Adoptive Transfer of Invariant NKT Cells as Immunotherapy for Advanced Melanoma: a Phase 1 Clinical Trial

Abstract

Purpose

Invariant NKT cells (iNKT) are innate-like CD1d-restricted T cells with immunoregulatory activity in diseases including cancer. iNKT from advanced cancer patients can have reversible defects including IFN-gamma production, and iNKT IFN-gamma production may stratify for survival. Previous clinical trials using iNKT cell activating ligand alpha-galactosylceramide have shown responses. Therefore, a phase 1 clinical trial was performed of autologous in vitro expanded iNKT cells in stage IIIB-IV melanoma.

Experimental Design

Residual iNKT cells (<0.05% of patient PBMC) were purified from autologous leukapheresis product using an antibody against the iNKT cell receptor linked to magnetic microbeads. iNKT cells were then expanded with CD3 mAb and IL-2 in vitro to obtain up to ~10⁹ cells.

Results

Expanded iNKT cells produced IFN-gamma, but limited or undetectable IL-4 or IL-10. Three iNKT infusions each were completed on 9 patients, and produced only grade 1–2 toxicities. The 4^th patient onward received systemic GM-CSF with their second and third infusions. Increased numbers of iNKT cells were seen in PBMC after some infusions, particularly when GM-CSF was also given. IFN-gamma responses to alpha-galactosylceramide were increased in PBMC from some patients after infusions, and DTH responses to Candida increased in 5/8 evaluated patients. Three patients have died, three were progression-free at 53, 60 and 65 months, three received further treatment and were alive at 61, 81, and 85 months. There was no clear correlation between outcome and immune parameters.

Conclusions

Autologous in vitro expanded iNKT cells are a feasible and safe therapy, producing Th1-like responses with anti-tumor potential.

INTRODUCTION

Invariant natural killer T (iNKT) cells constitute a lymphocyte lineage with anti-tumor potential that displays a highly restricted T cell antigen receptor (TCR) repertoire, in humans consisting of a specific Vα24-Jα18 chain rearrangement preferentially paired with a Vβ11 chain (1–4). iNKT cells recognize antigen in the context of the non-polymorphic MHC class 1-like CD1d antigen-presenting molecule, and are characterized by their capacity to rapidly produce large amounts of immunoregulatory cytokines. iNKT cells have been shown to play crucial roles in various model immune responses, including the regulation of antitumor immune responses. T helper (Th) 1 cytokine production by iNKT cells (e.g. IFNγ) is important in the initiation of antitumor immune responses, in tumor immune surveillance, in inhibiting angiogenesis, and in mediating the antitumor effects of IL-12 (1–4). Importantly, human pre-clinical studies have found the iNKT cell population to be selectively and functionally, but reversibly, defective and decreased in size in cancer patients (5–12). Decreases in circulating iNKT cells are found in a wide range of cancers (9–12). Decreased iNKT cell numbers are accompanied by reversible functional defects, such as decreased proliferation and IFNγ production by iNKT cells in advanced prostate and other solid tumors, myelodysplastic syndrome, and myeloma patients, resulting in a detrimental Th2 cytokine profile of iNKT cells in advanced cancer (5–12). Unlike conventional T cells, the presence of circulating iNKT cells capable of making IFNγ has positive prognostic value for survival in advanced cancer patients (9–14).

Pre-clinical murine models have shown similar defects and that iNKT cell stimulation or adoptive transfer can induce strong antitumor immune responses (15–22). iNKT cells were shown to be essential for mediating the anti-tumor effects of low and moderate-dose IL-12 therapy in models (16,20,21). iNKT cells are highly and specifically activated by the glycolipid α-galactosylceramide (α-GalCer), originally isolated from marine sponges in a screen for antitumor agents (22). α-GalCer stimulates iNKT cells to rapidly produce a cytokine ‘storm’ including large amounts of IFNγ, that stimulates NK cells, B cells, and that also enhances the generation of classical cytotoxic T cell responses (1–4;16,19,21). Strong antitumor immune responses to α-GalCer have been observed in murine models, including colon carcinoma, lymphomas, sarcoma, melanoma, prostate, and lung carcinoma (1–4;16–22). Together, these observations indicate that restoration of iNKT cell function in humans with cancer may stimulate potent antitumor immune responses.

