Adoptive cellular therapies: the current landscape

In TCR gene therapy, tumor antigen recognition is achieved by the introduction of a novel T cell receptor into T cells. Autologous T cells are redirected to recognize tumor antigens by engraftment of genes encoding TCR-α and β chains. TCR-modified T cells exert antigen recognition in an MHC-dependent manner [23, 51]. Examples of targetable antigens are tissue-specific antigens like melanoma differentiation antigens, cancer/testis (C/T) antigens, overexpressed antigens, and viral antigens. In most clinical trials, peripheral blood T cells for genetic modification are obtained via leukapheresis and are transduced by gamma-retroviral or lentiviral vectors that incorporate the TCR genes into the host genome, which results in high-level expression of the introduced TCR [36, 52, 53]. Other means of genetic engineering that are currently in development include the transposon/transposase system, such as Sleeping Beauty [54], or Crispr/Cas9 based technology [55]. These technologies do not require the production of lenti- or gamma-retroviral vectors, and may therefore provide a more flexible and cheap platform. As in ACT with TIL, most clinical protocols with TCR gene therapy have incorporated preconditioning of the patient with a lymphodepleting regimen prior to T cell infusion, to facilitate engraftment and expand the lifespan of the modified T cells. In addition, IL-2 administration following T cell infusion has been used [56].

The first evidence of the feasibility and clinical potency of TCR gene therapy targeting the melanoma differentiation antigen MART-1, present in approximately 80–95% of melanomas [57, 58], was demonstrated in 17 patients with progressive metastatic melanoma. Gene transfer efficiencies of 17–62% were achieved and an objective partial tumor response was seen in two (13%) of the treated patients [25]. In a subsequent clinical trial, 36 patients with metastatic melanoma were treated with high-avidity TCRs targeting the melanoma differentiation antigens gp100 or MART-1 and objective tumor responses were observed in 19% and 30% respectively [16].

Next to melanoma differentiation antigens as a target for TCR gene therapy, notable clinical responses have also been achieved when targeting C/T antigens. The genes encoding these antigens are normally expressed during embryogenesis, but are epigenetically silenced later in life, except in spermatocytes. Cancers oftentimes aberrantly re-express these genes, hence the name C/T antigens. These antigens include NY-ESO-1, MAGE-family, and SSX [59]. NY-ESO-1 is expressed in up to 52% of melanomas [60], 82% of neuroblastomas [61], 80–100% of synovial sarcomas and mixoid and round cell liposarcomas [62], 43% of ovarian cancer [63] and to a lesser extent in multiple other tumor types. Because of its restricted expression in normal tissues in combination with a widespread expression in cancer, NY-ESO-1 has been frequently used as a target in TCR gene therapy. In a phase II clinical trial, 5/11 (45%) patients with progressive advanced melanoma and 4/6 (67%) patients with synovial sarcoma had an objective tumor response after infusion of 1.6–130 × 10⁹ retrovirally transduced autologous T cells with a NY-ESO-1 targeting high affinity TCR [64]. Similar response rates have been observed when targeting MAGE-A3, which is expressed in 62% of melanomas [65]. In a clinical phase I/II trial, nine pretreated patients with either advanced melanoma (n = 7), synovial sarcoma (n = 1), or esophageal cancer (n = 1) were treated with anti-MAGE-A3 TCR gene-modified T cells in a dose-escalating manner. The first three patients were treated with 3 × 10¹⁰ transduced cells and the remaining patients were treated with 1 × 10¹¹ transduced cells, with a transduction efficiency of 70% (CD8⁺ T cells). Of these patients, five (56%), including four melanoma patients and the single synovial sarcoma patient, had an objective tumor response [66].

Tumor regression has also been seen in patients with T cells targeting carcinoembryonic antigen (CEA), which is overexpressed in colorectal adenocarcinoma but is also present in normal epithelial cells [67]. Three patients were treated with 2 × 10⁸–4 × 10⁸ CEA reactive-TCR transduced T cells, and one of these patients achieved a partial response. However, all 3 patients developed a severe transient colitis [68].

Viral antigens such as human papillomavirus (HPV) as possible targets for TCR therapy have also been explored for HPV-associated epithelial cancers including cervical, oropharyngeal, vulvar, vaginal, anal, and penile cancer. T cells engineered with the TCR recognizing the HPV-16 E6 epitope from metastatic anal cancer also showed recognition of HPV-16 positive cervical and head and neck cancer cell lines and this epitope may thus be a potential target for TCR therapy [69].

