CAR T cells require optimization to be effective in patients with solid tumors. There are many barriers affecting their ability to succeed. One barrier is persistence, as to achieve an optimal antitumor response, infused CAR T cells must engraft and persist. This singular variable is impacted by a multitude of factors—the CAR T cell design, lymphodepletion regimen used, expansion method to generate the T cell product, and more. Additionally, external agents can be utilized to augment CAR T cells, such as the addition of novel cytokines, pharmaceutical drugs that bolster memory formation, or other agents during either the ex vivo expansion process or after CAR T cell infusion to support them in the oppressive tumor microenvironment. This review highlights many strategies being used to optimize T cell persistence as well as future directions for improving the persistence of infused cells.

Background: Responsiveness to chimeric antigen receptor (CAR) T cell therapy correlates with CAR T cell expansion and persistence in vivo. Multiple strategies improve persistence by increasing stem-like properties or sustaining CAR T cell activity with combination therapies. Here, we describe the intrinsic ability of CAR T cells to differentiate into memory T cells, the effect of cytokine armoring, and neoadjuvant CD4 depletion therapy on CAR and tumor-specific endogenous memory T cells. Methods: TRP1-specific or NKG2D CAR T cells alone or with Super2+IL-33 (S233) armoring and/or CD4 depletion were evaluated in immunocompetent B16F10 melanoma or MC38 colon cell carcinoma models without preconditioning. We characterized CAR and endogenous tumor-specific memory T cell precursors, establishment of circulating (T_CIRC) and resident (T_RM) memory T cell subsets, and ability to protect against secondary tumors. Results: TRP1-specific or NKG2D CAR T cells had no effect on primary tumor growth in immunocompetent mice unless they were combined with S233 armoring or CD4 depletion. Unarmored CAR T cells expressed a stem-like phenotype in the tumor-draining lymph node and differentiated into CAR T_CIRC memory cells in lymphoid organs and CAR T_RM cells in the skin. In contrast, S233-armored CAR T cells exhibited an activated effector phenotype and differentiated inefficiently into CAR effector and central memory T cells. Combining CD4 therapy with unarmored CAR T cells increased CAR T_CIRC and T_RM memory T cells. Either CD4 depletion therapy or S233-armored CAR T cells induced activation of tumor-specific endogenous T cells that differentiated into both T_CIRC and T_RM memory T cells. CD4 depletion and S233-armored CAR T cell combination therapy synergized to increase endogenous memory T cells. Conclusions: Unarmored TRP-1-specific or NKG2D CAR T cells have intrinsic stem-like properties and differentiate into memory T cell subsets but are non-protective against primary or secondary tumors. S233 cytokine armoring alone or with CD4 depletion improved effector responses but limited CAR memory T cell generation. S233-armored CAR T cells or CD4 depletion therapy induced endogenous tumor-specific T_CIRC and T_RM T cells, but the combination potentiated endogenous memory T cell generation and resulted in improved protection against B16F10 rechallenge.

Immuno-oncology has revolutionized cancer treatment by mobilizing the immune system to eliminate tumors. Although immune checkpoint inhibitors and T cell therapies have mediated durable responses in hematologic malignancies and select solid tumors, most patients still relapse or fail to respond. To overcome these limitations, novel, next-generation immune constructs, including bispecific-cell therapy combinations, armored cells, tethered cytokines, immune-stimulatory antibody conjugates, and in vivo gene editing, are being developed to enhance specificity, persistence, and immune activation. This Society for Immunotherapy of Cancer roadmap highlights emerging technologies that integrate, redirect, or potentiate immune response. We examine advances in construct design, strategies for clinical translation, and opportunities for combinatorial approaches. By addressing translational barriers and real-world challenges, we outline how innovative engineering can unlock a new era of safe, durable, and accessible immunotherapies. Realizing this potential will require coordinated efforts from researchers, clinicians, industry leaders, and policymakers to deliver curative outcomes to patients worldwide.

Summary: Systemic administration of homeostatic γ-chain cytokines mediates antitumor responses in some patients treated with adoptive immunotherapy. Yet many patients experience toxic side effects. New work presented herein suggests these limitations can be overcome by membrane-tethering IL15 and IL21 to T-cell products. This finding has major implications in advancing medicine. See related article by Nguyen et al., p. 1555

Type I interferons (IFNs) signal by forming a high affinity IFN-IFNAR2 dimer, which subsequently recruits IFNAR1 to form a ternary complex that initiates JAK/STAT signaling. Among the 12 IFNα subtypes, IFNα1 has a uniquely low affinity for IFNAR2 (<100 × of the other IFNα subtypes) and commensurately weak antiviral activity, suggesting an undefined function distinct from suppression of viral infections. Also unique in IFNα1 is substitution of a serine for phenylalanine at position 27, a contact point that stabilizes the IFNα:IFNAR2 hydrophobic interface. To determine whether IFNα1-S27 contributes to the low affinity for IFNAR2, we created an IFNα1 mutein, IFNα1-S27F, and compared it to wild-type IFNα1 and IFNα2. Substitution of phenylalanine for serine increased affinity for IFNAR2 ∼4-fold and commensurately enhanced activation of STAT1, STAT3, and STAT5, transcription of a subset of interferon stimulated genes, and restriction of vesicular stomatitis virus infection in vitro. Structural modeling suggests that S27 of IFNα1 disrupts the IFNα:IFNAR2 hydrophobic interface that is otherwise stabilized by F27 and that replacing S27 with phenylalanine partially restores the hydrophobic surface. Disruption of the hydrophobic IFNα:IFNAR2 interface by the unique S27 of IFN α1 contributes to its low affinity and weak antiviral activity.