In the first related phase 1 clinical study, administration of α-GalCer in advanced-stage cancer patients was not accompanied by dose limiting toxicity (23). As in other analyses (9–12), circulating iNKT cell numbers were found to be decreased in these cancer patients (23). The relevance of the decreased size of the iNKT cell pool was demonstrated, as immunological responses to α-GalCer administration (increases in serum GM-CSF and TNF-α levels) were only observed in patients with at least normal iNKT cell levels (23). These initial studies implied that antitumor effects of α-GalCer in cancer patients could be limited by both qualitative and quantitative defects in iNKT cells, necessitating the evaluation of alternative approaches to exploit this natural antitumor system.

In mice, administration of α-GalCer-loaded dendritic cells (DCs) resulted in a more powerful antitumor immune response than α-Galcer alone (17,18). Hence, a number of phase 1 clinical studies used α-GalCer-loaded monocyte-derived ‘MoDC’ or other PBMC-derived antigen-presenting cell (APC) preparations, leading to clinically-relevant antitumor responses with activation of iNKT cells, activation of T and NK cells, and enhanced NK cell cytotoxicity (12,13;24–29). Chang et al. evaluated the effects of intravenous administration of purified α-GalCer-pulsed matured monocyte-derived DCs in 5 patients with cancer (26). They observed more than 100-fold expansion of circulating iNKT cell numbers in all 5 patients, sustained for up to 6 months post-vaccination. This was apparently associated with enhanced adaptive T cell immunity, as it was accompanied by an increase in memory CD8⁺ T cells. No more than grade 1 toxicity was observed. Although one patient developed rheumatoid factor and transient positive antinuclear antibody, no clinical evidence of autoimmunity was observed (26).

Several further trials have used APC (e.g. adherent PBMC treated with GM-CSF and IL-2) loaded with α-GalCer and shown increasing evidence of effectiveness as dose, in vivo targeting, and combinations have been improved (12,13;27–29). Finally, another previously tested approach entailed the adoptive transfer of activated iNKT cells to restore iNKT cell numbers in cancer patients. This approach has been tested in preclinical models of melanoma and lung cancer and shown to be more effective compared to the i.v. administration of α-GalCer (19). Trials of iNKT-enriched PBMC, with or without α-GalCer-pulsed matured monocyte-derived DC, have supported direct use of iNKT cells, with evidence for immunological and objective clinical responses (30–32).

Since we and others have demonstrated that iNKT of cancer patients can be expanded and functionally restored in vitro (5–9), we chose a complementary therapeutic approach using adoptive transfer of in vitro expanded autologous iNKT cells to restore iNKT cell numbers and activity. iNKT cells were isolated after leukopheresis by a protocol based on a monoclonal antibody that specifically recognizes the invariant TCR of iNKT cells (33,34), and were then expanded over several weeks in vitro. We evaluated the toxicity of these adoptively transferred autologous iNKT cells as well as their capacity to induce immunologic responses in cancer patients in a phase 1 clinical trial. All patients received three infusions of autologous in vitro expanded iNKT (up to 250 million cells/infusion) spaced 2 weeks apart. Since iNKT cells are activated via interaction with CD1d on APC, after the first 3 patients suffered no significant toxicities, subsequent patients were pre-treated with GM-CSF to enhance DC functions with iNKT cycles 2 and 3. This study treated patients with advanced melanoma.

MATERIALS AND METHODS

Reagents

Reagents, including iNKT-specific mAb 6B11, mock and CD1d C1R transfectants and their use in iNKT cell manipulation have been described (5,33–36). Commercial FACS antibodies were from eBioScience, Inc., except TCR mAbs including Vα24 and Vβ11 were from Coulter. Pure 6B11 mAb was biotinylated with Pierce/Endogen NHS-LC-biotin (lot #95022864; 2mg/ml in dimethyl formamide), as per manufacturer’s recommendations. GMP anti-biotin magnetic beads were from Miltenyi Biotec, Inc. Purified iNKT cells were stimulated with OKT3 CD3 mAb (Ortho. Immune, Inc.) and irradiated autologous PBMC ‘feeders’ as described (33–36). T cell media was as follows: RPMI-1640, 5% Human AB Serum (HAB), additional amino-acids, β-mercaptoethanol, antibiotics, and 100 U/mL IL-2 (ProLeukin) for expansion, 20 U/ml for assays.

