Achievements and challenges of adoptive T cell therapy with tumor-infiltrating or blood-derived lymphocytes for metastatic melanoma: what is needed to achieve standard of care?

Abstract

Adoptive cell therapy (ACT) based on autologous T cell derived either from tumor as tumor-infiltrating lymphocytes (TILs) or from peripheral blood is developing as a key area of future personalized cancer therapy. TIL-based ACT is defined as the infusion of T cells harvested from autologous fresh tumor tissues after ex vivo activation and extensive expansion. TIL-based ACT has so far only been tested in smaller phase I/II studies, but these studies consistently confirm an impressive clinical response rate of up to 50 % in metastatic melanoma including a significant proportion of patients with durable complete tumor eradication. These remarkable results justify the need for a definitive phase III trial documenting the efficacy of this type of T cell-based Advanced Therapy Medicinal Product in order to pave the way for regulatory approval and implementation of TIL therapy as a new treatment standard in oncology practice. TIL-based ACT can, however, only be offered to a limited group of patients based on the need for accessible tumor tissue, the complexity of TIL production procedures, and the very intensive nature of this three-step treatment including both high-dose chemotherapy and interleukin-2 in addition to T cell infusion. To this end, adoptive T cell therapy using peripheral blood mononuclear cell-derived T cells could be a welcome alternative to circumvent these limitations and broaden up the applicability of ACT. Here, we discuss current initiatives in this focused research review.

Introduction

During the last decade, the field of cancer therapy has been transformed by the development of new and effective treatment modalities with a major focus on immune therapeutic strategies. Monoclonal antibodies not directly attacking the tumor cells but instead modulating immune response mechanisms to release anti-tumor cytotoxic activity have been developed, e.g., by blocking suppressive immune check point components such as CTLA-4 and PD-1 [1]. Alongside these therapeutics being developed by the pharmaceutical industry, a huge effort from non-commercial research groups around the world has been made to develop highly individualized types of cellular cancer therapies, particularly those based on transfer of autologous T cells.

Transfer of immune cells such as tumor-infiltrating lymphocytes (TILs) has on occasion proven effective especially for metastatic melanoma (MM), where this form of immunotherapy holds the promise to markedly change the very poor long-term prognosis of those patients, showing response rates similar to those obtained with the immune check point antibodies.

MM is the most aggressive and deadliest form of skin cancer with around 20,000 persons dying from MM every year in Europe alone. More traditional approaches such as chemotherapy and radiation have failed to impact survival of MM patients significantly, and only recently the first targeted therapy (vemurafenib targeting the BRAF V600 mutation present in 50 % of MM) was approved based on improved progression-free and overall survival as compared to dacarbazine [2]. The capacity of BRAF inhibitors to significantly influence long-term prognosis of MM patients is, however, doubtful. Thus, being highly resistant to traditional treatment strategy and with an average life expectancy of less than 1 year following diagnosis and treatment, MM has been a focus for development of alternative immune strategies. Besides the successful immune check point antibodies, different forms of adoptive cellular immunotherapies for MM patients have been exploited in the past few years. Among these, adoptive cell therapy (ACT) based on the infusion of ex vivo expanded TILs has been the most often applied and most successful, but also ACT based on expanded peripheral blood-derived T cells has been pursued.

TIL-based ACT is defined as the infusion of T cells isolated from autologous tumor tissue after ex vivo activation and multiple rounds of expansion taking advantage of the enriched numbers of tumor-specific T cells which can be found at the tumor site [3]. Hence, a new T cell product is established for each individual patient by employing a standard two-step production protocol [4]. The final cell infusion product consists of a polyclonal T cell population comprising a highly variable amount and diversity of tumor-specific T cells.

In vitro reactivity against autologous tumors can be detected in TILs from the majority of patients with melanoma including both CD8⁺ and CD4⁺ tumor-reactive T cells [4–6]. By culturing these naturally occurring tumor-reactive TILs ex vivo in the presence of factors supporting activation and proliferation, such as anti-CD3 antibody and interleukin-2 (IL-2), a renewed activation state of the TILs is achieved. In addition, expansion of the population leads to around 1,000-fold to 5,000-fold increase in their number, creating a large amount of highly activated tumor-specific T cells.

