Adoptive cell transfer: a clinical path to effective cancer immunotherapy

The role of ACT in human cancer immuotherapy

Current efforts in the immunotherapy of human solid cancers fall into three main categories.

Adoptive cell transfer in animal models

Shortly after the demonstration that the cellular arm of the immune system was responsible for tissue rejection15, attempts were made to treat established rodent tumours by the transfer of immune cells. Examples of the development of ACT in animal models are summarized in TIMELINE 1. As T lymphocytes could not be grown in vitro, early efforts at ACT in the 1960s were limited to the use of cells obtained directly from immunized animals. Studies by Alexander and colleagues in the mid-1960s showed that sarcomas a few millimetres in size could be treated in rats by the administration of large numbers of lymphocytes from immunized syngeneic animals16. Extensive studies by Fefer and colleagues beginning in 1969 showed that intraperitoneal instillation of immune lymphocytes along with chemotherapy could effectively treat mice bearing intraperitoneal virus-induced lymphomas17. The ability to expand populations of anti-tumour immune cells using in vitro sensitization techniques in the mid-1970s freed these studies from the constraints imposed by the need for fresh cells from immunized hosts18. Eberlein et al. took this a step further by showing that the intravenous injection of immune cells grown in culture in IL2 could treat disseminated tumours in mice19 and subsequent studies showed that the concurrent administration of IL2 could further enhance the effectiveness of these IL2-dependent cells in vivo20.

The need for immunization of lymphocyte donors limited the application of this approach until 1986 when it was shown that tumour-infiltrating lymphocytes (TIL) from non-immunized mice bearing sarcomas or melanomas could be expanded in vitro in IL2 and used to successfully treat established lung and liver tumours21.

More recently, T-cell receptor (TCR) transgenic mice, all of whose lymphocytes express anti-gp100 tumour antigen TCRs, have provided a constant source anti-tumour T cells that are valuable for defining host factors and cell properties associated with effective ACT of large, vascularized, subcutaneous tumours in mice22. The importance of host immunosuppression as part of ACT was shown to be due to both the elimination of T regulatory cells23 as well as the elimination of ‘cytokine sinks’ that compete with the transferred cells for homeostatic cytokines, such as IL7 and IL15, that are produced by host stromal cells24. The greater the degree of host lymphodepletion the more effective was the treatment25. The characteristics of the transferred cells themselves had a profound effect: anti-tumour T cells with a CCR7⁺,CD27⁺,CD28⁺,CD62L⁺ phenotype that is characteristic of central memory cells were more effective than highly differentiated cells that lost these markers26. Immunization of the host with a vaccine expressing a tumour antigen recognized by the transferred cells enhanced their therapeutic effect22. Antigen-presenting cells (APCs) also have a role in effective ACT as the injection of APC-stimulating molecules such as Toll-like receptor agonists enhanced the efficacy of treatment27. More recently, it was demonstrated that normal murine splenocytes transduced with a retrovirus encoding an anti-melanoma TCR resulted in the generation lymphocytes that recognized the melanoma in vitro and, when adoptively transferred, could mediate tumour regression in vivo28.

ACT is currently the most effective immunotherapy capable of mediating the rejection of large tumours in mice and optimization of the treatment in mouse models has had an important role in the design of ACT approaches in the human.

ACT in metastatic melanoma

Selected highlights in the development of ACT in patients with cancer are shown in TIMELINE 2.

An important step in the development of human ACT was the finding in 1987 that lymphocytes infiltrating melanomas could be grown in IL2 and exhibit major histocompatibility complex (MHC)-restricted recognition of the autologous melanoma29. Over 50 different antigenic epitopes either unique to the autologous tumour or widely shared among melanomas were characterized using these TIL12. Studies using improved culture methods capable of generating up to 1011 TIL showed that melanoma-specific activity could be detected in TIL from 81% of 36 consecutive patients30.

