Solid tumours: Building bridges to CAR-T success

1 INTRODUCTION

Immunotherapy using chimeric antigen receptor (CAR) engineered T-cells has shown tremendous promise in recent years, particularly in the management of refractory B-cell and plasma cell-derived cancers.¹ Spurred on by this, cell therapy has now become the leading area of drug development in the immuno-oncology sector, with CAR T-cell immunotherapy at the vanguard of this advance.² However, non-haematological cancers account for 90% of all malignant diagnoses and remain stubbornly impervious to this emerging therapeutic modality. Challenges that need to be addressed are summarised in Figure 1. How can we build on the gains made against blood cancers to secure impact where need is greatest—in patients with relapsed refractory solid tumours?

2 TARGET SELECTION

Target selection remains the first issue to consider. In blood cancers, a “lineage ablation” strategy has applied whereby a cell surface molecule shared by malignant and normal cell types is targeted, accepting that loss of normal B lymphocytes can be offset with immunoglobulin replacement therapy. Unfortunately however, this type of approach cannot generally be undertaken safely in patients with solid tumours. The challenge of target selection is highlighted by the fact that almost 100 candidates have been investigated in solid tumour CAR research.³ Thirty of these targets have been evaluated in clinical trials, an experience that has revealed the frequent occurrence of on-target off-tumour toxicity.⁴ A further related difficulty is heterogeneity of target expression, favouring the emergence of antigen loss leading to therapeutic failure.⁵ To overcome these issues, more complex targeting strategies require consideration. One approach entails the engagement of ligand families that are commonly dysregulated in cancer, such as the eight NKG2D ligands. These ligands are upregulated by cell stress events such as DNA damage that are highly prevalent in transformed cells.⁶ Accordingly, it is estimated that at least 80% of human cancers express these ligands.^7-12 Moreover, co-expression of multiple NKG2D ligands within a single tumour is generally detected.^7-12 Targeting this family using NKG2D-containing CARs offers the dual advantage that risk of antigen loss is minimised while an evolutionarily optimised ligand receptor interaction is harnessed for therapeutic benefit.¹³ NKG2D ligand-targeted CAR T-cells are not inhibited by soluble NKG2D ligands at concentrations found in patients with advanced malignancy.¹³ An alternative strategy entails the use of split chimeric receptors that provide complementary signalling upon engagement of two or more tumour targets.¹⁴ A particular advantage of such dual receptor strategies is that CAR T-cell docking on tumour cells is driven by both receptor-ligand interactions, creating an avidity effect that can enhance target antigen sensitivity.¹⁵ More recently, this platform has been refined to deliver an activating and dual co-stimulatory signal using a so-called parallel (p)CAR, an approach that has been described by five independent research groups within a short time frame.^16-20 The pCAR approach not only provides a logic-gated strategy to ensure greater precision of cancer targeting but it also delivers synergistic dual co-stimulation, minimising exhaustion in favour of functional CAR T-cell persistence. A further elegant strategy to achieve precision targeting of tumours involves synthetic (syn)Notch receptors whereby engagement of a tumour-associated target by synNotch triggers the transcriptional upregulation of a second CAR.²¹ By this means, specificity is dependent on the co-localisation of both targets within the tumour microenvironment (TME). Alternatively, the propensity of tumours to contain an aberrant vasculature can be harnessed through the engineering of hypoxia-sensing CARs, whereby CAR cell surface expression is contingent upon entry to the hypoxic TME.²² Additionally, the abnormal tumour vasculature may be targeted using CARs directed against vascular endothelial growth factor receptor, prostate-specific membrane antigen, CLEC14, integrins such as αvβ3 or extracellular matrix components such as fibronectin fibronectin with extra domain B (EDB) or fibronectin with extra domain A (EDA). SUPRA CAR provides another flexible approach whereby a universal CAR is combined with complementary leucine zipper-containing zipFv proteins. This system allows combinatorial antigen detection as well as drug-controlled inhibition of CAR function, albeit at the price of requiring the administration of two or more therapeutic agents.²³ To access intracellular antigens, neoantigen-specific CARs have been described.²⁴ These CARs are generally human leucocyte antigen (HLA) restricted and thus require matching with patient HLA antigen status as well as maintained HLA antigen expression by tumour cells.