The final cell product is infused iv after 1 week of intensive lymphodepleting chemotherapy (cyclophosphamide and fludarabine) and followed by treatment with high iv doses of the T cell growth factor IL-2. Clinical efficacy is achieved when migration of tumor-reactive T cells to the metastatic sites occurs, leading to tumor cell killing due to a broad and patient-specific recognition of both defined and undefined antigens presented by a variety of major histocompatibility complex (MHC) antigens.

TIL-based ACT for treatment of disseminated melanoma

The immunogenicity of MM has been recognized for many years with an established clinical efficacy of high-dose IL-2 therapy in a small subset of patients (aprox 15 %) including some long-term responders [7]. Researchers have since then explored alternative immunotherapeutic strategies which ultimately led to the development of TIL-based ACT capitalizing on the concept of using a preconditioning non-myeloablative lymphodepleting chemotherapy regimen prior to infusion of TIL and followed by cytokine support therapy using high-dose IL-2 [8, 9]. This kind of TIL therapy has demonstrated consistent efficacy with the potential to induce impressive objective response rates of about 50 % in metastatic melanoma confirmed in multiple phase II single-institution trials [10–14]. Noteworthy, among responding patients prolonged complete tumor eradication in up to 20 % remaining disease-free 7 years or more after treatment has been described [15].

TIL preparation procedures

A major drawback of the methods employed for TIL generation in the earliest TIL therapy trials was the need for selection of TIL cultures from a single or few tumor fragments grown individually based on documented anti-tumor reactivity. Thus, only TIL cultures showing tumor reactivity over a predefined cutoff value were further expanded and infused [4]. This selection procedure constituted a significant problem especially as the ideal target for screening, and the autologous tumor cells in most cases are not available at the time point when they are needed for testing of TIL cultures. Alternative strategies using, e.g., testing with defined tumor antigens or allogeneic melanoma cell lines were employed, but both strategies harbor obvious disadvantages. To this end, it has been shown that TILs displaying no reactivity against allogeneic melanoma cell lines were indeed able to induce tumor regression upon infusion, suggesting that the identification of reactive TILs largely depends on the availability of appropriate target tumor cells. Hence, these limitations in methods for in vitro screening of anti-tumor reactivity could significantly underestimate the actual level of anti-tumor effectiveness [16, 17] and lead to inappropriate exclusion of patients.

Another incorporated drawback of selected TIL manufacturing was that the long overall cell production time of 5–8 weeks when the required minimum number of TILs to initiate the final expansion step was to be grown from a single or a few individual tumor fragments [4, 16, 17]. Inevitably, this prolonged TIL production time led to an unnecessary high patient dropout rate due to clinical deterioration [18] in the early TIL trials. Facing these problems, a modified method of TIL production named ‘Young TILs’ was developed. This new approach abandoned culture selection and instead used a single bulk TIL culture generated from the whole tumor specimen. Thus, all the unselected TILs produced undergo further expansion using the standard rapid expansion protocol (REP) followed by infusion into the patient. This change resulted in a significantly shorted preparation time, increased the number of successful TIL preparations, and thereby reduced the dropout rate significantly [16, 17, 19–21]. In addition to these advantages, the final infusion product of Young TILs contains cells expressing a younger and therefore improved phenotype and also displays high in vitro anti-tumor recognition [16, 22].

Young TIL production consists of two different expansion steps. The initial ‘pre-REP’ phase is of variable duration normally lasting 2–4 weeks during which TIL are expanded using IL-2 to at least 40–50 million cells in total. At this stage, it is possible to cryopreserve the cells for future use. During the final more standardized ‘REP’ phase, pooled T cells are cultured for 2 weeks with anti-CD3, irradiated allogeneic peripheral blood mononuclear cell (PBMC)-derived feeder cells, and IL-2 resulting in a 1.000-fold to 3.000-fold expansion of the T cells. Using this method, it is possible to generate clinical grade TIL products from about 60 % up to more than 90 % of tumor samples [4, 17, 22, 23]. The whole process of young TIL generation takes 4–6 weeks, and typically a total of 2–20 × 10¹⁰ TILs can be produced and infused.

Additional production aspects

TIL manufacturing is indisputably a resource-consuming, technically complex and labor-intensive process which needs to be carried out by a specialized team following cGMP compliant rules. Thus, several initiatives could be appropriate to reduce procedure complexity, labor, and costs.