The infusion of autologous TIL grown from the resected tumour nodules of patients with metastatic melanoma represents the clearest example of the effectiveness of ACT for the treatment of patients with a metastatic solid cancer and has helped to elucidate the cellular and host characteristics required for effective treatment in humans. The original reports of this approach, first published in 1988 (REF. 1) and summarized in 1994 (REF. 31), described an overall objective response rate of 34% in 86 patients treated with autologous TIL plus high-dose IL2 (REF 31). Patients received two cycles of treatment separated by 2 weeks. Eighty-eight percent of patients received >1011 cells in the first cycle. Response rates were similar in patients treated alone (31%) or with low-dose cyclophosphamide (35%) and in patients who had not received prior IL2 (34%) or in patients refractory to prior IL2 therapy (32%). In this trial, TIL were administered regardless of their in vitro activity, although a retrospective analysis revealed that there was a highly significant correlation between clinical response and the ability of the TIL to lyse the autologous fresh tumour (P = 0.0008)32,33. Traffic of indium-111-labelled TIL to tumour deposits correlated with clinical response (P = 0.022). Shorter times in culture and shorter doubling times were also positively associated with response (P = 0.0001 and 0.03)33, in accord with the findings in mouse models that increased proliferative potential was an important property of clinically active cells.

Several shortcomings became apparent in these initial studies of ACT using TIL. Persistence of the transferred cells in vivo was short. Studies using retroviral insertion of the neomycin phosphotransferase gene to mark TIL and sensitive PCR assays to detect the transgene revealed that at 1 week barely 0.01% of cells in the circulation were the transferred cells34. Of the 29 patients that exhibited an objective clinical response, five were complete though only two were ongoing at 21 and 46 months. The median duration of the partial responders was 4 months31.

Emerging information from murine models of ACT emphasized the need for prior lymphodepletion to eliminate regulatory T cells as well as normal endogenous lymphocytes that compete with the transferred cells for homeostatic cytokines23,24,35. Studies using highly selected tumour reactive CD8⁺ clones administered following lymphodepletion did not result in objective tumour regression, suggesting that the polyclonal nature of tumour reactivity and possibly the presence of CD4⁺ cells were necessary to mediate tumour rejection3,36. This led to a new generation of ACT clinical protocols with altered techniques for cell growth and with profound host lymphodepletion before cell transfer2,3.

In the latest Surgery Branch, National Cancer Institute trials multiple TIL cultures were initiated from freshly resected metastatic melanomas (a mass of ≥1cm³ was required) and as soon as anti-tumour activity could be detected in vitro against specific unique or shared antigens, cells underwent a rapid expansion using the T-cell-stimulating antibody OKT3 and IL2 (FIG. 1). Approximately 5×10¹⁰ cells were infused immediately following a non-myeloablative preparative regimen consisting of 60 mg/kg cyclophosphamide for 2 days followed by 5 days of fludarabine at 25 mg/m². Twenty-five patients each also received either 2 Gy or 12 Gy plus the chemotherapy regimen. IL2 was administered for 2–3 days at 7.2×10⁵ IU/kg every 8 h.

Objective responses by standard response evaluation criteria in solid tumours (RECIST) were seen in 21 of 43 patients (49%), who received no total body irradiation, in 13 of 25 patients (52%) who received 2 Gy and 18 of 25 patients (72%) who received 12 Gy (TABLE 1). Patients with stable disease were not considered as responders as it is not possible to evaluate this criterion in the absence of randomized patient comparisons. Examples of anti-tumour responses in melanoma patients treated with ACT are shown in FIG. 2. Many of these responses at multiple metastatic sites are durable. None of the 10 complete responders have recurred at times from 8 to 63 months. Fifteen of the partial responders have ongoing responses from 4 to 64 months. The actuarial 3-year survival of patients receiving ACT with the chemotherapy non-myeloablative regimen alone or with 2 Gy total body irradiation is 25% and 42% respectively, compared with 14% for the no lymphodepletion group. The trend for increasing survival as a function of increasing lymphodepletion is highly significant (P = 0.007), but this should be interpreted with caution as this was not a randomized comparison79.

