Successful ex vivo expansion of tumor infiltrating lymphocytes with systemic chemotherapy prior to surgical resection

Tumor infiltrating lymphocytes (TIL) have demonstrated efficacious clinical outcomes for many patients with various types of solid cancers, including melanoma, gastrointestinal cancer, lung cancer, and head and neck cancer. Currently, the majority of clinical trials require that patients did not receive systemic therapy right before tumor tissue resection to avoid the interference of chemotherapy in the ex vivo TIL expansion. The primary disadvantage of this strategy is limiting the accessibility of TIL therapy for many eligible cancer patients. Over the past decade, substantial progress has been made for ex vivo expansion technologies in T cells. In this study, we investigated the possibility of enrolling patients who underwent chemotherapy prior to surgical resection. We collected seventeen tumor tissues from treatment naive cases, and five from cases that underwent chemotherapies. Cancer indications enrolled in this study were colorectal and lung cancers from both primary and metastatic sites, such as liver and brain. TILs from these tumors were expanded ex vivo to 2.1E8 (total viable lymphocytes counts) on average, with an overall success rate of 90.9%. Subsequently, TIL phenotypes and cytokine production were analyzed using flow cytometry and ELISA, respectively. We demonstrated functional TIL expansion from tumor tissues despite chemotherapy prior to surgical resection. We observed no significant phenotypic or functional differences between groups with and without chemotherapy. TIL expansion rate and characteristics were similar regardless of chemotherapy prior to resection, thereby providing a possibility to recruit patients with the most recent chemotherapy history in TIL therapy trials.


Introduction
Adoptive cell therapy (ACT) has emerged as one of the most promising cancer immunotherapies [1][2][3]. One such ACT, tumor infiltrating lymphocytes (TIL) therapy has been widely studied and has exhibited favorable clinical outcomes for solid tumors [4][5][6]. For example, a recent study by Chesney et al. achieved an objective response rate (ORR) of 36% in 66 advanced melanoma patients, with tumor progression after immune checkpoint inhibitors (ICI) and targeted therapies [4]. A two arm, randomized phase 3 trial study by Rohaan et al. showed a median overall survival (OS) of 25.8 months in melanoma patients receiving TIL, in contrast to an OS of 18.9 months in the 2nd group of patients treated with ipilimumab (anti-CTLA-4), demonstrating a strikingly better efficacy with TIL than with ICI [5,6].
One of the biggest challenges in recruiting patients for immunotherapy clinical trials has been a history of chemotherapy prior to enrollment. Given that chemotherapy is one of the standards of care in cancer, exclusion of patients who have received chemotherapy from a TIL therapy trial would greatly impact patients due to lack of treatment options. A study by Aydin et al. shows that TIL can be successfully expanded from a rare indication, penile cancer regardless of neoadjuvant therapy status or HPV status of the patient [7]. However, successful generation of TILs from tumor tissues following systemic therapies in other types of solid cancer has remained elusive. We hypothesized that chemotherapy would not significantly impact ex vivo TIL generation. Thus, we procured tumor tissues from patients with or without chemotherapy and examined TILs from these two groups.
Here we report the effective generation and expansion of

Patient samples
Tissue samples were provided by the cooperative human tissue network (CHTN (RRID: SCR_004446), https:// www. chtn. org/) funded by the National Cancer Institute. Other investigators may have received specimens from the same subjects. The tissues were obtained as either anonymized, de-identified, or a limited data set. The de-identified pathological reports were provided for each sample. Sixteen treatment naïve cases and five cases with treatment were included in this study. All cases with available treatment information were analyzed in this study due to the sample availability. This indicates that the statistical calculations for grouping were not conducted. The patient demographics and baseline characteristics are listed in Table 1.