To this end, new GMP compliant closed systems of TIL manufacturing have recently been introduced [24] including the Wave Bioreactor. Notably, we have observed a considerable increase in expansion rate up to more than 5.000-fold using this system [30].

The above-mentioned possibility to cryopreserve pre-REP TILs offers increased flexibility to the treatment which is of great advantage for logistical planning and, as REP carries the majority of the production costs, for minimization of economic loss due to REP production of TILs for patients who eventually will not be eligible for treatment. As an additional advantage of this approach, pre-REP TIL preparation could be established and cryopreserved for patients with non-progressive disease; TILs could then be thawed upon disease progression and be ready to use within 14 days.

T cell factors associated with clinical response

Exploratory studies performed in the context of TIL-based ACT clinical trials have identified a number of factors which seem to be associated with clinical response. To this end, a ‘younger’ phenotype, including high expression of co-stimulatory receptor molecules (such as CD27) or longer telomere length, and the persistence of TILs in the circulation 1 month after treatment have been recognized as a potential predictive factor for achieving a clinical response [25]. Additional analyses of TILs infused in the setting of clinical trials at the Sheba Medical Center, Israel; the M.D. Anderson Cancer Center, USA; and CCIT in Denmark have shown that the absolute number of TILs infused, and in particular the number of CD8⁺ TILs, is another critical factor influencing the outcome of ACT [14, 21] (CCIT, unpublished results). In addition, our data indicate that it may be even more critical to infuse a large number of in vitro tumor-reactive CD8⁺ TILs (Donia et al., unpublished) [26].

TIL products mainly consist of CD4⁺ and CD8⁺ T cells at highly variable ratios, as well as a very small but consistent fraction of γδ T cells [5]. Most published data suggest that the efficacy of TIL therapy is primarily attributed to CD8⁺ T cells; however, the role of other subsets in TIL products should not be ignored. Several studies have shown that tumor-reactive CD4⁺ T cells can be identified in TILs of some preparations [5, 6], and a smaller number of clinical responders to ACT, including patients achieving long-lasting complete remission, were actually treated with TIL products largely dominated by CD4⁺ T cells [14, 17] (CCIT unpublished). These results may suggest a positive influence in cases where TIL products contain a significant fraction of tumor-reactive CD4⁺ T cells.

The question of IL-2 dosing

The rationale for treatment with high doses of IL-2 after TIL infusion has been to support in vivo TIL growth and activity. This regimen is, however, associated with severe toxicity and is likely to significantly limit the dissemination of TIL therapy to cancer centers outside highly specialized units. It was recently published that an increasing number of IL-2 (high-dose bolus) doses administered after TIL promoted peripheral CD4⁺ T regulatory cell reconstitution, which was negatively associated with clinical response [27].

At the CCIT, at the time of writing, we have treated 25 patients with TIL therapy. In our initial small pilot study, we treated six patients with TILs preceded by standard lymphodepleting chemotherapy but followed by a low-dose regimen of IL-2, consisting of sc IL-2, 2 MIU/day given daily for 14 days. Two of these patients achieved complete and long-lasting responses [12]. Furthermore, in our ongoing phase II trial, despite using an intermediate decrescendo IL-2 schedule (continuous infusions of IL-2: 18 MIU/m2 over 6, then 12, and then 24 h followed by 4.5 MIU/m2 over 24 h for 3 days [28] ), a response rate comparable to what has previously been published with high-dose IL-2 is achieved (CCIT, unpublished data). When changing from a low-dose subcutaneous IL-2 regimen to an intravenous decrescendo IL-2 schedule, we did observe increased toxicity. However, the side effects were still manageable at a conventional Department of Oncology.

Despite the limited number of patients in these trials, they do show that durable complete responses can in fact be induced even with the use of lower doses of IL-2 with an advantageous toxicity profile. Evidently, the role of IL-2 treatment following TIL requires further clarification which probably could lead to dose reduction in the future.

