Prospects to improve chimeric antigen receptor T-cell therapy for solid tumors

Introduction

In recent years, cancer immunotherapy has achieved remarkable successes through the introduction and fast US FDA approval of checkpoint blockade antibodies and advancements in the use of patient-derived chimeric antigen receptor (CAR) T-cells. Adoptive transfer of CAR T-cells against the pan-B-cell marker CD19 has been highly successful in treating B-cell malignances, especially acute lymphoblastic leukemia (ALL) [1–4], but has also led to fatal toxicities [5]. Except from a promising study using CAR T-cells against the disialoganglioside GD2 where three out of 11 neuroblastoma patients went into complete remission [6], CAR T-cells have not yet been successful for treatment of solid tumors. This can be attributed to the fact that solid tumors have physical and biochemical barriers that restrict CAR T-cell infiltration. It has become clear that the immunosuppressive microenvironment inside solid tumors dampens the activity of CAR T-cells [7,8]. Immune suppression is mediated by regulatory T-cells (Tregs), myeloid-derived suppressor cells (MDSCs) and tumor-associated macrophages (TAMs) as well as IL-10 and IL-4, and other immunosuppressive soluble factors such as TGF-β, indoleamine 2,3-dioxygenase (IDO), arginase and elevated extracellular potassium concentration [9]. Tumor cells and stromal cells within solid tumors can also express programmed cell death protein (PD) ligand-1 (PD-L1), which can interact with PD-1 on CAR T-cells leading to inhibition of CAR T-cell activation [10,11]. Alteration of the immunosuppressive tumor microenvironment (TME) and induction of bystander immune responses are therefore of utmost importance to improve CAR T-cell therapy for solid tumors. In this short article, we discuss recent advancements in the field and these advancements are summarized in an illustration (Figure 1).

Approaches to target & modify the tumor microenvironment

CAR T-cells targeting tumor stroma or tumor vasculature

Tumor growth is not just determined by malignant cells, but also by the tumor stroma that consists of non-malignant cells: fibroblasts, mesenchymal cells, cells forming new blood and lymph vessels, and various immune cells together with extracellular matrix and inflammatory mediators [12]. Cancer-associated fibroblasts (CAFs) are important promoters of tumor growth by production of growth factors, chemokines and extracellular matrix components, which recruit endothelial cells and pericytes [13]. Therefore, directing CAR T-cells to fibroblast activation protein (FAP), which is abundantly expressed on CAFs, is an attractive therapeutic strategy. Pre-clinical studies using FAP-specific CAR T-cells have exhibited encouraging results in a mouse model [14] and a clinical trial for patients with malignant pleural mesothelioma is currently ongoing (NCT01722149). Blood vessels within solid tumors aberrantly express antigenic growth factors and receptors. Therefore, CAR T-cells targeting VEGFR-2, which is overexpressed in tumor vasculature and is associated with tumor angiogenesis, have been developed to direct a specific immune response against the tumor vasculature [15,16]. Preclinical data with VEGFR-2-specific CAR T-cells have shown promising antitumor effects in a variety of human cancers [16].

CAR T-cells engineered to secrete enzymes or cytokines

The hostile TME with abundant reactive oxygen species (ROS) impairs CAR T-cell activity [17]. CAR T-cells with co-expression of catalase (CAR-CAT), an enzyme that catalyzes hydrogen peroxidase (H₂O₂) to water and oxygen, has been developed to counterattack ROS [18]. CAR-CAT T-cells demonstrated a reduced oxidative state with less ROS accumulation and maintained their antitumor activity despite high H₂O₂ levels [18]. Moreover, CAR-CAT T-cells exerted a substantial bystander protection to other infiltrating immune effector cells (T-cells and NK-cells) by protecting them from exposure to the high concentration of H₂O₂. This approach provides a new strategy for protecting tumor-infiltrated immune cells from oxidative stress-mediated suppression [18].

Tumor-infiltrating lymphocytes (TILs) have been isolated and engineered to constitutively express and secrete IL-2 to reduce the need for exogenous cytokine support [19,20]. However, IL-2-secreting TILs showed toxicity, and clinical efficacy did not improve [21]. T-cells have also been engineered to constitutively express IL-15 [22]. However, since IL-15 expressing T-cells could be leukemogenic as they retain proliferative capacity without the need of exogenous cytokines [23], such T-cells were later safety-modified by including the iCaspase9 suicide gene [24]. Some studies have investigated CAR T-cells with constitutive or induced expression of the T helper type (Th)-1-associated cytokine IL-12 upon target cell recognition [25]. It was shown that IL-12 secreting CAR T-cells enhance antitumor response by being less susceptible to Tregs, MDSCs and TAMs [26–28]. Induced IL-12 not only enhanced CAR T-cell activation, it also resulted in a Th1 polarization that attracted endogenous T-cells and innate immune cells to eradicate CAR antigen-negative cancer cells [25,29]. Recently, CAR T-cells with a synthetic Notch (synNotch) receptor were constructed for precision tumor cell killing. SynNotch CAR T-cells are only activated in the presence of dual antigens on tumor cells and can, therefore, be used to target high-risk antigens or carry toxic payloads [30–32].