3 TUMOUR TRAFFICKING OF CAR T-CELLS

Trafficking of CAR T-cells to tumours represents a second important obstacle. Whereas intravenously infused CAR T-cells have direct access to malignant cells in blood cancers, they must leave the bloodstream and access diverse tissues to enter solid tumour tissue. Imaging studies have shown that when tumour trafficking does occur in man, CAR T-cells may accumulate at the periphery of the tumour lesion, with inadequate penetration into the tumour core.²⁵ Intra-tumoural delivery of CAR T-cells provides one route to overcome this^{26, 27} and may find particular application in the context of malignancies affecting the central nervous system.²⁸ Alternatively, chemokine receptors may be co-expressed alongside a CAR to leverage tumour derived chemokines, an approach that improves not only efficacy but also safety of this approach in pre-clinical models.²⁹ A third strategy to increase CAR T-cell recruitment to solid tumours entails the administration of radiotherapy³⁰ or immunogenic chemotherapy such as oxaliplatin that favours chemokine release within the TME.³¹ Another approach to boost infiltration of CAR T-cells within the TME involves the co-expression of matrix-disrupting enzymes such as heparinase³² or targeting of stromal cells using chemotherapy (e.g., nab paclitaxel) or a second population of CAR T-cells (e.g., specific for fibroblast-activating protein).³³

4 ADDRESSING THE TUMOUR MICROENVIRONMENT

The solid TME is the ultimate battleground for CAR T-cells and imposes an impressive array of physical, chemical and biological hurdles to anti-tumour immunity. High interstitial pressure within solid tumours tends to hinder drug delivery, including that of CAR T-cells. However, nanoparticle albumin-bound (nab) paclitaxel can exploit the enhanced permeability and retention effect to achieve improved drug delivery to tumour sites,³⁴ where it eliminates cancer-associated fibroblasts and remodels stromal collagen content.³⁵ With this in mind, nab paclitaxel has been included in conditioning regimens employed in some CAR T-cell clinical trials.³⁶ A wide range of CAR T-cell armouring technologies are also available to modulate the TME including cytokines such as interleukin (IL)-12,³⁷ or IL-18,³⁸ although toxicity relating to their pro-inflammatory properties requires consideration. Alternatively, oncolytic viruses may be employed to reverse the immunosuppressive nature of the TME, rendering it more favourable to CAR T-cells.³⁹ More recently synthetic Notch T-cells engineered to produce IL-2 within the TME have achieved highly impressive pre-clinical efficacy in supporting CAR T-cell immunotherapy⁴⁰ and clinical trials of this approach are eagerly awaited.