There were no added toxicities in these trials owing to the cell administration, although the expected toxicities of IL2 administration were seen most often owing to the capillary leak syndrome caused by IL2.Neutropenic fevers that subsided when white cells recovered were seen in several patients.

Several findings in this trial are of note. The transferred cells expanded in vivo and persisted in the peripheral blood in many patients, sometimes achieving levels of 75% of all CD8⁺ T cells at 6–12 months after infusion2. Persistence of the transferred T-cell clonotypes correlated with cancer regression37. Eleven of 13 responding patients had ≥5% persistence of transferred clonotypes compared with only 1 of 12 non-responders (P = 0.001). In accord with this, the telomere length of the infused TIL correlated with cancer regression38. The mean telomere length in TIL administered to responding patients was 6.3 kb compared with 4.9 kb in TIL given to non-responders (P < 0.01). TIL clonotypes that persisted in vivo had mean telomeres of 6.2 kb compared with 4.5 kb in non-persisting clonotypes (P < 0.001). Telomere length is related to the proliferative history of the cell. Each time the cell divides telomeres at the ends of chromosomes shorten. Thus cells with longer telomeres have a greater proliferative potential.

In contrast to murine models, the antigen-reactive administered TIL were of effector phenotype, CD27⁻,CD28⁻,CD45RA⁻,CD62L⁻,CCR7⁻ (REF. 39). By 2 months after transfer, however, circulating tetramer-positive cells had a less differentiated phenotype and were CD27⁺,CD28⁺,CDRA⁺,IL7R⁻ but remained CD62L⁻,CCR7⁻, suggesting that highly activated tumour-reactive cells in culture can re-express many of these differentiation markers in vivo39. Re-expression of CD27 in TIL is an indication of a less differentiated cell and, following removal from IL2 in vitro, correlates with the effectiveness of these cells in vivo. These findings are in accord with studies in HIV-positive patients showing that CD27 expression promotes the in vivo survival of administered T cells40.

The high rate of cancer regression seen in these trials convincingly demonstrates the anti-tumour efficacy of ACT therapy and the importance of transferring cells with a high degree of antigen recognition and a high proliferative potential. Other issues of importance that might account for the lack of response in some patients relate to the traffic of the cells to draining lymph nodes or to poor expression of antigens by the tumour. Opportunities for improving ACT for patients with cancer based on the genetic modification of T cells are considered later in this Review.

Other examples of effective ACT therapy

ACT using gene-modified lymphocytes

The success of ACT for the treatment of patients with metastatic melanoma has formed a foundation on which to build improvements of this approach. TIL with high avidity for tumour antigens can only be generated from some patients with melanoma and a need exists for the generation of T cells with broad reactivity against shared cancer-associated antigens present on multiple tumour types. The ability to introduce genes into circulating human lymphocytes provides the flexibility to introduce antigen receptors as well as molecules that can provide the cell with enhanced properties required for effective ACT therapy (reviewed in REF 55,REF 56). Genes encoding TCRs can be isolated from high avidity T cells that recognize cancer antigens and retroviral or lentiviral vectors can be used to redirect lymphocyte specificity to these cancer antigens57–60. Mispairing of the inserted chains with endogeneous TCR chains can be greatly reduced by insertion of murine constant region sequences or by insertion of cystine residues that favour pairing of only the transduced chains61,62. High-affinity TCRs can be obtained from rare reactive human clones or from transgenic mouse cells following immunization against human cancer antigens, which thus avoids the tolerance that can limit the generation of these cells in the cancer patient. High-affinity TCR against a p53 epitope63–65 and against carcinoembryonic antigen (M. R. Parkhurst, personal communication) that is present on common epithelial cancers have been generated in transgenic mice and used to redirect the specificity of human lymphocytes. Phage display techniques have been used to generate TCRs with 10⁶× the affinity of a natural TCR directed against the cancer–testis antigen NY-ESO-1, which is expressed on many common cancers66. Chimeric TCR that use the combining site of antibodies genetically fused to intracellular T-cell signalling chains such as CD3ζ can redirect the recognition specificity of lymphocytes to cell surface tumour-associated antigens and thus avoid the limitations of MHC restriction imposed by the use of α–β TCRs67–71.