Study design and treatment

This study was designed to assess the feasibility of purifying and in vitro expanding iNKT cells from cancer patients, and to assess whether the expanded cells could be administered safely. The study protocol (Figure 1A,B) was approved by the Dana Farber/Harvard Cancer Center Institutional Review Board, and all patients signed informed consent. Inclusion criteria included advanced stage melanoma (III-IV), age 18 or over, and predicted life span of 6 months or more to complete the study and follow-up. Exclusion criteria included active systemic infection, positive HIV, HBV (unless previously immunized), or HCV serology, or immune deficiency disease; autoimmune disease requiring systemic therapy with immunosuppressive agents; and known hypersensitivity to GM-CSF or DMSO. Patient demographics and status on entry to trial are shown in Table S1.

Figure 1. Overview of iNKT trial and iNKT cell enumeration.

(A) Simplified iNKT preparation protocol overview. (B) Clinical Trial Protocol Flow Chart. (C) Analysis of patient leukapheresis product PBMC iNKT by 3 methods, Vα24-Vβ11 antibodies, Vα24-6B11 antibodies, and αGalcer-analogue loaded-CD1d tetramer, shown with Vβ11 antibody. Representative patient #1. Last panel on the right shows negative control unloaded-CD1d tetramer for comparison.

Patients underwent leukapheresis to collect circulating iNKT cells, which were isolated using a monoclonal antibody (6B11) against the invariant TCRα chain expressed by iNKT cells (5,33,34) and expanded in culture over 6–8 weeks in the Connell/O`Reilly Cellular Manipulation Core Facility (CMCF) at Dana-Farber Cancer Institute (DFCI). The expanded cells were harvested and divided into 3 aliquots and frozen for subsequent intravenous infusion every 2 weeks. Thawed cells were assessed for viability (>90%) and purity (minimum Vα24 positive cells >5%), and were administered by a 15-minute intravenous infusion. The first 3 patients (patients 1, 2, and 4, since patient 3’s infusions were delayed and overtaken by patient 4) did not receive additional cytokine, while the subsequent 6 patients were also treated with GM-CSF (250 μg/M²/day subcutaneously for 10 days) from the 2^nd and 3^rd administration of iNKT cells. Vital signs were monitored following the completion of the infusion, and blood was collected before, during, and after treatment for clinical and immune monitoring. The latter included determining the frequencies and functions of various leukocyte subsets and serum cytokine levels. Delayed type hypersensitivity reaction was measured by standard Candida skin prick test before starting therapy and after completion of infusions.

iNKT cell product preparation and qualification

Approximately 10⁹ PBMCs were isolated by leukapheresis and iNKT cells were then isolated using a scaled up version of a previously described method (36,37). Cells were labeled with biotinylated 6B11 monoclonal antibody, followed by addition of anti-biotin microbeads and application to a magnetic column (CliniMACS) to separate iNKT cells. Purified iNKT cells were then cultured in 24-well culture plates pre-coated with mitogenic human therapeutic grade anti-CD3 antibody (OKT3) in RPMI-1640 medium supplemented with 5% Human AB Serum (HAB), L-glutamine, gentamicin, essential and non-essential aminoacids, and β-mercaptoethanol for approximately 6–8 weeks. Autologous irradiated PBMC feeder cells were added to cultures during weeks one and four of culture. During the culture period, medium was replenished weekly and cultures expanded from single wells to as many as possible (maintained at approximately 2 × 10⁶/well). The iNKT cells were then harvested and cryopreserved after an aliquot had been removed for immunophenotyping and functional testing, and for sterility, endotoxin and mycoplasma testing while the remaining aliquots were kept frozen.

The purity of the iNKT cell product was assessed by staining with combinations of Vα24, Vβ11, and 6B11 antibodies (33–36), and with αGalcer-analogue loaded-CD1d tetramers (NIH Tetramer Facility, Emory, GA), used as described (37,38). Functional status of the iNKT cell product was assessed by determining the cytokine profile (IL-4, IL-10 and IFNγ) after in vitro polyclonal stimulation, and after CD1d-specific stimulation, as described (5,33–36). To assess cytokine production after the in vitro expansion by FACS, thawed cells were rested overnight and stimulated with phorbol myristic acid and ionomycin for 4 hours before cytokine capture staining (Miltenyi Inc.) and analysis by 5-color flow cytometry for iNKT subsets (33,34). To assess CD1D-specific responses, the cells from the iNKT cell cultures were incubated 1:1 with CD1d transfected C1R cells and phorbol myristic acid (PMA; 1 ng/ml) or α-GalCer (100 ng/ml). ELISA was performed on culture supernatants harvested at the peak response time, which was day 1 for IL-4 and day 3 for IFNγ (Endogen Inc.).