A further important hindrance to efficacy against solid tumours is CAR T-cell exhaustion. This is exacerbated by exposure to high antigen burden, inhibitory factors within the TME and also tonic signalling by CARs themselves.⁴¹ Several strategies have been advanced to minimise exhaustion. Tonic signalling may be minimised by reduced⁴² or regulated, rather than constitutive, CAR expression. The latter may be achieved by insertion of the CAR-encoding cDNA into the T-cell receptor α locus⁴³ or under the transcriptional control of a synNotch receptor.⁴⁴ Alternatively, CAR expression may be pharmacologically regulated, thereby reducing tonic signalling and allowing the rapid termination of CAR-mediated toxicity.⁴⁵ Another approach to reduce exhaustion involves expansion of cell products over a shorter period⁴⁶ or in the presence of a chemical agent that maintains the CAR T-cells in a less differentiated fitter state (e.g., dasatinib⁴⁷ or a p38 MAP kinase inhibitor).⁴⁸ Signalling attributes of the CAR are also important in optimising fitness of cell products. Illustrating this, delivery of a calibrated activating signal⁴⁹ or optimised co-stimulation^{16, 50} both reduce exhaustion in the face of repeated antigen exposure. Inhibitory immune checkpoints operating in the TME may be disabled, for example by co-expression of PD1-specific scFv,⁵¹ shRNA directed against immune checkpoints (e.g., PD1, LAG3 and TIM3⁵² or adenosine 2A receptor),⁵³ or expression of arginine synthetic enzymes⁵⁴ in the CAR-engineered cells. Dominant negative transforming growth factor (TGF)-β receptor has also been co-expressed in CAR T-cells to effectively block immunosuppressive effects of this cytokine. However, a clinical trial in which this technology was employed was terminated prematurely owing to two patient deaths, raising caution about this approach.⁵⁵ Recent evidence also implicates epigenetic programming in the induction of CAR T-cell exhaustion, a process that can be reversed using gene editing of DNA methyltransferase 3⁵⁶ or decitabine treatment.⁵⁷ Exhaustion can also be modulated by enforced expression of some transcription factors (e.g. TCF1,⁵⁸ c-JUN⁵⁹) or knockdown of others (e.g., GATA3,⁶⁰ transcription factor transducing-like enhancer of split 4 [TLE4],⁶¹ Ikaros family zinc finger protein 2 [IKLF2],⁶¹ TOX,⁶² ARID1A⁶³ or NR4A transcription factors⁶⁴). In considering this last approach, consequences of global dysregulation of transcription factor networks require careful consideration from a safety perspective.⁶⁵

A further issue is tonic signalling, particularly by scFv-based CARs, which promotes T-cell exhaustion and therapeutic failure.⁴¹ This can be mitigated by a number of strategies such as targeted gene delivery to the T-cell receptor α locus using CRISPR-Cas9 gene editing.⁴³ CRISPR-Cas9 may also be used to rapidly screen large CAR T-cell libraries, enabling the selection of superior architectures for solid tumour immunotherapy.⁶⁶

5 HOST CELLS FOR CAR-BASED IMMUNOTHERAPY OF SOLID TUMOURS

While greatest success has been achieved with CAR-expressing αβ T-cells, there has been increasing recent interest in the use of alternative cell hosts. Peripheral blood-derived γδ T-cells of the δ2 subtype possess intrinsic antitumour immunity, which can be further boosted by the simple expedient of expansion in the presence of TGF-β.⁶⁷ Alternatively δ1 subtype γδ T-cells^{68, 69} and invariant natural killer (NK) T-cells⁷⁰ also possess anti-tumour activity and can be expanded at high efficiency from peripheral blood. Natural killer cells represent yet another option, and encouraging clinical data have been reported using umbilical cord derived CAR NK cells in the context of CD19 targeting.⁷¹ Macrophages represent another potential host cell, which demonstrates particular tropism for solid tumours and pre-clinical data also supports their utility as CAR hosts.⁷² All of these formats are of potential utility in the allogeneic setting, although clinical efficacy data in the solid tumour arena remain to be achieved with each of these platforms.

6 CONCLUSIONS

Immunotherapy of solid tumours using CAR T-cell immunotherapy lags considerably behind blood cancers, for which six products have been approved in the US and one additional product approved for use in Spain. Here, recognised challenges to solid tumour CAR-based immunotherapy have been summarised together with a range of potential solutions that may help to bridge the gap to success. Ultimately, opportunities for advanced engineering of the cells, deployment of combinatorial therapeutic approaches and advances in manufacturing technologies offer considerable promise for the development of successful CAR T-cell drugs for solid tumours in the near future.

CONFLICT OF INTEREST STATEMENT

J.M. is CSO, scientific founder and shareholder of Leucid Bio, is a member of the scientific advisory board of Arovella Therapeutics Ltd, and has undertaken consultancy work for Bristol-Meyers-Squibb, Juno, Celgene, Ellipses Pharma, Poolbeg Pharma and Biotest.