Why the Same CAR T-Cell Therapy Can Behave Differently: The Patient’s DNA May Matter

Every autologous CAR T-cell product begins with a patient’s own immune cells. T cells are collected from the blood, engineered to recognize a tumor target, expanded in the laboratory, and infused back into the same person. The resulting treatment is therefore not a completely standardized drug. It is a living cell population that carries both an artificial receptor and the inherited genetic background of its owner.

That genetic background may help determine how vigorously the cells expand and whether their immune response becomes dangerously inflammatory.

In a study published in Science Immunology, researchers analyzed whole-genome data from patients with aggressive B-cell lymphomas who received axicabtagene ciloleucel in the ZUMA-1 and ZUMA-7 clinical trials. The principal genomic analyses included 78 participants from ZUMA-1 and 115 from ZUMA-7.

When an immune attack becomes excessive

CAR T cells need to activate, multiply, and kill malignant cells. Excessive activation, however, can produce cytokine release syndrome, a systemic inflammatory reaction that may cause fever, low blood pressure, and organ dysfunction. The most severe cases can resemble hemophagocytic lymphohistiocytosis, a hyperinflammatory condition in which immune cells sustain a damaging feedback loop.

The researchers first examined 17 genes already associated with these syndromes. In ZUMA-1, potentially damaging inherited variants in STXBP2 were found in about 15% of patients with severe toxicity and in none of the control patients who avoided high-grade complications.

STXBP2 helps immune cells release the contents of cytotoxic granules, a process known as degranulation. This allows a T cell to deliver a lethal hit to its target. When the process is impaired, the cell may kill less efficiently yet remain stimulated for longer, continuing to release inflammatory signals.

Functional experiments supported this mechanism. CAR T cells in which STXBP2 was suppressed, or which carried certain patient-derived variants, showed reduced degranulation and produced more interferon-gamma, a major inflammatory mediator.

The clinical association was observed primarily in ZUMA-1 and was not reproduced as clearly in ZUMA-7. STXBP2 therefore cannot yet be treated as a validated predictive test for CAR T-cell toxicity.

Genetic signals linked to safety and activity

A genome-wide analysis produced two additional signals.

Potentially deleterious variants in ADAMTSL3 were more common among patients without severe toxicity in both trials, suggesting a possible protective effect. The gene may influence inflammatory signaling through the TGF-β pathway. However, the association was nominal and did not meet the stringent threshold required after correction for the large number of genome-wide comparisons.

Variants in PTPN22 were associated with greater CAR T-cell expansion after infusion in both cohorts. PTPN22 encodes a protein that restrains T-cell activation. Reduced activity of this molecular brake could allow engineered cells to respond and proliferate more strongly. Notably, PTPN22 variants occurred in some of the patients with the highest measured CAR T-cell expansion.

Expansion is an informative pharmacokinetic measure and often correlates with therapeutic response, but it is not equivalent to a clinical benefit. The study does not establish that carriers of a particular PTPN22 variant are more likely to enter remission or survive longer.

The therapeutic cell carries the patient’s biology

Differences in CAR T-cell outcomes are usually attributed to tumor burden, the patient’s general condition, manufacturing quality, or the design of the engineered receptor. This work adds another variable: inherited variants present in the patient’s original T cells may affect the behavior of the final therapeutic product.

The study was a retrospective genomic analysis of relatively modest cohorts, and some rare variants occurred in only a small number of patients. Several findings did not survive stringent correction for multiple testing. The results also concern one CAR T-cell product used against B-cell lymphoma and may not apply to other cancers or cellular therapies.

The findings are therefore not ready to guide treatment selection or routine genetic screening. Their broader importance is conceptual: in personalized cell therapy, individuality may reside not only in the receptor chosen to recognize the tumor, but also in the underlying biology of the cells into which that receptor is introduced.