Viability after thawing (>90%) and percent Vα24⁺ (>5%) were used as release criteria, and the thawed products were also re-tested for sterility and endotoxin. On the last two patients, under an approved amendment, the first infusions were given using freshly-harvested cells that were not frozen.

Immune monitoring

Levels of iNKT cells and other immune cells from patient blood were assessed by 5-color flow cytometry with panels of mAb for the various lineages, as described (39–43). Means ± SD shown.

Serum antibodies against melanoma antigens

Pre- and post-treatment serum IgG were bound to protein-G Sepharose, chemically cross-linked with DSP (Pierce/Endogen), and briefly acid-strip washed to remove noncovalently coupled IgG and prevent subsequent loss from beads in sample buffer and contamination of the gels. Cell surface and whole cell lysates from several melanoma cell lines (ATCC) were biotinylated with NHS-LC-biotin (Pierce/Endogen), as per manufacturer’s recommendations, and immunoprecipitated with each pair of pre- and post-treatment IgG. Precipitates were eluted under non-reducing conditions and analyzed by SDS-PAGE and streptavidin-HRP western blotting. Patient sera were also tested for specific IgG reactivity against a panel of known melanoma antigens by ELISA, as previously described (41).

Statistical Analysis

Means and standard deviations of triplicate samples shown, differences were considered significant for p < 0.05. Standard Kaplan-Meier analysis was applied.

RESULTS

iNKT cell isolation and in vitro expansion

All 9 patients accrued to the study underwent leukapheresis, from which iNKT cells were successfully isolated and expanded in culture (Figure 1A,B). Cells harvested for iNKT cells were characterized by FACS analysis before and after purification and the in vitro expansion (Figure 1C). iNKT cells can be identified based on their use of a unique Vα24-Jα18 encoded TCR that is recognized by the 6B11 mAb. Most iNKT cells co-express Vβ11, but not all Vα24/Vβ11 T cells are iNKT cells, and a subset of iNKT cells may express other TCRβ chains (1–4;35,38). Therefore, several reagent combinations were used to comprehensively monitor iNKT cells. These included co-staining with Vα24 and Vβ11 Abs, 6B11 and Vβ11 Abs (Figure 1C). Finally, αGalcer-analogue loaded-CD1d tetramer identifies high affinity αGalcer-reactive T cells restricted by CD1d that are chiefly, although again not exclusively, Vα24⁺Jα18⁺ TCR⁺ (37,38) (Figure 1C). As negative control, unloaded tetramer was used, as also shown in Figure 1C. Therefore, double-positive cells in each case were used to estimate the size of the corresponding iNKT populations.

The iNKT frequencies as assessed by each method in the leukapheresis products prior to purification were typically comparable (6B11⁺ ≥ αGalcer-CD1d tetramer⁺ ~ Vα24⁺/Vβ11⁺), and yielded a frequency of from ~0.02 to 0.20% iNKT cells/total T cells (Figure 1C; not shown). iNKT cell purifications resulted in a marked 100 – 1000-fold enrichment, and iNKT cell purity post-purification (measured as % Vα24⁺ T cells) varied from 4% to 73 % (median 47%) (Table 1). Based on estimates of total iNKT cells in the leukapheresis product and numbers after purification, losses of iNKT cells appeared to be negligible (not shown). The purity of iNKT cells after the in vitro expansion was 13 – 87% (median 66 %). Comparing cells pre and post-expansion, iNKT cell purity was increased after expansion in 4 patients, decreased in 1 patient, and unchanged in 3 patients (Table 1). The total number of iNKT cells after expansion ranged from 1.1 × 10⁷ to 1.26 × 10⁹, and up to 1/3 of these were used per infusion (Table 1). Numbers of iNKT cells infused each cycle were therefore from 4 – 217 × 10⁶ (median 7.3 × 10⁶). iNKT cell CD4 expression was also assessed pre- and post in vitro expansion and varied considerably (from 5 – 77 % in the post expansion product; median 70%) (Table 1).

Table 1. Summary of iNKT cell recovery and purity pre- and post-expansion.

The iNKT product was characterized as described in M&M.