CAR T-cells with chemokine or signaling receptors

The endothelial cells in the tumor bed can also act as a barrier by blocking effector T-cells from entering while allowing Tregs through [33]. CAR T-cells can be modified to express a chemokine receptor along with the CAR for increased migration to tumor. CD30-targeted CAR T-cells co-expressing CCR-4 resulted in increased migration and antitumor efficacy for Hodgkin lymphoma with CCL-17 and CCL-22 expression [34]. GD2-targeted CAR T-cells co-expressing CCR-2b resulted in increased CAR T-cell infiltration in neuroblastomas that express CCL-2 [35].

Epstein-Barr virus (EBV)-specific T-cells modified with a CAR against GD2 and IL-7Rα had an increase in proliferation and resistance to Treg-mediated immune suppression, and restored T-cell responsiveness to IL-7, an important cytokine for T-cell homeostasis [36]. EBV-specific T-cells modified with a dominant negative TGF-β receptor (DN TGF-βR) were able to resist the inhibitory effects of tumor-derived TGF-β, both in vitro and in vivo [37,38].

CAR T-cells with costimulatory molecules or chimeric switch-receptors

Co-stimulatory signals are of importance for T-cell survival, proliferation and persistence. PSMA-specific CAR T-cells have been modified to constitutively express the costimulatory ligands CD80 and 4-1BBL. Those CAR T-cells can stimulate each other resulting in significantly improved antitumor responses [39]. In order to interfere with the inhibitory PD-1 signaling for CAR T-cells, a genetically engineered chimeric switch-receptor was constructed, comprising the truncated extracellular domain of PD-1 and the transmembrane and intracellular signaling domain of CD28. Mesothelin-specific CAR T-cells with the chimeric switch-receptor led to enhanced CAR T-cell infiltration and significant regression of solid tumors [40]. In one study, tumor-associated antigen (TAA)-specific T-cells were transduced with a hybrid receptor comprising of IL-4 exodomain and IL-7 endodomain [41]. When these T-cells encounter IL-4 they signal in an IL-7 fashion, which supports T-cells persistence in the TME [41].

Combination strategies for CAR T-cells

Based on the composition of the TME, including the immune and inflammatory landscape, tumors can be distinguished as immunogenic (hot) and non-immunogenic (cold) [42,43]. An immunogenic TME consists of infiltrating CD8⁺ T-cells, proinflammatory cytokines and often the presence of tumor-adjacent organized lymphoid aggregates, while a non-immunogenic TME lacks these components and is populated by immunosuppressive cells and soluble factors, while the tumor cells and infiltrating stromal cells express ligands for inhibitory receptors [43]. Combining CAR T-cells with agents that can reduce immune suppression in the TME is therefore attractive.

Inhibitory interaction can be blocked by monoclonal antibodies against PD-1 or PD-L1, which has shown encouraging results in many solid tumors [44,45]. However, the success of this approach depends on the pre-existence of TAA-specific T-cells [46]. Therefore, combination anti-PD-1 or anti-PD-L1 antibodies with CAR T-cells may be beneficial for treatment of non-immunogenic ‘cold’ solid tumors. It may also be possible to arm the CAR T-cells so that they secrete anti-PD-1 single-chain antibody fragments upon target cell recognition. Other inhibitory molecules such as CTLA-4, Tim3, LAG3 may be targeted in a similar fashion. Alternatively, CAR T-cells may be produced with siRNA that block expression of the inhibitory receptors.

Blockage of soluble tumor suppressive mediators is another way to enhance CAR T-cells function. Dual IL-10 and PD-1 blockage has been demonstrated to enhance expansion and function of TAA-specific T-cells [47]. Adenosine disturbs T-cell function by interacting with the adenosine receptor A2aR [48]. Dual A2aR and PD-1 blockage enhanced function of CD8⁺ TILs with an increased release of IFN-γ and granzyme B [49]. IDO inhibits T-cell function by depletion of the essential amino-acid tryptophan and activates Tregs through kynurenine production [50]. Dual IDO and PD-1 or CTLA-4 blockage have shown significant clinical activity in patients without adverse side effect [51,52]. In a recent study it was found that IDO inhibits CD19 CAR T-cells [53]. This could however be counteracted by applying lymphodepleting drugs or a specific IDO inhibitor [53]. Therefore there is great interest in combining IDO inhibitors with CAR T-cells for the treatment of solid tumors.