Dissemination of TIL-based ACT to multiple centers

Given the success of TIL-based ACT, several institutions in the US and Europe have shown interest in this treatment, and a handful of highly specialized centers have succeeded in establishing methods for TIL manufacturing and treatment. It can be expected that additional centers will establish TIL treatment in the near future, pointing toward the need for establishing validated standardized manufacturing procedures. Previous comparable studies have promoted the notion that broad application of TIL methods can be achieved through the implementation of a blood banking model, where specialized facilities for TIL production are integrated into existing blood banks at specialized cancer hospitals [29]. In this way, decentralized manufacturing of T cell products according to cGMP compliant rules can be established at the lowest costs in a non-commercial setup in close proximity of the treatment institutions. However, the feasibility of centralized TIL production necessitating longer transportation of the infusion product has not yet been clinically validated. To this end, CCIT has established a practical and simplified protocol of TIL manufacturing using a closed system bioreactor. This protocol will ensure higher applicability of TIL production in laboratory routine, reducing work load while still allowing the generation of high-quality TIL products [30].

Phase III trial for documentation of TIL therapy effectiveness

The present level of technical and clinical development of TIL therapy warrants a pivotal phase III trial, aiming at providing conclusive efficacy data in a randomized and controlled setting, which are required for regulatory approval of this effective treatment strategy. It is crucial and urgent to obtain robust documentation of TIL therapy effectiveness. First of all, TIL therapy seems to offer a clear advantage over other therapeutic options currently available for patients with advanced MM. It is therefore vital to obtain regulatory approval (and thereby reimbursement) of TIL therapy in order to secure dissemination and accessibility of the treatment. Secondly, many new initiatives to further improve TIL therapy are upcoming, leading to an apparent need for testing these strategies against a ‘classical’ and standardized TIL therapy. Moreover, the cost/efficacy ratio of TIL treatment is considerably lower than any other available treatment for MM. Thus, the estimated total cost of TIL therapy is less than 50,000 €/patient, in comparison with a drug cost alone of approximately 100,000 € for a course of anti-CTLA4 antibody therapy. Undoubtedly, this cost level will be maintained or even exceeded by new promising drugs for MM such as anti-PD-1 antibodies. Thus, in the interest of society, cost-effective treatments need to be pushed forward.

‘Classical’ TIL therapy is currently performed in three European academic institutions: CCIT, Denmark; Netherlands Cancer Institute–Anthoni van Leeuwenhoek Hospital, Amsterdam, The Netherlands; and The Christie NHS Foundation Trust, Manchester, UK. An alliance between these experienced cancer centers has been established, and TIL production has been fully harmonized. Based on this, a randomized controlled trial is being established and has most recently been approved by the EU Heads of Medicines Agencies. In this multicenter phase III trial, patients with advanced MM will be randomization between TIL-based ACT and ipilimumab, non-blinded, as first- or second-line treatment. Ipilimumab has been chosen as the comparator based on the fact that it is an approved drug implemented as standard of care for MM patients as first- or second-line treatment with a well-described short and long-term efficacy.

A major objection to reported response rates from previous smaller TIL therapy trials has been the unclear data management in regard to patients enrolled in the studies who eventually did not receive the treatment due to either deterioration while waiting for treatment or insufficient TIL growth. It is therefore pointed out that true response rates taking the intention-to-treat population into consideration would be significantly lower. Evaluation of results of TIL transfer studies performed in four different institutions that did report dropout rates shows that on average one-third of the enrolled patients are not treated with TIL resulting in a response rate for enrolled patients that is 9–15 % lower than the response rate obtained when only treated patients are considered (Table 1, [12, 13, 31–33]). The majority of the dropouts are caused by rapid disease progression or failure to culture (enough) TIL for treatment (60 and 30 % of all dropouts, respectively). To this end, it is important to stress that recent method optimization allows specialized centers to generate clinical grade T cell products from up to 90 % of resected tumor samples [17, 18, 20, 23]. At CCIT, it is our experience that TIL can be successfully generated from more than 90 % of the patients using the simplified procedure described above. Also, introduction of the Young TIL procedure has significantly reduced dropout rates due to reduced production time and no requirement for TIL selection. The planned phase III trial will be based on intention-to-treat data analysis.

Table 1.