Transduction of T cells in an early differentiation stage (central memory CD8⁺ T cells) seems to result in greater anti-tumor responses when combined with tumor-antigen vaccination and exogenous IL-2 in preclinical murine models [70]. It has also been shown that the cytokines IL-7 and IL-15, which play a role in the development of central memory T cells in vivo, favor the generation of this T cell subset in culture systems [70, 71]. In most cancer immunotherapy studies, CD8⁺ T cells have been the main focus because of their known strong cytolytic capacity [72]. However, recent evidence shows that CD4⁺ T cells also exert anti-tumor efficacy [44] and CD4⁺ T cells have already been part of the TCR infusion product used in early clinical trials [16, 64, 66]. In a validated good manufacturing practice (GMP) production process, CD4⁺ as well as CD8⁺ T cells are retrovirally transduced and expanded in the presence of IL-7 and IL-15 in combination with anti-CD3/CD28 bead selection and activation [73]. This protocol is being evaluated in a phase I/IIa trial in HLA-A:*02-01 MART-1 positive patients with advanced melanoma, including patients with uveal melanoma (NCT02654821), and should provide more insight into the feasibility and safety of TCR therapy in melanoma.

In summary, these results demonstrate that TCR therapy can be a potent anti-tumor treatment in various cancer types. However, as the antigens explored up until now are not solely expressed by the tumor, the identification of antigens restricted to tumors is essential to further increase the efficacy and safety of TCR therapy.

CARs are hybrid receptors and are currently genetically constructed to contain a scFv of a monoclonal antibody as the antigen-binding extracellular domain, an intracellular CD3ζ chain as the TCR signaling domain and an additional co-signaling domain, mainly CD28 and 4-1BB (CD137) or others, to deliver co-stimulation [23, 74]. Multiple methods to transfer CARs to T cells have been developed, but most commonly used is transfer by retroviral infection, which has proven to be efficacious and safe [75]. Induction of cytotoxic activity of the manufactured T cell is a result of antigen-binding to the scFv, leading to downstream signaling through phosphorylation of CD3ζ and additional signaling cascades via the co-stimulating domains [76], similar to signaling following T cell activation through the TCR complex. Unlike TCR gene therapy, CAR T cells show target recognition in an MHC-independent manner, as was first demonstrated by the groups of Kuwana and Eshhar in the late 1980s in the first generation CARs [18, 77]. Since this first discovery, CAR therapy has undergone major improvements and thus far most research has been performed in hematological malignancies such as B cell lymphoma and leukemia. The co-receptor CD19 showed to be an optimal target [78‐80] as it is expressed early during B cell development and expression is maintained until plasma cell differentiation. B cell malignancies originating from these B cell differentiation states also express CD19. As CD19 is also expressed on normal B cells, treatment with CD19 CAR T cells will result in a transient or lasting B cell aplasia and hypogammaglobulinemia [81]. In 2003, the group of Sadelain at the Memorial Sloan-Kettering Cancer Center (New York, US) were the first to show successful transduction of peripheral blood lymphocytes with CD19 CARs in immunodeficient mice with various B cell malignancies resulting in tumor reduction and even long-term eradication [82].

The engineering of CARs has evolved over time and resulted in four generations of CAR molecules. In 1993, first-generation CAR consisted of a scFv and intracellular CD3ζ domain which mediated the production of IL-2 and non-MHC-restricted cell lysis upon activation in murine models [83]. However, the presence of costimulatory signals lead to better T cell activation (by providing signal two) and resulted in better T cell proliferation [84]. These second and third generation CARs additionally contained costimulatory domains to enhance T cell survival, activation and expansion [20, 85, 86]. Second generation CARs carry the costimulatory domains CD28 [87] or 4-1BB [85]. These showed enhanced TCR signaling, production of cytotoxic cytokines such as IL-2, proliferation and survival [20, 85, 87, 88]. Third generation CARs aim to encompass the signaling capacity of two costimulatory domains, mostly CD28 in combination with 4-1BB or OX-40. Addition of proliferative cytokines such as IL-12 or costimulatory ligands such as 4-1BBL have proven to further potentiate the anti-tumor capacity of second generation CAR T cells in preclinical studies and are currently known as the fourth generation CAR T cells [89, 90]. These CAR T cells can also be referred to as T cells redirected for universal cytokine killing (TRUCKs), which can deliver a transgenic product to the targeted tissue. By using nuclear factor of activated T cells (NFAT) to induce cytokines such as IL-12, the area around the CAR-targeted tissue is made more favorable for an immune response [91].