Expansion protocols & specific T-cell subsets to improve CAR T-cell therapy

For adoptive T-cell transfer, large amounts of CAR T-cells need to be generated ex vivo under good manufacturing practices (GMP) conditions in a relatively short period of time. Most clinical expansion protocols use anti-CD3/CD28 beads in combination with IL-2 or other cytokines for expansion of CAR T-cells [54]. The anti-CD3/CD28 beads deliver signal-1 and costimulatory signal-2 to the CAR T-cells leading to cell proliferation without provoking anergy or early apoptosis. Irradiated artificial antigen-presenting cells (aAPCs) have also been developed and used for expansion of T-cells. The aAPCs can be modified in different ways to support T-cell function, such as expressing co-stimulatory ligands (CD28, 4-1BB) [55], CAR ligands [56] or membrane-bound cytokines such as IL-15 and IL-21 [57]. Irradiated aAPCs have the advantage of mimicking stimulation from professional antigen-presenting dendritic cells (DCs) leading to fitter CAR T-cells. For TIL expansion, a rapid expansion protocol (REP) has been developed, with irradiated allogeneic peripheral blood cells usually from three different donors together with a stimulating anti-CD3 antibody (OKT-3) and IL-2 [58]. The allogeneic feeder cells promote a mixed leukocyte reaction where TILs expand and proliferate due to the allogeneic reaction [59]. The feeder cells also act by anchoring the Fc-portion of the OKT-3 antibody for optimal stimulation of T-cells. Ex vivo-expanded T-cells obtained with the above-mentioned protocols have potent anti-tumor activity in vitro, but have been shown to be less effective in triggering solid tumor regressions in vivo [60]. The expanded T-cells can be exhausted and sensitive to hostile factors in the TME such as ROS and immunosuppressive cytokines.

For CAR T-cell therapy, especially for solid tumors, it is important to make sure that the ex vivo-expanded T-cells are able to further proliferate in vivo after transfer, as this has been shown to translate to better antitumor efficacy. For example, it has been reported that T-cells with a less-differentiated phenotype expanded in the presence of IL-7 or IL-15 have a better proliferation capacity in vivo [61]. Exogenous IL-21 was used for CAR T-cells expansion by K562-based aAPCs to provide full activation signals to T-cells, which resulted in an increased T-cell yield, increased granzyme B expression and increased IFN-γ release [57]. We have developed a novel CAR T-cell expansion protocol, named AEP, where irradiated and pre-activated allo-sensitized allogeneic lymphocytes (ASALs) are used as helper cells to license OKT-3-armed allogeneic mature DCs to expand CAR T-cells. The allogeneic ASAL/DC combination yields an enriched Th-1 polarizing cytokine environment, including IFN-γ, IL-12 and IL-2, and optimal co-stimulatory signals for CAR T-cell stimulation [62]. AEP-expanded CAR T-cells showed better survival and cytotoxic efficacy under oxidative stress and exposure to immunosuppressive cytokines compared to CD3/CD28 bead-expanded and REP-expanded CAR T-cells [62].

It has been demonstrated that purified central memory (T_cm), naive (T_n) and stem like central memory (T_scm) cells are more potent in elimination of tumors compared with effector memory T-cells (T_em cells) [63–65]. Those T-cells have a better capability to proliferate and persist in vivo after ex vivo expansion and adoptive transfer [64]. The optimal proportion of CD8⁺ versus CD4⁺ T-cells in the CAR T-cell pool is not yet known. However, recently CD19 CAR T-cells of a defined ratio of CD8⁺ (T_cm) and CD4⁺ T-cells showed improved antitumor response and progression-free survival for patients with non-Hodgkin's lymphoma [66].

Conclusions & future perspective

CD19 CAR T-cell therapy has demonstrated promising clinical activity in B-cell malignancies. However, limitations still exist in terms of toxicity in B-cell ALL patients while responses are not yet sufficiently long-lasting for patients with B-cell lymphomas. For CAR T-cell therapy for solid tumors, increasing strategies must be explored to armor T-cells with different agents to overcome the immunosuppressive TME and further enhance bystander anti-tumor killing. We envision that emphasis will be on trying to convert immune ‘cold’ to immune ‘hot’ tumors in order to alter the immune and inflammatory landscape in the TME and thereby make these tumors accessible for checkpoint blockade antibody and/or CAR T-cell therapies. Other factors that will be further addressed for CAR T-cell therapy are: vector and CAR design; T-cell subsets to be used for engineering; T-cell transduction and ex vivo expansion procedures; pre-conditioning strategies for patients before CAR T-cell transfer and the use of cytokine support after CAR T-cell transfer.