Objective response rates on intention-to-treat basis and causes for patient dropout

					Dropout n (% of total dropouts)
Reference	TIL production	Culture time (week)	Enrolled	Treated (%)	PD or sBM	No TIL	Cr-SAE	Other	OR (n)	% OR enrolled	% OR treated
10	Selected	6–8	Unknown	43 (?)	n.r.	n.r.	n.r.	n.r.	21	n.r.	49
13	Selected	7	19	13 (68 %)	4 (67 %)	1 (17 %)	1 (17 %)	None	5	26	38
12	Selected	5–7	11	6 (55 %)	4 (80 %)	1 (20 %)	None	None	2	18	33
32	Unselected young*	7	101	69 (68 %)	15 (47 %)#	17 (53 %)	None	None	19	19	28
33	Unselected young	2–4	80	57 (71 %)	11 (48 %)	8 (35 %)	1 (4 %)	3 (13 %)	23	29	40

n.r. not reported, PD or sBM rapid disease progression or appearance of symptomatic brain metastasis, cr-SAE lymphodepleting conditioning regimen-related SAE

^*TIL in both cohorts were unselected ‘young’ TIL either used as such or after enrichment of CD8⁺ cells

^#includes patients lost for follow-up as well as patients with rapid disease progression

ACT based on peripheral blood-derived lymphocytes

In addition to TIL, ACT can also be performed using tumor-specific T cells that are obtained from PBMC, thus bypassing the need to acquire surgically removed tumor tissue. Tumor Antigen (Ag)-specific T cells can be expanded from PBMC after repeated in vitro stimulation with autologous whole tumor cells, or specific peptides that can be used as such or after loading onto autologous dendritic cells. Optionally, these enriched Ag-specific T cell populations can be cloned or further purified, e.g., by using peptide-HLA-multimer-coupled beads [34]. Thus, well-defined T cells can easily be obtained and have been shown to exert clinical benefit in a number of trials [35–37].

A major drawback of transferring T cells with defined but restricted Ag specificity is the enormous heterogeneity and plasticity of tumor cells with respect to expression of the targeted antigen. Indeed, loss of targeted Ag expression has been described in residual or relapsing tumors in non-responder patients or even in patients with a mixed response after treatment with gp100- or Melan-A/Mart-1-specific T cells [35, 38]. Loss of targeted Ag seems to occur more frequently when the targeted Ag is a differentiation Ag, probably because expression of these Ags is not essential for the oncogenic phenotype. Down-regulated expression of melanoma differentiation Ags is not a genetic deficit but an epigenetic process. Paradoxically, interferon-γ (IFN-γ), which is produced by activated tumor-infiltrating effector T cells, is capable of down-regulating the expression of differentiation Ags [39]. In contrast, expression can be enhanced by stimulation of tumor cells with MEK or BRAF inhibitors [40] arguing in favor of combined or sequential BRAF inhibitor and T cell transfer protocols. The observation of Ag loss or down-regulation in progressing or relapsing tumor has emphasized the need to use T cells specific for either multiple different antigens that are unlikely to be down-regulated simultaneously, or T cells specific for gene products essential for tumor cell proliferation and oncogenic phenotype.

In order to identify tumor Ags that are involved in durable regression observed after TIL-based ACT, which could be suitable targets for selecting and in vitro expanding the most effective T cells, the specificities of TIL products was investigated using an elegant approach allowing the parallel detection of Ag-specific T cells against a large array of antigenic peptides [41]. Reactivity against overexpressed antigens was rare, whereas reactivity against differentiation antigens (especially MART-1 and gp100) was common but did not correlate with clinical response [42]. In contrast, CD8⁺ T cell responses against cancer-testis antigens, albeit present in low frequencies, did correlate with clinical response, suggesting that preferential ex vivo expansion of these T cells might improve outcome. Overall, only a very small percentage of responding TIL populations displayed specificity for any of the tested previously identified melanoma antigens. These data correspond with our own observations that T cells obtained after repeated in vitro stimulations of PBMC with autologous irradiated tumor cells prior to ACT resulting in durable partial or complete responses or disease stabilization [43] contain only very few T cells with specificity for any of the tested large panel of known melanoma antigens (unpublished data from the LUMC), whereas they consist of up-to 85 % of tumor-reactive CD8+ T cells. These data suggest that the majority of the tumor-specific T cells are targeted against mutated antigens that are highly variable between patients and therefore considered ‘personal.’ Interestingly, the dominant T cell populations in TIL products that induced durable tumor regression in melanoma patients were recently also demonstrated to be specific for mutated/neo-antigens [44–46], suggesting that this may be a more general phenomenon. In this respect, autologous tumor cells would be the ideal source for stimulation of PBMC providing all possible personal/neo-epitopes capable of inducing potent tumor-specific T cells. However, establishing an autologous tumor cell line is time-consuming and not possible for all patients. Therefore, extensive evaluation of T cell specificities determining tumor control are needed to select the proper (mix of) antigen(s) used for ex vivo isolation and expansion for effective PBMC-based ACT. The rapidly developing technical possibilities to identify patient-specific mutated genes in tumor biopsies by exome sequencing and subsequent prediction and synthesis of mutated peptide epitopes within a short time frame may provide the means to culture a population of effective T cells for optimized ACT.