Subject	Pre-Expansion % iNKT	Post-Expansion % iNKT	% CD4+ iNKT	Total iNKT	iNKT/Infusion
001	54	54	74	5.88E+8	1.35E+08
002	73	66	5	2.24E+8	7.19E+07
003	19	28	72	1.13E+07	3.60E+06
004	70	72	47	5.83E+07	2.49E+07
005	47	66	70	2.46E+8	1.53E+08
006	31	74	50	9.89E+8	1.85E+08
007	4	13	72	1.36E+8	2.73E+07
008	21	87	21	1.26E+9	2.17E+08
009	62	30	77	2.70E+8	7.28E+07

Characterization of expanded iNKT cells and pre-infusion cell product

To assess cytokine production after the in vitro expansion, multi-color flow cytometry of cytokines among CD4⁺ and CD4⁻ iNKT cells (identified with Vα24 antibody) was performed. As we observed previously following the expansion protocol (33,34), substantial fractions of both the CD4⁺ and CD4⁻ iNKT cells produced IFNγ (Table 2 and Figure S1). There were no consistent differences in CD4⁺ and CD4⁻ iNKT cells with respect to IFNγ production. Fewer cells expressed IL-4, and the expression of IL-10 was detected in only one patient (Table 2).

Table 2. Functional activity of the iNKT product.

Functional activity of the iNKT product was characterized as described in M&M. NA: not applicable (no significant CD4⁺ population); ND: not detectable; NT, not tested.

Subject	%IFNγ+ CD4-iNKT	IFNγ+ CD4-iNKT MFI	% IFNγ+ CD4+ iNKT	IFNγ+ CD4+ iNKT MFI	%IL-4+ iNKT (MFI)	%IL-10+ iNKT (MFI)
001	52	10	43	6	NT	ND
002	52	10	NA	NA	ND	ND
003	60	3	70	10	15 (1.2)	NT
004	21	3.5	22	20	NT	ND
005	68	3.3	61	3.3	ND	ND
006	18	2.4	13	2.6	10 (2.6)	7 (2.5)
007	11	2.0	24	0.7	17 (0.8)	ND
008	NT	NT	NT	NT	NT	NT
009	55	1.5	30	2.5	NT	NT

Cytokine release in response to αGalcer presented by CD1d⁺ autologous APC was also assessed by ELISA in 4 patients. In patients 4 and 5, αGalcer induced IFNγ responses were equivalent to ~20% of mitogen (Table 2 and Figure S1). In patients 6 and 7 we were able to assess IFNγ and IL-4 production in isolated CD4⁺ versus CD4⁻ subsets. Comparable levels of IFNγ were produced by both subsets, while IL-4 production was greater in the CD4⁺ subset (Table 2 and Figure S1). These results show that the in vitro expanded iNKT cells are functional and produce IFNγ and IL-4 at levels comparable to healthy donor iNKT cells, and suggest a Th1-bias in the CD4⁻ population that is also seen in healthy donors (1–4;34).

Effects of iNKT cell infusion on PBMC populations

PBMC harvested prior to, during, and after treatment were characterized by flow cytometry. Gating strategies on a representative case are shown in Figure S2, and analysis of iNKT cell numbers for each patient throughout treatment are shown in Figure 2A. Baseline levels (prior to the first iNKT cell infusion) were generally low (median ~0.02%), which is below those observed in healthy controls (median ~ 0.05%) and comparable to the low levels reported in other cancer patients (9–13;39). iNKT cell proportions remained at or below normal levels during the 3 cycles of infusion for the first 3 patients infused (patients 1, 2, and 4) (Figure 2A). However, numbers were modestly increased on cycle 2 and notably increased on cycle 3 and in follow-up for patient 3, the fourth patient treated (who received GM-CSF with their 2^nd and 3^rd infusions). Subsequent patient’s circulating iNKT cells were also generally increased after their 2^nd and 3^rd infusions (given with GM-CSF). Patient 8 was unusual in peaking at high levels in all 3 cycles and follow-up, and patient 6 peaked above normal levels only in cycle 1. Significantly, peak size did not correlate with input, since patients 1, 2 and 6 received higher numbers of iNKT cells per infusion while patients 3 and 7 received amongst the lowest levels. Overall, iNKT infusions appeared to cause transient peaks of circulating iNKT cells that were enhanced by GM-CSF (although we cannot exclude mobilization of endogenous iNKT cells by GM-CSF). As expected due to their tissue-homing nature (5–8;15–17), circulating iNKT cell levels returned close to baseline at follow-up.

Figure 2. Patient PBMC iNKT and other immune cell levels through infusions.

Immune cells were monitored before, during and after iNKT infusions, as shown using gating in figure S2. (A) iNKT levels (Vα24+6B11+). (B) Dendritic cells (lineage-negative CD45+ DR+). (C) Activated CD4 T cells (CD69+). (D) Treg (CD3+CD4+CD25+ FoxP3+). (E) NK cell levels (CD56+CD3-negative). (F) Proportion of activated (CD69+) NK cells. (G) CD14+ monocyte levels. (H) Activated (CD69+) monocytes.

