Adoptive cell transfer immunotherapy for patients with solid epithelial cancers

Abstract

Efforts to apply adoptive cell transfer (ACT) immunotherapy to patients with common epithelial cancers have been stimulated by the demonstration that the majority of these patients contain lymphocytes reactive against the expressed products of their cancer mutations. Early efforts to specifically target these antigens have been promising.

Adoptive cell transfer (ACT) immunotherapy refers to the in vivo transfer of lymphocytes, designed to mediate anti-tumor immune reactions in cancer patients. These naturally occurring or genetically modified cells represent a “living drug” capable of expanding many thousand-fold inside the body when they encounter and destroy cells bearing their cognate antigens.

ACT has several unique features as a treatment approach for cancer. Because the original cells used for treatment are removed from the patient, it is possible to manipulate the host prior to cell transfer to provide an optimal tumor microenvironment in which the cells can function. Large numbers of anti-tumor lymphocytes (greater than 100 billion) can be readily grown in vitro under conditions suitable for in vivo administration. Populations of lymphocytes with optimal phenotype and anti-tumor properties can be selected and grown and the infused cells can be analyzed to identify the cell properties and effector functions required to mediate cancer regression in vivo.

Although ACT can be highly effective for the treatment of patients with metastatic melanoma, the major challenge is its application to patients with metastatic solid epithelial cancers that cannot be cured by any available systemic treatment and result in 90% of all deaths from cancer (over 600,000 deaths per year in the US). Recent developments have provided evidence that tumor-infiltrating lymphocytes (TILs) can recognize antigens on most common epithelial cancers and these findings have opened possibilities for improvement of this treatment approach.

In 1988, the first demonstration of effective ACT in patients utilized autologous lymphocytes infiltrating into the stroma of growing metastatic melanomas that were expanded in the laboratory prior to infusion.¹ These TILs could mediate the regression of established growing, vascularized metastatic melanomas in patients. Subsequent refinements involved pre-treatment of patients with lymphodepleting chemotherapy prior to the cell infusion, designed to eliminate regulatory T cells and myeloid derived suppressor cells that can interfere with the activity of the transferred lymphocytes. Lymphodepletion also increases the exposure of the transferred cells to cytokines without competitive binding by endogenous cells. Additional improvements in ACT have resulted from the addition of interleukin-2 (IL-2) administration to promote the in vivo growth of the infused cells as well as the development of growth techniques to increase the diversity and decrease the senescence of the infused population.

Trials in the National Cancer Institute, Surgery Branch of 192 patients with metastatic melanoma who received ACT with TILs prior to the administration of anti-PD-1 checkpoint therapy revealed objective response rates by RECIST criteria of 56%, with 48 patients (25%) showing durable complete and likely curative cancer regressions with a median potential follow-up exceeding 8 years.² Forty-six of the 48 patients achieving complete response received only a single treatment. A recent randomized clinical trial demonstrated improved response rates and survival using ACT with TILs compared to ipilimumab, a checkpoint inhibitor, in patients with metastatic melanoma.³

Major questions concerning the improvement of ACT cancer immunotherapy for patients with metastatic solid epithelial cancers revolve around identifying appropriate antigens to be targeted, the appropriate “fitness” of the cells to enable them to proliferate in the patient following infusion, as well as improving the ability of the transferred cells to infiltrate and attack the growing cancer.

The ability of ACT using TILs to mediate cancer regression in patients with metastatic melanoma provided us with an opportunity to study the mechanisms of action of this therapeutic approach. Single-cell transcriptome analysis of up to 10,000 cells in the infusion products administered to a group of patients, who either demonstrated a complete response to ACT or did not respond to treatment, identified a gene signature of “stem-like” cells marked by lack of expression of CD39 and CD69 that showed a high correlation with objective cancer regression in vivo.⁴ This finding is in concert with multiple findings in the ACT of established murine cancers showing that highly differentiated lymphocytes are less effective than stem-like cells in mediating regression of established cancer.

The ability of melanoma TILs to mediate tumor regression in the absence of on-target, off-tumor toxicities led us to explore the recognition by TILs of the expressed products of mutations present in the cancer and not in normal cells.⁵ Targeting non-mutated cell surface proteins can lead to life-threatening destruction of normal tissues. In this effort, whole-exome sequencing of cancer as well as normal tissue identified all non-synonymous mutations present in the cancer. We constructed peptides or mini-genes encoding all of the non-synonymous cancer mutations as 25-mers with the non-synonymous mutated amino acid in the middle flanked on each side by 12 normal amino acids. These 25-mers thus contained all of the possible 8–13 mer peptides likely to bind to the patient’s surface major histocompatibility complex (MHC) molecules. These 25-mers were pulsed or transfected into autologous antigen-presenting cells and cocultured with TILs to identify the antigen recognized by the melanoma TILs. Updated results in 86 consecutive patients with metastatic melanoma showed a median of 288 mutations (mean 556) per cancer and identified 47,827 mutations.⁶ We screened 15,622 of the expressed mutations for recognition by autologous TILs. Eighty-five percent of these patients showed reactivity against at least a single antigen and 1.4% of all mutations were immunogenic, i.e., recognized by the patient’s autologous T cells. Approximately 80% of the neoantigens recognized by melanoma TILs were MHC class I-restricted and all of the 218 neoantigens that were identified in these 86 patients were unique to the individual patient and not shared between any two patients.