Optimizing patient conditioning and T cell-supporting cytokine administration

Persistence of T cells after transfer has been shown to be a factor predicting clinical effectiveness of ‘classical’ TIL therapy comprising lymphodepleting chemotherapy and IL-2. However, the transfer of Ag-specific T cells, allowing the in vivo monitoring of T cell persistence by tetramer (Tm)- staining, revealed that even without a lymphodepleting conditioning regimen, transferred antigen-specific T cells persisted in vivo from 2 up to 8 weeks after transfer (see Table 2, [35–38, 47, 48] ). Persistence was short in the absence of IL-2 but clearly improved when at least low-dose IL-2 was administered after T cell transfer [38]. From these studies, it remains questionable whether chemotherapy-induced lymphodepletion prior to T cell transfer, although clearly improving the clinical effectiveness after TIL transfer [10], is essential. The benefit of lymphodepletion is thought to depend on the selective depletion of suppressive immune cell subsets like Treg, the creation of space for transferred cells and removal of cellular sinks consuming the induced homeostatic cytokines required for in vivo expansion of transferred T cells. Specifically, the latter two aspects may be essential when large numbers of up to 0.5–2 × 10¹¹ TIL are infused but not when almost 100-fold lower numbers of antigen-specific T cells are infused, as is the case in the above-mentioned studies. In this respect, it is noteworthy that in 4 out of 5 studies where 5 × 10⁹ or more T cells are given per infusion, no objective responses were obtained irrespective of prior high dose, low dose, or absence of lymphodepleting conditioning (see Table 2, [38, 47, 49–51] ). In contrast, the objective clinical response obtained after infusion of relatively low numbers (<5 × 10⁹) of tumor-reactive or Ag-specific T cells per infusion without lymphodepleting conditioning is on average 38 % (range 11–100 %, see Table 2, [35–37, 43, 48] ). This indicates that clinical responses can be obtained after infusion of much smaller numbers of T cells requiring no or at least a milder conditioning regimen and reduced supportive care consisting of low dose IL-2 or IFN-α or a combination thereof. Such a treatment schedule will reduce the treatment burden and acute morbidity and potential toxicities related to intensive lymphodepleting conditioning and high-dose IL-2 support, and consequently may improve the patient eligibility. In this respect, the use of conditioning regimens or adjuvants aiming at selective depletion of myeloid cells involved in immunosuppression at the tumor microenvironment level, rather than reduction of total leukocyte numbers, are of interest. The major components of an immunosuppressive tumor environment, including regulatory T cells and myeloid-derived suppressor cells (MDSC), have been described to be elevated and associated with disease progression in melanoma [52]. Low-dose chemotherapeutics including cyclophosphamide, temozolomide and gemcitabine or targeted therapies like sunitinib are reported to reduce the numbers of peripheral as well as tumor-infiltrating Treg and/or reduce peripheral frequencies and suppressive function of MDSC [53, 54]. Thus, the combined use of these types of agents and ACT without additional intensive unselective host lymphodepletion might be sufficient to overcome local immunosuppression and thereby improve the clinical effectiveness of transferred T cells which obviously deserves further exploration.

Table 2.

