What Is Antigen Escape, and Why Do Some Cancers Stop Responding to CAR-T Therapy?

What This Article Covers

CAR-T (chimeric antigen receptor T-cell) therapy has produced some of the most striking remissions in modern oncology, particularly in B-cell leukemias and lymphomas targeted through the CD19 protein. But a meaningful share of patients who respond initially will later relapse. One major driver of that relapse — antigen escape — is fundamentally different from the T-cell exhaustion mechanism covered elsewhere on this site. This article explains what antigen escape is, the several distinct biological routes tumors use to achieve it, how it differs from exhaustion, and where the field stands on countering it.

What Antigen Escape Actually Is

A CAR-T cell is built around a single receptor that recognizes one specific protein on the surface of cancer cells — for many B-cell cancers, that protein is CD19. This precision is the therapy's strength, but it is also its vulnerability: if the cancer cell population evolves so that CD19 is no longer present, or no longer detectable, the CAR-T cell has nothing to grab onto, regardless of how many CAR-T cells are circulating or how well they are functioning otherwise. Antigen escape refers collectively to this family of tumor-side changes. It is a form of Darwinian selection: CAR-T therapy applies intense pressure on a cancer cell population, and any pre-existing or newly arising subpopulation that lacks the target antigen has a survival advantage and can eventually repopulate the disease.

The Several Biological Routes to Escape

"Antigen escape" is not one mechanism but a category of several. A 2026 Frontiers in Immunology review lays out multiple distinct pathways by which CD19-negative relapse arises. In some patients, small populations of CD19-negative leukemic cells exist even before CAR-T treatment begins, and these clones simply expand once CD19-positive cells are eliminated. In other cases, CD19 is downregulated through active biological processes: mutations and aberrant RNA splicing (including retention of intron 2 in the CD19 gene), epigenetic silencing of the CD19 promoter, and altered protein trafficking. A related and distinct phenomenon is epitope masking, where CD19 protein is still present but is bound or obscured in a way that shields it from CAR-T recognition — a resistance route documented in laboratory studies of CD19-positive leukemia blasts.

Perhaps the most dramatic route is lineage switch. In a subset of B-cell ALL patients — associated in particular with KMT2A gene rearrangements — the leukemia doesn't just lose CD19, it transforms into an entirely different disease: acute myeloid leukemia. Work published in Nature Communications showed this isn't simple clonal selection of a pre-existing myeloid clone but active epigenetic reprogramming: loss of the B-cell transcription factors PAX5 and EBF1 under CAR-T pressure was shown, using gene editing, to be sufficient to recapitulate this lineage conversion in the laboratory. The Frontiers review found lineage switch occurred in roughly 8% of relapsed patients following CD19 CAR-T therapy in a cohort of 420 children and young adults with relapsed/refractory B-ALL, compared with less than 1% in chemotherapy-treated patients. Similar lineage-plasticity escape has also been documented after CD7-targeted CAR-T therapy in T-cell ALL, where a case report described relapse as acute myeloid leukemia following elimination of CD7-positive disease.

How This Differs From T-Cell Exhaustion

It's worth being precise about the distinction, since both mechanisms can produce the same outcome — relapse — through opposite routes. T-cell exhaustion (covered in depth in our companion article, "What Is CAR-T Cell Exhaustion, and Why Does Relapse Happen?") describes a state in which the CAR-T cells themselves become functionally depleted after prolonged antigen stimulation, losing cytotoxic capacity even while the tumor still expresses the target antigen normally. Antigen escape is essentially the mirror image: the CAR-T cells can remain persistent and functional, but the tumor has changed so there is no longer a target for them to engage. Research reviewed here makes this distinction explicit, noting that CD19-positive relapses tend to correlate with reduced CAR-T potency and poor persistence, while CD19-negative relapse occurs through active immune selection pressure despite persistent, functional CAR-T cells. In practice, exhaustion and escape aren't always neatly separable in an individual patient, and both can contribute over the course of treatment — but they call for very different scientific solutions.

What's Being Done About It — and Real Reasons for Optimism

This is where there's genuinely encouraging news. Because antigen escape often exploits the "one lock, one key" design of first-generation CAR-T cells, one of the most active areas of CAR-T research is building constructs that target two antigens simultaneously — commonly CD19 together with CD22, another protein broadly expressed on B-cell malignancies. The logic is straightforward and compelling: a tumor cell would need to independently lose both markers at once to escape, a substantially higher evolutionary bar than losing one.

The early clinical results are worth paying attention to. A Phase 1 trial of CD19/CD22 dual-targeting CAR-T cells published in Nature Medicine in 2021 enrolled 39 heavily pretreated adults with relapsed or refractory B-cell malignancies (17 with B-ALL, 22 with large B-cell lymphoma) — including patients who had already progressed after CD19-directed therapy — and reported that all 17 evaluable B-ALL patients achieved minimal-residual-disease-negative complete remission, alongside a favorable safety profile. Separately, a Phase 2 investigator-initiated trial of a bicistronic (dual-gene) CD19/CD22 CAR-T construct in 343 pediatric patients with relapsed/refractory B-ALL, reported in late 2024, showed a one-year event-free survival of 75.5% and one-year overall survival of 93.5%, with outcomes improving further among patients who proceeded to bridging bone marrow transplant.

These are genuinely promising signals that dual- and multi-antigen targeting can raise the bar against antigen escape as a resistance strategy — real, hard-won progress against a problem that has limited durable remissions for years.

That said, the caveats matter. These are still early- to mid-phase trials, not yet a routine standard of care, and none of them eliminate relapse entirely — even dual-targeted cells can eventually be evaded if a tumor loses one antigen while persisting via the other, less dominant target. Other experimental strategies, including approaches designed to reduce epitope masking, remain largely preclinical or early-stage. None of this should be read as evidence that antigen escape has been "solved."

Bottom Line

Antigen escape is a tumor-side resistance mechanism distinct from CAR-T cell exhaustion: rather than the T cells wearing out, the cancer itself changes — losing, hiding, or downregulating its target antigen, or in some cases transforming into a different disease lineage entirely — to survive the immune pressure CAR-T therapy applies. It is a well-documented and clinically significant cause of relapse, particularly in CD19-directed therapy for B-cell leukemia. The dual-antigen CAR-T constructs now moving through clinical trials represent a genuinely promising and rational countermeasure, with early efficacy data that is worth feeling encouraged about. But this remains an active area of investigation rather than a solved problem, and patients and clinicians alike should treat current results as encouraging early evidence, not a guarantee against future relapse.