Considerable effort is being devoted to the extension of the ACT immunotherapy approach to the treatment of patients with metastatic solid epithelial cancers that are not responsive to checkpoint inhibitors. We studied antigen identification in 205 consecutive patients with a variety of common epithelial cancers (Table 1). Of these patients, 151 (76%) showed reactivity of lymphocytes against one or more autologous non-synonymous mutations. A total of 363 cancer neoantigens were identified and all were unique except for a KRAS p.G12D mutant peptide recognized in the context of HLA-C*0802 by T cells from 2 patients. Thus, in contrast to prior dogma that melanomas were uniquely immunogenic, the majority of patients with common solid epithelial cancers can develop immune responses against the neoantigens expressed in the autologous cancer that can potentially serve as targets for ACT immunotherapy. Of interest, in contrast to melanomas, 50%–75% of neoantigen reactive cells in patients with gastrointestinal and breast cancers recognized MHC class II-restricted neoantigens. From 1.3% to 2.1% of mutations were immunogenic.^7,8

Table 1.

Mutated antigens recognized by TILs from 205 patients with epithelial cancers

Cancer	# of patients screened	# of patients with neoantigen reactivity	Total # of neoantigens recognized
Colorectal	99	81 (82%)	161
Anal	1	1 (100%)	1
Cholangiocarcinoma	15	11 (73%)	17
Pancreatic	11	7 (64%)	12
Esophageal	7	6 (86%)	9
Endometrial	3	3 (100%)	4
Breast	43	29 (67%)	100
NSCLC	14	9 (64%)	35
Ovarian	7	6 (86%)	16
Stomach	4	2 (50%)	6
Prostate	1	1 (100%)	2
TOTAL	205	151 (76%)	363

The ability to select and grow cells with specific anti-tumor reactivity against identified cancer antigens has resulted in durable responses in some patients with a variety of solid cancers such as cholangiocarcinoma, breast cancer, cervical cancer, and colorectal cancer, thus showing the potential application of this ACT approach to these common tumors.^8–10

The natural lymphocyte reactivity targeting non-shared unique antigens in each patient emphasizes the highly personalized nature of this ACT immunotherapy. Patients are treated with their own cells that target antigens unique to their own cancer. The targeting of mutated proteins is likely the “final common pathway” underlying the effectiveness of most cancer immunotherapies including IL-2, checkpoint modulators, and TILs.

An alternate approach to ACT immunotherapy in humans involves the retroviral insertion into autologous or allogeneic lymphocytes of receptors recognizing cancer antigens, thereby converting normal lymphocytes into anti-tumor effectors. Although natural TILs contain conventional alpha beta T cell receptors (TCRs), many efforts have attempted to use non-MHC-restricted chimeric antigen receptors (CARs). Construction of a single chain of the antigen-combining region of the heavy and light chain of antibodies connected to intracellular signaling molecules in lymphocytes can redirect lymphocyte recognition to that of antibodies. The use of CARs thus depends on the use of monoclonal antibodies that recognize the three-dimensional structure of molecules presented on the cell membrane. In the almost five decades since the description of monoclonal antibodies, there has been a paucity of antibodies described with unique recognition of molecules on the surface of cancer cells and not on corresponding normal cells. Targeting molecules on normal cells has resulted in serious clinical toxicities. CAR approaches have been effective in the treatment of patients with hematologic malignancies but not as yet for the solid cancers because of a lack of specific cell-surface targets.

Improved methods of targeted screening and in vitro sensitization have enabled the identification of libraries of TCRs targeting the expressed products of shared KRAS¹¹ and p53 mutations¹² across multiple MHC-I and MHC-II restriction elements. The genetic modification of normal lymphocytes to express these TCRs represents an approach to develop “off-the-shelf” reagents for use in ACT immunotherapy. Early efforts targeting shared KRAS and p53 mutations using TCR transfer has mediated cancer regressions in patients with pancreatic cancer¹³ and breast cancer.¹²

Improved methods to identify lymphocytes with appropriate stem-like qualities and selected to identify mutated epitopes are likely necessary to develop effective ACT for the common epithelial cancers that express fewer mutations than melanomas. Recent studies identified a gene signature of lymphocytes from freshly resected epithelial tumors that could identify the sub-populations of lymphocytes and their TCRs capable of recognizing autologous cancer antigens.^14,15

The creation of a highly personalized drug for each patient represents a considerable departure from the established norm in drug development. Traditional pharmaceutical companies depend on the development of “drugs in a vial” applicable to large numbers of patients and easily distributed widely. The development of the first vial can cost hundreds of millions of dollars and can be commercially pursued if subsequent vials can be produced for a few pennies or dollars. This approach using cytotoxic or targeted agents has been highly effective in prolonging the life of many cancer patients but has largely failed to cure patients with metastatic solid epithelial tumors. Virtually all patients with detectable metastatic epithelial cancers will die of their disease despite the best available systemic treatments.

The generation and administration of cells at each institution to treat their own patients is necessary for discovery and innovation in cell-based therapies but is not likely to bring personalized cell therapies to large populations in need. A practical model for treatment, pioneered by Kite Pharma and Novartis, is the development of central laboratories that can receive lymphocytes and/or tumors and prepare the “personalized drug” as cryopreserved cells for delivery to the home institution for infusion.

The administration of immunotherapy using checkpoint inhibitors has had a major impact on cancer treatment though most patients with solid epithelial cancers do not respond. The use of ACT with TILs or genetically modified lymphocytes represents an important area of recent progress in the development of cancer immunotherapies.