CAR-Based Cell Therapy Moves From Hematology Toward Gastrointestinal Tumors
Gastrointestinal malignancies still drive more than one-third of cancer-related deaths worldwide, and advanced disease remains difficult to control even after years of progress in systemic therapy and immunotherapy. In that setting, the emergence of CAR-T and CAR-NK strategies marks a clear attempt to push cellular therapy beyond blood cancers and into the biologically harsher terrain of solid tumors.
The review centers on six targets that have drawn the most clinical attention in GI oncology: CLDN18.2, GUCY2C, MSLN, GPC3, CEA, and HER2. Each target reflects a different tumor type, antigen profile, and safety problem, but the clinical question is the same: can engineered immune cells locate the cancer, persist long enough to matter, and avoid injuring normal tissues along the way?
The answer is strongest so far for CLDN18.2.
CLDN18.2 Sets the First Randomized Benchmark in a Solid Tumor
Among the reviewed targets, CLDN18.2 is the most advanced. In the randomized phase II CT041-ST-01 trial, satricabtagene autoleucel improved progression-free survival compared with physician’s choice therapy in previously treated gastric or gastroesophageal junction cancer. That result matters because it provides the first randomized evidence that CAR-based therapy can work in a solid tumor setting, not just in hematologic disease.
The broader implication is biological as well as clinical. CLDN18.2 is a tight-junction protein with a tumor-associated expression pattern that offers a more tractable target than many broadly shared antigens. Solid tumors have repeatedly frustrated cellular therapies because they are not only antigenically messy but physically and immunologically suppressive; a target with more selective tumor enrichment gives engineered cells a better starting position.
Even so, the leap from a randomized signal to routine practice remains substantial. Manufacturing, access, toxicity management, and patient selection all shape whether this strategy can move from a research achievement to a reproducible treatment option in everyday oncology.
Other Targets Show Early Activity, but the Field Is Still Uneven
Beyond CLDN18.2, the review points to encouraging early-phase activity for GUCY2C in metastatic colorectal cancer and GPC3 in hepatocellular carcinoma. Those signals are important because they show the field is not relying on one antigen or one disease site. Instead, investigators are testing whether different tumor biologies can be matched with different cellular products.
This is where the promise of the platform meets its main limitation. Solid tumors differ sharply in antigen density, immune exclusion, stromal barriers, and vascular accessibility. A therapy that reaches one tumor type may fail in another, not because the CAR concept is flawed, but because the local biology blocks trafficking, expansion, or durable antitumor function.
The review also discusses mesothelin, CEA, and HER2 as active targets in the GI space. Each has obvious appeal because of known tumor expression, yet each also brings safety concerns tied to expression on nonmalignant tissues or the risk of off-tumor recognition. In solid tumors, the difference between a useful target and a dangerous one can be a narrow band of antigen expression rather than a clean on-off switch.
Why Solid Tumors Resist Engineered Immune Cells
The review places the clinical data in the context of several recurring barriers: antigen heterogeneity, immunosuppressive tumor microenvironments, on-target/off-tumor toxicity, and manufacturing challenges. Those obstacles explain why CAR therapies have transformed hematologic malignancies far more quickly than they have changed solid tumor care.
Tumor heterogeneity means a single antigen may not cover every malignant clone. Immune suppression can blunt the activity of infused cells even after successful delivery. Off-tumor toxicity remains a constant concern when the same antigen exists, even at lower levels, in normal tissue. Manufacturing adds another layer of complexity because individualized cell products require time, infrastructure, and consistent quality control.
These limitations are not abstract. They shape whether a patient can actually receive treatment, whether treatment can be delivered quickly enough for aggressive disease, and whether the therapeutic window is wide enough to justify the risk.
CAR-T and CAR-NK Are Not Interchangeable Strategies
Although both CAR-T and CAR-NK platforms aim to redirect immune cytotoxicity, they are not identical tools. The review treats them as related but distinct approaches within GI oncology, and that distinction is clinically meaningful. CAR-NK products may offer theoretical advantages in safety and manufacturing flexibility, while CAR-T products have the deepest clinical track record across oncology.
In practice, the field is still testing which cellular backbone is best suited to which antigen and tumor type. A target like CLDN18.2 may be the clearest current proof of concept, but broader success will probably depend on matching the right target, the right cell product, and the right disease context rather than assuming one construct will fit all GI cancers.
What the Current Evidence Can and Cannot Support
The review is a narrative synthesis, not a head-to-head comparative trial program. Its value lies in organizing an expanding clinical literature around the targets that appear most plausible and the disease states where early evidence has begun to accumulate. The data remain strongest for gastric and gastroesophageal junction cancer through CLDN18.2-directed therapy, while signals in colorectal cancer and hepatocellular carcinoma are still earlier and less definitive.
For clinicians, the most concrete takeaway is that cellular immunotherapy for GI cancers has moved beyond speculation. There is now randomized evidence for at least one target, and several others have produced early-phase activity that justifies continued development. For patients, that may eventually mean more precise immunotherapy options in diseases where standard systemic therapy still leaves too much residual risk.
The next phase will depend on whether these products can be scaled, made safer, and integrated into treatment pathways without losing the biological specificity that makes them attractive in the first place.