Overview of conditioning, T cells dose, cytokine support, and clinical response data of ACT trials performed with PBMC-derived T cells

Treatment				Response
Type of T cells	Nr. of infusions mean cell nr./infusion total cell nr. infused	Conditioning/persistence	Support after infusion	N evaluable	CR/PR	SD/MR	PD	Remarks	Reference
TIL and PBMC derivedgp100 CD8+ T cell clones	4 infusions 10.4 × 10e9 total 35x10e9	No conditioning No persistence	No 1st cycle Low- /High-dose IL-2 in 2nd cycle	13	0	2 (HD IL-2)	11	Ag loss NR Autoimm. NR SAE NR	49
PBMC derived MART-1 and gp100 CD8+ T cell clones	4 infusions 5.94 × 10e9* total 24x10e9*	No conditioning Persistence only with IL-2 support	None or increasing low-dose IL-2	10	0	8	2	Ag loss 3×, Autoimm. 1×, SAE None	38
PBMC-derived MART-1 CD8+ T cell lines	3 infusions 0.07 × 10e9 total 0.2 × 0e9	No conditioning Persistence (2 week)	3 × 6 day low-dose IL-2 (3 × 106 U daily)	11	2	2	7	Ag loss 2× Autoimm. None SAE none	35
PBMC-derived Vaccine-induced gp100-enriched CD4+ and CD8+ T cells	1 infusion 66 × 10e9 total 66 × 10e9	Yes† Persistence (≥ 1 month)	High-dose IL-2	9	0	0	9	Ag loss No Autoimm. 2× SAE No	50
PBMC-derived NY-ESO-1 CD4+ T cell clone	1 infusion 5 × 10e9 total 5 × 10e9	No conditioning Persistence (≥ 1 month)	None	1	1	na	na	Ag loss No Autoimm. No SAE No	36
PBMC-derived MART-1 CD8+ T cell clone(s)	1 infusion 0.9 × 10e9 total 0.9 × 10e9	No conditioning Persistence (≥ 1 month)	IL-2 (9 × 106/day 2 × 5 days) +IFN-α (9 × 106 U 3×/week during 1 month)	14	6	1	7	Ag loss NR Autoimm. 1× SAE None	37
PBMC-derived MART-1, gp100, tyrosinase CD8+ T cell clone(s)	2 infusions 18 × 10e9* total 36 × 10e9*	1st inf No conditioning Short persistence 2nd inf Fludarabine Persistence (≥ 1 month)	IL-2 low dose (250.000 U 2×/day during 2 week)	10	0	3	7	Ag loss NR Autoimm. 1× mild SAE None	47
PBMC-derived Tumor-reactive CD4+ and CD8+ T cells	3–6 infusions 0.29 × 10e9 total 1.0 × 10e9	Mild transient with (IFN-α 3x106 U/dy 1-week Persistence (CR, n = 1, ≥1 month)#	IFN-α (3 × 106 U/day during 11 weeks)	10	2	3	5	Ag loss NR Autoimm. 1× SAE no	43
PBMC-derived MART-1 CD8+ T cell clones (skewed Tcm phenotype)	2 infusions 2.29 × 10e9 total 4.32 × 10e9	No conditioning Persistence (≥ 1 month)	None	9	1	4	4	Ag loss NR Autoimm. None SAE None	48
PBMC-derived MART-1, gp100, tyrosinase CD8+ T cell clones	1 infusion 1.8 × 10e10* Total 1.8 × 10e10*	Cyclophosphamide high-dose Persistence (LD IL-2 ≥ 1 month) Persistence (HD IL-2 ≤ 1 month)	Low dose (LD) IL-2 High-dose (HD) IL-2	10	1	5 week 8	4	Ag loss NR Autoimm. 6 × mild SAE in all HD IL-2 n = 3	50

NR not reported, na not applicable

^*3,3x10e9 c/m² corrected to average surface of 1,8 m² per adult healthy person

^†Lymphodepleting conditioning according to TIL regimen NCI, i.e., 2 day cyclophosphamide (60 mg/kg) followed by 5 day fludarabine (25 mg/m²)

^#in CR patient (unpublished data)

Another important alternative to sustain T cell persistence in vivo without the need for conditioning or cytokine support is to obtain antigen-specific T cells with a memory phenotype, e.g., by selection of Ag-specific T cells with a high IL-2 versus IFN-γ cytokine profile [55] or by stimulation of PBMC with specific antigen-loaded artificial-APC expressing the required co-stimulatory molecules plus IL-15 [48]. T cells obtained in this way showed long in vivo persistence, sustained tumor-reactivity, homing to the tumor site, and durable complete response in 1 of 8 treated patients. Five patients received additional anti-CTLA-4 therapy upon disease progression that resulted in partial response (n = 3) or stable disease (n = 2) that coincided with in vivo expansion of transferred T cells. These exciting data imply that adoptive transfer of Ag-specific T cells with a memory phenotype without any conditioning or cytokine support can establish an in vivo anti-tumor memory that can be triggered to induce tumor regression after combination with immune checkpoint-blocking antibodies or probably also with subsequent peptide vaccination.