Characterization of other immune cell populations before, during, and following treatment did not reveal any marked consistent alterations (Figure 2 and Figure S3,4). We particularly focused on whether the iNKT cell infusions would increase circulating DCs (lineage-negative, DR⁺), which can express CD1d (39,40), but their levels were generally not altered during the therapy (Figure 2B). As is normal, myeloid DC were the most common of the DC and approximately half of these expressed maturation markers CD80/86 (data not shown). CD1d⁺ CD14⁺ monocytes also remained present at normal frequencies of 5 – 10 % (data not shown). T cells examined included activated T cells (CD69⁺ CD25^lo) (Figure 2C) and CD25^hi Treg cells, the latter confirmed to express FoxP3 (Figure 2D). Expression of the transient activation marker CD69 on CD4⁺ T cells appeared to respond to each cycle of infusion, with most patients peaking at about 0.5 – 1 % above baseline, and returning to baseline in between treatment cycles (Figure 2C). Activated CD69⁺ CD4⁺ T cells were more prominent within cycles 2 and 3. Since this applied to patient 2, who did not receive GM-CSF, this is likely to be in response to the iNKT cells. Activation declined in follow up. CD4⁺CD25⁺FoxP3⁺ Treg were found at frequencies ~2 – 6% at baseline and generally rose, at least during cycles 2 and 3 (including in patients 2 and 4, so again apparently GM-CSF independent), although declining to baseline or below in follow-up (Figure 2D). Although apparently responding to infusions, no consistent effects were noted among B cells or activated B cells (Figure S5).

NK cells were quantified in PBMC (Figure 2E). Of note, patient circulating NK cell levels remained approximately stable or spiked, but these were largely different patients to those whose iNKT clearly spiked (Figure 2E). This distinct behavior of circulating iNKT and NK cells is reminiscent of responses to high dose IL-2, where levels of iNKT and NK were approximately reciprocal (39). Activated NK cells (CD69+) were also measured (Figure 2F). Briefly, the fraction of activated NK cells did not change much on first iNKT infusions, but tended to spike over first few hours after second infusions (including patient #2, 1 of the 3 patients who did not receive GM-CSF on second and third infusions). There was somewhat less response of NK cells to the third infusions (Figure 2F). Patient #5, who had progressive disease and was withdrawn from study, had elevated levels of several populations including notably and progressively activated NK cells.

Finally, CD14+ monocytes and the activated fraction of these monocytes were also quantified (Figures 2G, H). Monocyte levels tended to rise and fall after especially the second and third infusions (Figure 2G). There was considerable variation between patients and within each patient over time among activated monocytes. Four patients had lower levels at the pre-bleed which increased up to nearly 100% in response to infusions and which remained high in 3 of 4 patients (Figure 2H). Those that had high levels at outset remained high or higher by the end of follow-up, despite spikes during infusions (Figure 2H).

In summary, iNKT cells (Figure 2A) and a range of other immune cells in peripheral blood responded to iNKT infusions with changes in numbers and activation state, including generally increased numbers of activated monocytes (Figure 2H). Most of the common changes in circulating cells lasted for a week or 2 after infusions before returning to approximately baseline levels for the majority of measurements.

Cytokine production by PBMC in response to iNKT cell stimulation

To assess whether the infusions had effects on iNKT cell function in vivo, we examined IFNγ production after stimulation with αGalCer in PBMC obtained at baseline and after completion of therapy (at least one month after the 3^rd infusion). The PBMCs were stored viably frozen, and aliquots were thawed and tested in parallel. Both ELISPOT and ELISA methods were used to measure cytokine production. We initially performed IFNγ ELISPOT assays, and then collected the supernatants to measure IL-4 and IL-10 by ELISA. Elevated levels of IFNγ production were seen after treatment in 5/8 patients tested (Figure 3A). By ELISA we found that αGalCer weakly stimulated the production of IL-4 in pre- and post-therapy PBMC’s from most patients (Figure 3B). However, the basal levels and fold-induction were generally comparable in pre- and post-therapy samples, with only patient 9 showing a significantly greater induction in the post-therapy sample. Basal IL-10 production also was comparable in most cases, but fold induction by αGalCer was increased in several of the post-therapy cases, with this increase being significant in patients 3 and 8 (Figure 3C). There were no increases in GM-CSF, TNFα, or MIP1α in response to αGalCer stimulation (Figure S5). Overall these results indicated that iNKT cell function was increased post-therapy in a subset of patients, with a more consistent increase in Th1 cytokine production.