Generation of TILs from a tumor tissue
A resected tumor tissue was stored in HypoThermosol FRS (StemCell Technologies) with 0.1% gentamicin sulfate (Gibco) and 1% amphotericin B (R&D Systems) and kept at 4 °C until processed. TILs were expanded as previously described [8]. Briefly, for the initiation phase, tumors were dissected into < 2 mm 3 fragments and cultured with AIM-V media (Gibco) containing 10 µg/mL gentamicin sulfate, 25 mM HEPES (Lonza), 2 mM l-glutamine (Gibco), 10% human AB serum (access cell culture) supplemented with 6000 IU/mL of rhIL-2 (R&D systems). Half media changes were performed every 3-4 days with fresh culture media with IL-2 (final concentration 6000 IU/mL) and cultured in 37 °C and 5% CO 2 until day 14. On Day 14, TILs were collected and counted and then proceeded to the REP. TILs were co-cultured with feeder cells which were irradiated peripheral blood mononuclear cells (PBMCs) from healthy donors at a ratio TIL:PBMC of 1:100 in AIM-V media with 5% Human AB serum, 2 mM l-glutamine, 25 mM HEPES, 30 ng/mL of anti-CD3 (OKT3 clone, ThermoFisher Scientific) and final concentration 3000 IU/mL rhIL-2. Half media changes were performed every 3-4 days with fresh media with IL-2 (3000 IU/mL) for 14 days. On day 28, generated TILs were harvested, counted, and downstream assays were conducted, or cryopreserved in LN2 till used. Since TIL expansion in this study was not performed in a closed bioreactor or scaled up compared to other studies that report significantly higher fold expansion. Furthermore, lymphocytes infiltration into tumor tissue greatly varies among patients, indications, stage, and tumor types. We have limited access to patients' pathological evaluation which disable to define absolute infiltrated lymphocytes in the tissues. Therefore, success of TIL generation in this study was defined as fold expansion ≥ 2 at the end of rapid expansion protocol (REP). . Live/dead fixable yellow dead cell stain kit (Ther-moFisher Scientific, Cat# L34959) was used to exclude dead cells from the analysis. Samples were fixed with 2% paraformaldehyde. Functional analysis of TILs: single cell suspensions were incubated with Protein Transport Inhibitor Cocktail (eBioscience, Cat# 00-4980-03), anti-CD107a-PE-Cy7 (BD Biosciences, Cat# 561348) with/without cell stimulation cocktail (eBioscience, Cat# 00-4970-03) for 4 h at 37 °C. After 4 h incubation, cells were washed, and stained with a master mix of Abs for surface stains after blocking with 10 μL of FcR blocking reagent (Miltenyi Biotec) for 10 min at RT. The following Abs were used: anti-CD3-FITC (Biolegend), anti-CD4-PerCP (Biolegend), anti-CD8-V500 (BD Biosciences), anti-CCR7-APC (Biolegend, Cat# 353214), anti-CD45RA-APC-H7 (Biolegend, Cat# 304128). After incubation, cells were permeabilized with BD cytofix/ cytoperm (BD Biosciences, Cat# 554722) for 20 min on ice, washed twice with BD perm/wash buffer (BD Biosciences, Cat# 554723), then intracellular cytokines were stained with anti-IFN-γ-BV421 (BD, Cat# 562988) and TNF-α-PE (Bio-Legend, Cat# 502909). Live/dead fixable yellow dead cell stain kit (ThermoFisher Scientific, Cat# L34959) was used to exclude dead cells from the analysis.

Characterization of generated TILs using flow cytometry
All samples were acquired using LSRFortessa (BD) and analyzed with FlowJo software (TreeStar).

Statistical analysis
We performed the Wilcoxon rank sum test to compare the features between the group with chemotherapy and the group without chemotherapy. P values less than 0.05 were considered to be statistically significant. Principal component analysis (PCA) was performed using a truncated singular value decomposition method [9] implemented in the R package of irlba (version 2.3.5). Pearson correlation coefficients were calculated using the R function, cor, from the package of stats (version 4.0.3). The heatmaps with hierarchical trees are drawn using the R function, pheatmap in the package of pheatmap (version 1.0.12) using euclidean distance metric and complete linkage rule. Spearman's correlation coefficient analysis was used to explore the association between TIL numbers and treatment days prior to resection. The P values are computed using algorithm AS 89 [10] Results