The future of ACT using in vitro expanded unmodified autologous T cells

Despite remarkable clinical efficacy, with impressive response rates and extraordinarily long survival especially of complete responders, the implementation of TIL-based ACT in current practice has been severely hampered for several reasons: (1) The technical complexity and labor intensiveness of cell production, (2) The toxicity profile of high-dose IL-2 requiring treatment at experienced cancer centers, and (3) The complete lack of investment from the pharmaceutical industry because of the low commercial potential of this personalized therapy.

Regardless of these barriers, a pivotal phase III study is indeed timely as regulatory approval of TIL therapy as an Advanced Therapy Medicinal Product (ATMP) is urgently needed. This will hopefully be achieved within 3–4 years based on the upcoming phase III study. Regulatory approval of TIL therapy will undoubtedly lead to a more rapid dissemination of the techniques necessary for clinical implementation on a routine basis. The initiatives mentioned above aimed at simplifying procedures, and increasing flexibility will facilitate this process despite the technical challenges. With the current level of available technologies, TIL production could, in the future, be offered by clinical institutions (e.g., hospitals, blood banks) using a model similar to that currently used for stem cell transplantation [56].

Recent investigations have indicated that the high doses of IL-2 used as cytokine support therapy in the classical TIL schedule might not be necessary. These promising results are being pursued in new trials and could potentially lead to a significant reduction in toxicity associated with TIL therapy, thereby allowing more patients to be offered this treatment. Another approach to reduce toxicity is to use an IFN-based cytokine support regimen as has been demonstrated at the LUMC using autologous tumor cell-stimulated PBMC-derived T cells for ACT. Interestingly, the use of low-dose IFN-α during 1 week prior to T cell infusion resulted in a transient leukopenia and neutropenia that correlated with response to treatment [43]. These data deserve further exploration since they suggest that the ACT-supportive features of the harsh high-dose IL-2 and lymphodepleting conditioning regimen might be equally well addressed by this far less toxic approach.

Other approaches might also contribute to the improved efficacy of TIL therapy. These strategies could specifically aim at increasing the anti-tumor activity of T cell products or at improving host conditioning for a better in vivo TIL survival and tumor targeting by immune sensitization of tumors. With regard to host conditioning, CCIT has recently shown that IFN may increase tumor recognition by TILs by increasing tumor cell antigen presentation [26]. As an alternative strategy for BRAFV600 mutant melanomas, several preclinical reports have suggested that administration of BRAF inhibitors before T cell infusion may increase the efficacy of adoptively transferred T cells by multiple mechanisms involving tumor immune sensitization as well as improved activity of infused TILs [5, 57–59].

Despite the success of emerging immune checkpoint blockade agents such as anti-CTLA-4 and anti-PD-1/anti-PD-L1 inducing clinical responses in 10–40 % of MM patients [60–62], many patients will fail to benefit from these treatments. Similarly, the high objective response rate obtained with BRAF inhibitors is only of short duration (median 5–7 months) and leads to durable complete responses in a minority (6 %) of patients [63]. So, there will still be a need for additional effective therapies such as TIL- or PBMC-based ACT therapy. Moreover, the combination or sequential treatment with ACT and checkpoint blockade or BRAF inhibitors may potentially lead to further improvement of clinical efficacy [48] and may also provide a window for production of T cells for ACT by preventing ineligibility of patients for ACT treatment due to rapid disease progression.

Abbreviations

ACT

Adoptive cell therapy

Ag

Antigen

ATMP

Advanced therapy medicinal products

CCIT

Center for cancer immune therapy

IL-2

Interleukin-2

IFN-α

Interferon-α

IFN-γ

Interferon-γ

MHC

Major histocompatibility complex

MM

Metastatic melanoma

PBMC

Peripheral blood mononuclear cells

REP

Rapid expansion protocol

TILs

Tumor-infiltrating lymphocytes