Figure 3. Patient PBMC iNKT cytokine responses pre- and post-treatment.

(A) Patient PBMC EliSpot responses to iNKT stimulation pre- and post-treatment. IFNγ spot forming cells in response to α-GalCer were detected by standard EliSpot overnight.

(B, C). Patient PBMC were assayed by EliSpot for iNKT cell responses to stimulation pre- and post-treatment as in (A). IL-4 and IL-10 cytokine production in the EliSpot supernatants in response to α-GalCer then were detected by standard ELISA. Supernatants from the EliSpot assay were lost, so there were no ELISA for patient #2.

Serum antibody responses to melanoma antigens

To determine whether iNKT cell infusions lead to B cell responses, we assessed melanoma cell lysates that were immunoprecipitated with patient serum IgG pre- and post-treatment. Pre- and post-treatment serum IgG were bound to protein-G Sepharose and chemically cross-linked to prevent subsequent loss from beads (and therefore contamination of the gels). Cell surface-biotinylated whole cell lysates from several melanoma cell lines were immunoprecipitated with each pair of pre- and post-treatment IgG. Precipitates were eluted under non-reducing conditions and analyzed by SDS-PAGE and streptavidin-HRP Western blotting. In one patient (patient 2) we found that post-, but not pre-treatment serum IgG, specifically precipitated an ~46 KDa protein that was expressed by several melanoma cell lines (Figure S6A).

Patient sera were also tested for specific IgG reactivity against a panel of known melanoma antigens by ELISA, as previously described (40). Increased reactivity in post-treatment serum was observed in patient 2 (for Angiopoietin-2, ~66 KDa), in patient 5 for Angiopoietin-1 and NY-ESO-1, in patient 7 for Angiopoietin-1, and in patient 1 for Progranulin (Figure S6B). Interestingly, patients 2 and 3 had the highest levels of baseline circulating dendritic cells (Figure 2B), which were maintained throughout the iNKT cell infusions of patient 2, who also had the largest fraction of CD4⁻ versus CD4⁺ iNKT cells, which could be related to the antibody responses (Figure S6A,B) in this patient.

Patient treatment and outcomes

Patients had metastatic melanoma, but with minimal disease or no evident disease at enrollment (Table S1). All patients received the planned 3 iNKT infusions, each spaced 2 weeks apart, with vital signs, symptoms, and blood collected before, at 1 and 4 hours post infusion, day 1, day 7, and day 14, as well as follow-ups. Patients received from 1.3 × 10⁶ to 2.5 × 10⁸ total T cells. With purity varying as described above, this translated into infusions of as low as 4 x10⁶ (patient 3) to 2.4 × 10⁸ iNKT cells (patient 8) per infusion (median 7.3 × 10⁶; mean 9.1 × 10⁶ iNKT cells).

Toxicities attributed to treatment were limited to Grades 1 and 2 (Tables S2 and S3). Some patients experienced low-grade fevers immediately after infusion despite pretreatment with acetaminophen and diphenhydramine, but treatment was still well tolerated (Table S2). Patients 3 and 5–9 also received GM-CSF prior to iNKT cycles 2 and 3 and experienced mild expected constitutional symptoms and localized injection site reactions attributable to GM-CSF. With patients 8 and 9, the first infusions were given using freshly-harvested cells that were not frozen. Again, no difference was seen in toxicity. Therefore, no treatment adjustments were necessary during the course of the trial. Tables SA1-5 list all treatment related and unrelated toxicities.

Change in delayed type hypersensitivity reaction was measured by Candida skin prick tests. Candida DTH increased by at least 1 measure with 5/8 evaluable patients post-NKT infusions (Table S4). Time on Study, Progression and Survival data were updated at completion of study in April 3, 2015. Median follow-up for this patient cohort was 63 months (53 to 85 months). As of 04/03/15, three patients had died, and three had disease progression. All patients who died also had disease progression. Disease-specific data gathered during treatment and follow-up are summarized in Tables S5 and S6, and in Figure 4. Patient 5 was found to be progressing during treatment and died of disease (DOD) within a month of treatment completion. This was judged to be unrelated to treatment, so no hold was put on the trial. Otherwise, the patients whose outcome in the short-term post-treatment was progression of disease were 4 and 6. Those who did not progress while on follow-up were patients 1 (53 months follow-up), 7 (65 months follow-up), and 8 (60 months follow-up). Those who began stably after treatment, although eventually progressed and moved to other treatments were patients 2 (11 months TTP), 3 (18 months TTP) and 9 (31 months TTP).