The steps involved in this process are shown in FIG. 3. High-avidity T cells that are reactive with tumour antigens are identified in the human (often after extensive in vitro sensitization) or from transgenic mice immunized with human cancer antigens. The genes encoding the TCRs from these T cells are cloned and inserted into retroviruses. Retroviral supernatants are then generated under good manufacturing practice conditions that enable their use in humans. These retroviruses can be used to transduce human T cells that express the receptor and can be expanded in vitro for infusion into cancer patients.

The first clinical trial to successfully mediate the regression of human cancer by ACT using genetically engineered autologous lymphocytes has recently been published4. Sixteen patients were treated with a TCR that was reactive with the MART1 melanoma antigen isolated from highly reactive TIL. Two patients with metastatic melanoma who received ACT of their autologous normal lymphocytes transduced with genes encoding this MART1 TCR underwent regression of liver and lung hilum metastases respectively and both are currently disease free over 2 years later. Two of 15 additional patients subsequently experienced objective tumour regressions (S.A.R., unpublished observation). TCRs with far greater affinity for the MART1 melanoma antigen have been identified and are now being evaluated in clinical gene therapy trials72. TCRs are now available against a broad array of cancer antigens present on common epithelial cancers and we have recently begun a trial treating patients with epithelial cancers using autologous T cells transduced with a TCR that recognizes a p53 epitope65 (FIG. 4).

As mouse models clearly indicated that increased lymphodepletion could improve the efficacy of ACT12 we are performing clinical trials in which patients receive non-myeloablative chemotherapy plus either 2 Gy or 12 Gy whole body irradiation. Other improvements in ACT are being studied (see TABLE 2 for selected examples). Murine models have shown that CD4⁺CD25⁺ FOX3⁺ regulatory T cells can inhibit ACT and thus the regimen can be modified to deplete CD4⁺ cells or selectively eliminate T-regulatory cells. Lymphocyte function may be improved using antibodies or genetic approaches that block inhibitory signals on lymphocytes such as CTLA4, PD-1 or TGFβ. Increased persistence and function of the transferred cells may be accomplished by the administration of alternative cytokines such as IL15, or by stimulating the cells in vivo by the administration of a vaccine or by activating host APCs with Toll-like receptor agonists.

The future of ACT

In contrast to common epithelial cancers, melanoma appears to be a tumour that naturally gives rise to anti-tumour T cells. However, other cancers are equally susceptible as the targets of reactive T cells. The susceptibility of melanoma to ACT provides optimism for the application of ACT to common epithelial cancers using TCR gene-modified lymphocytes.

A major problem with the application of ACT is that it is a highly personalized treatment and does not easily fit into current modes of oncological practice. The treatment is labour-intensive and requires laboratory expertise. In essence, a new reagent is created for each patient and this patient-specific nature of the treatment makes it difficult to commercialize. Pharmaceutical and biotechnology companies seek off-the-shelf drugs, easy to produce, vial and administer. From a regulatory standpoint, ACT might be more appropriately delivered as a service rather than as a ‘drug’. Blood banks have been instrumental in providing CD34⁺ haematopoietic stem cells for clinical studies and might be the ideal location for the generation of the anti-tumour T cells needed for ACT.

As modern science increasingly provides the physician with sophisticated information about the unique aspects of an individual cancer, changes in the modes of care delivery need to accommodate this. The ability to use this patient-specific information can lead to a new era of personalized medicine in which individual treatments, such as ACT, are devised for each patient.

Studies of ACT have clearly demonstrated that the administration of highly avid anti-tumour T cells directed against a suitable target can mediate the regression of large, vascularized, metastatic cancers in humans and provide guiding principles as well as encouragement for the further development of immunotherapy for the treatment of patients with cancer.