In xenograft models comparing the efficacy of different CAR constructs, CARs consisting of two signaling domains (CD3ζ plus CD28) and 4-1BB ligand showed the greatest anti-tumor efficacy and also increased persistence in the peripheral blood compared to first generation CAR constructs [92]. With second generation CARs, complete response rates of around 40% have been demonstrated in acute lymphoblastic leukemia (ALL) murine models treated with 5–10 × 10⁶ CD19 CD28 or 4-1BB CAR T cells [82, 93, 94]. The first clinical trial to show clinically significant responses in patients with ALL was performed by Sadelain and co-workers in 2013. In this trial, five patients with relapsed B-ALL not previously treated with allogenic hematopoietic stem cell transplantation (allo-HSCT) were treated with 1.5–3 × 10⁶/kg CD19 CD28 CAR T cells after prior conditioning treatment with cyclophosphamide (1.5–3.0 g/m²) and subsequent allo-HSCT as per protocol and complete remissions were seen in all treated patients (n = 4) [26]. Finalization of this clinical trial in 2014 with a total of 16 patients resulted in a complete response rate of 88% [95]. In a case series by Grupp et al., two children with relapsed and refractory pre-B cell ALL received 1.4 × 10⁶–1.2 × 10⁷/kg CD19 4-1BB CAR T cells and both patients showed a complete remission, one of which was ongoing 11 months post-treatment (current status unknown) [80]. In a following phase I dose-escalation trial, 21 patients with relapsed or refractory ALL or non-Hodgkin lymphoma (NHL) were either treated with 1 × 10⁶/kg/dose, 3 × 10⁶/kg/dose or the entire CD19 CAR T cell product if the product did not meet the required dosage amounts. The maximum tolerated dose was 1 × 10⁶/kg/dose, all toxicities were temporary and a complete response rate of 67% (14/21 patients) was reached [96]. More recently in 2017, patients with refractory diffuse B cell lymphoma, primary mediastinal B cell lymphoma, or transformed follicular lymphoma were treated in a multicenter phase II trial with 2 × 10⁶/kg CD19 CD28 CAR T cells following low dose preconditioning regimens with cyclophosphamide (500 mg/m²/day) and fludarabine (30 mg/m²/day) for 3 days. Of 101 treated patients, an objective response was seen in 82% and 54% of patients showed a complete response, of which 40% were durable complete responses [97]. These response rates were reproducible, as in another study complete response rates of 57% were seen in 28 patients with refractory B cell lymphoma treated with a median of 5.79 × 10⁶/kg/dose CD19 4-1BB CAR T cells [98].

As in ACT with TIL, preconditioning lymphodepletion is commonly used in the clinical treatment protocol with CAR therapy. When patients with chemotherapy-refractory chronic lymphocytic leukemia were treated solely with CAR T cells (without lymphodepleting regimen), no clinical benefit and less persistence of the CAR T cells was observed. However, it is important to note that these patients in this small study also received a lower dose of CAR T cells [78]. The clinical successes with CAR T cell therapy has recently led to the FDA approval of two CD19 CAR therapies for ALL and NHL in 2017, namely axicabtagene ciloleucel (Yescarta) with costimulatory molecule CD28 and tisagenlecleucel (Kymriah) with costimulatory molecule 4-1BB [99]. Although high complete remission rates have been demonstrated with the use of CD19 CAR T cells, resistance via the loss of CD19 has been observed in 28% of young adult and pediatric patients with acute leukemia in an international trial [100].

As stated above, most of the research with CAR T cell therapy has been performed in hematological malignancies, but also other B cell lineage-restricted targets like CD22 and B cell maturation antigen (BCMA) are currently under investigation. Moreover, CAR T cell technology is being explored in solid tumors, however achieving limited clinical activity thus far. For example in sarcomas targeting ERBB2/HER2 [101], renal cell cancer targeting carbonic anhydrase IX (CAIX) [102], non-small cell lung cancer and cholangiocarcinoma targeting epidermal growth factor (EGFR) [103], and neuroblastoma targeting GD2 [104] and other solid tumors (targeting shared antigens including mesothelin and CEA) [105]. Recently, CAR T cells directed against the colorectal cancer antigen GUCY2C were investigated in murine models showing increased antigen-dependent T cell activation, cytokine production and killing of GUCY2C-expressing tumor cells [106].