Figure 4. Patient time to progression and overall survival data.

(A) Patient time to progression. (B) Patient overall survival.

DISCUSSION

Initial cancer clinical trials using systemically administered αGalcer or αGalcer-loaded in vitro-generated dendritic cells demonstrated some immunological responses, and most recent trials have exploited αGalcer-loaded PBMC-derived matured antigen-presenting cells and/or PBMC cultured with αGalcer to enrich for iNKT (11–13;26–32). In combination, these latter approaches have led to some objective clinical responses (13,26–32), as well as immunological activity. However, low numbers of iNKT cells in cancer patients, and defects in their ability to produce Th1 cytokines, may limit the efficacy of these approaches. We previously developed a mAb (6B11) that is specific for the invariant TCRα chain that defines iNKT cells, and found that it could be used to enrich for these cells (33,34). Therefore, this study was carried out to assess the feasibility of purifying and expanding in vitro large numbers of iNKT cells, to determine whether the expanded cells would produce Th1 cytokines, and to assess whether the autologous expanded cells could be safely infused. We confirmed that this approach is feasible and safe, although further studies are clearly needed to assess for efficacy.

Recent progress in cancer immunotherapy has demonstrated proof-of-principle and curative potential (44–49). Compared to tumor-infiltrating T lymphocyte (TIL) trials where TILs are expanded in vitro from resections and re-infused at high doses following conditioning regimens (50–53), fewer iNKT cells were transferred in the current trial. We speculated that transfer of relatively rested iNKT cells may allow them to persist in the circulation, and that by acting as immunoregulatory cells (rather than direct effector cells) they may have immunological effects even at low levels. However, based on blood sampling, iNKT post-infusion appeared to circulate only transiently, and we do not know the extent to which they marginated successfully into tissues or were cleared by the spleen. Moreover, while suggestive increases in IFNγ responses were detect in post-treatment PBMCs in a subset of patients, the extent to which this can be attributed to the iNKT cell infusions remains uncertain.

It is possible that higher doses might have more effect in vivo, but the need for conditioning to create immunological ‘space’ might then become a factor. Interestingly, there appeared to be a transient rise in circulating FoxP3⁺ Treg-type cells in the majority of patients, something which has been a concern in TIL and iNKT cell therapy (11–13;47,49–51), which is at least partly addressed by some conditioning regimens (47,49–51). Most recently, in a murine model it was shown that CD4⁺ iNKT cells (present in most iNKT infusions in this trial) induce Treg which protect against lethal graft versus host disease, although potent graft versus tumor activity was retained (54). In any case, it is also possible that CD4⁻ iNKT cells may drive more robust Th1 responses and provide more effective anti-tumor activities. Indeed, patient 2 had the highest induction of anti-melanoma antibodies and was an outlier with only ~5% CD4⁺ iNKT cells in their infusion product.

Overall, iNKT cell therapy for advanced melanoma was feasible, well tolerated, produced evidence of increased iNKT cell activity and consequent activation of other immune cells. Infusion of more cells, and/or enriching for CD4⁻ iNKT cells are possible future directions. Combination with other therapies that may enhance iNKT cell activity in vivo (such as αGalcer or IL-12), or synergize with iNKT cells in driving T cell activity (such as check point blockers) are also options to consider. Finally, iNKT cells expressing a chimeric antigen receptor have been explored and may provide more potent anti-tumor activity (55).

STATEMENT OF TRANSLATIONAL RELEVANCE.

Invariant NKT cells (iNKT cells) can contribute to protective anti-tumor immune responses. iNKT cells are reduced in cancer patients and have defects in Th1-cytokine production, but these defects are reversible during in vitro expansion. Therefore, a phase 1 clinical trial of infusion with autologous in vitro expanded iNKT cells was initiated in advanced melanoma to determine safety and feasibility. Autologous iNKT cells could be expanded in vivo and produced IFNγ, suggesting potential for improved anti-tumor activity. Three planned infusions were completed for 9 patients, and GM-CSF was added with the 2^nd and 3^rd infusion in the last 6 patients, with no major toxicities. There was evidence for systemic T cell and myeloid cell activation, and patients showed variable rates of progression. iNKT cell therapy is feasible and has potential for anti-tumor effects.