ROR1-Directed Chimeric Antigen Receptor T Cell Recognition of Self-Antigen Is Associated with Acute Toxicity, T Cell Dysfunction, and Poor Tumor Control.

Chimeric antigen receptor T cells (CAR T) have demonstrated robust clinical efficacy in B cell and plasma cell malignancies. In contrast to success in these hematological malignancies, the application of CAR T cells as a treatment for solid tumors has yielded limited efficacy. While there are multiple mechanisms whereby CAR T cells targeting solid tumors could be hindered in their efficacy, of primary concern is the specificity of recognition. Ideally, investigators desire to select target tumor antigens that are uniquely expressed by the neoplasm to avoid normal tissue ("on-target/off-tumor") toxicity, which can have serious clinical compilations, including death. We investigated the consequences of on-target/off-tumor CAR T cell recognition of the Receptor Tyrosine Kinase Like Orphan Receptor 1 (ROR1) gene, which has been proposed as a clinical target for CAR T cells. The potential for CAR T-mediated toxicity can be informed by the preclinical application of rigorous bioinformatics, and direct expression analysis using in situ hybridization (ISH), immunohistochemistry, (IHC), and flow cytometry. Using these approaches, we found that in humans and mice, ROR1 was expressed in the lungs and other normal tissues. As a specificity control, we used B-cell maturation antigen (BCMA) and confirmed that BCMA expression was limited to plasma cell-containing lymphoid tissues. Next, we generated anti-ROR1 CAR T cells (cross reactive with both human and mouse ROR1) and human-specific anti-BCMA CAR T cells. Using in vitro culture models, we demonstrated that both the ROR1 and BCMA CAR T cells robustly recognized and killed antigen expressing tumor cell lines. To follow the fate of CAR T cells in vivo, we applied ISH and IHC to study T cell activation and trafficking of anti-ROR1 and control anti-BCMA CAR T cells. ROR1-expressing multiple myeloma cells (ROR1+/BCMA+-RPMI-8226) were injected into the flanks of immune deficient NSG mice, and after tumor establishment, CAR T cells were administered by IV injections. Immediately following administration, CAR T cells traffic to the lung before release and subsequent trafficking to tumor. We observed markedly different behavior of the two CAR T cell populations. The anti-ROR1 CAR T cells accumulated in lung where they caused vasculitis and interstitial pneumonia, whereas anti-BCMA CAR T cells did not. Mice receiving anti-ROR1 CAR T cells lost weight and appeared ill. Within the lung, anti-ROR1 CAR T cells expressed the T cell activation / exhaustion marker PD-1. Lung accumulation and activation required a functional CAR molecule, as no pathology was observed with T cells expressing a CAR containing ROR1 binding elements in which the T cells signaling domains were deleted. Anti-ROR1 CAR T cell trafficking to the tumor was delayed and, compared to anti-BCMA CAR T cells, T cell infiltration of xenografts was sparse and tumor growth inhibition was not observed. By contrast, mice receiving anti-BCMA CAR T cells had robust T cell infiltration of xenografts and tumor clearance within 12 days. These findings raise concern about the effectiveness and safety of anti-ROR1 CAR T cells in humans and demonstrate that this type of preclinical safety assessment paradigm may be useful to de-risk potential CAR targets and for modeling CAR T cell toxicity and efficacy.