 2026. Abstract 1946O. ESMO congress summary; trial results filed with the SEC.
  3. Hou AJ, Chen LC, Chen YY. Navigating the CAR-T cells through the solid-tumour microenvironment. Nature Reviews Drug Discovery. 2021;20:531–550. Journal article.

Leave a Comment

Your email address will not be published. Required fields are marked *