Pre-surgery chemotherapy shows no impact on in vitro TIL generation and expansion
A study demonstrated that active TILs were successfully expanded from 78% of patients with advanced metastatic melanoma despite prior chemotherapy history [11] and another recent study revealed successful TIL generation from patients with penile cancer [7]. However, there has been no evidence of treatment effect on TIL generation in other indications. Therefore, it remains unclear whether TIL can be initiated from patient samples with other cancer types after chemotherapy. To verify this, we generated TILs from tumor tissues with 17 colorectal and 5 lungs, including 9 metastasized site tissues, such as liver and brain ( Table 1). The tissue weight varied from 50 to 2560 mg. Therefore, we normalized the number of TIL per mg of tissue to be able to compare samples. First, we sought to determine the correlation between the number of TILs generated and the time period between chemotherapy and surgical resection. We observed a trend of negative correlation between TIL numbers and the length of time after chemotherapy with spearman correlation coefficient = − 0.7, although it was not statistically significant (Supplemental Fig. 1). We then examined the cell numbers and expansion rate with and without chemotherapy. The average number of TIL per mg *F: female, M: male #Removed from downstream analysis due to lack of in vitro TIL expansion of tumor tissue was 6.2E + 5 cells for patients with chemotherapy prior to resection and 4.9E + 5 for patients without chemotherapy before surgery (Fig. 1A). The mean and SD of expansion rate of the TILs that were generated from tissues with and without chemotherapy were 23 ± 20 times and 54 ± 74 times, respectively (Fig. 1B). These generated TILs proceeded to REP and were subsequently used for further analyses.

TILs display similar phenotypic profiles regardless of pre-surgery chemotherapy
Next, we performed flow cytometric analysis to further characterize TILs derived from chemotherapy treated and untreated samples. Expanded TILs exhibited predominantly CD3+ T cells, with average of 94.6% with chemotherapy and 85.6% without chemotherapy ( Fig. 2A). In the CD3+ T cell populations, CD4+ T cells were dominant with an average of 68.2% and 65.3%, treatment+ and treatment naive, respectively (Fig. 2B) while CD8+ T cells were 27.4% and 25.6% with and without chemotherapy respectively (Fig. 2C). A study has shown that TILs with stem-like phenotype (CD8+CD39−CD69-) were associated with cancer regression and that CD39+ CD69+ TILs have poor persistence [12]. We examined our TILs and found that there were no significant differences observed in CD39/CD69 double staining between the chemotherapy and no chemotherapy groups (18.2% vs. 18.6% for the CD39-CD69-, 32.9% vs 25.6% for the CD39+ CD69+ with and without chemotherapy respectively, Fig. 2D). We then examined the T cell activation marker 4-1BB which was equally expressed on TILs with or without chemotherapy, both in CD4+ (15.4% vs. 27.4%) and CD8+ (17.7% vs. 21.6%) populations (Fig. 2E). Programmed cell death receptor 1 (PD-1) is highly expressed on tumor specific T cells and plays a vital role in immune responses [13,14]. We found that the PD-1 expression of our TILs was comparable in CD4+ (14.4% with chemotherapy vs. 23% without chemotherapy, Fig. 2F) and CD8+ populations (5.8% with chemotherapy and 11.3% without chemotherapy, Fig. 2F). Similarly, T cell immunoreceptor with immunoglobulin and ITIM domain (TIGIT) which is upregulated by immune cells and expressed by activated CD8+ T and CD4+ T cells [15], is a promising new target for immunotherapy. Our results indicated that it was expressed in 63.3% and 65.1% of CD4+ TILs with and without chemotherapy respectively. TIGIT expression in CD8+ T cell population demonstrated a similarly unremarkable profile of 57.9% with chemotherapy and 61.6% without chemotherapy (Fig. 2G). Interestingly, we found that the vast majority of TILs were T effector memory cells (Tem: CD45RA-CCR7-, Fig. 2K), and other phenotypes were T central memory (Tcm: CD45RA-CCR7+, Fig. 2J), Naïvelike-phenotype T cells (CD45RA+ CCR7+, Fig. 2I), and Terminally Differentiated Effector Memory (TEMRA: CD45RA+ CCR7−) T cells (Fig. 2H). Finally, we confirmed TILs with and without chemotherapy were undistinguishable when combining the effects of all surface marker expressions. PCA was performed to search for a multi-variable pattern based on the expression of 10 surface markers. Figure 3 shows that TILs with and without chemotherapy are not separable by top 2 principal components (PCs) that explain 64.1% of total variances. Pearson correlation coefficients between each pair of surface marker expressions are shown in Supplemental Fig. 2A.

TILs exhibit comparable cytokine producing capacity irrespective of chemotherapy status
We confirmed that chemotherapy did not change TIL phenotypes and then sought to examine their functionality. The generated TILs were harvested after REP (on day 28) and stimulated with anti-CD3/anti-CD28 coated beads for 4 h in the presence of protein transport Inhibitor and anti-CD107a antibody. Anti-CD107a was used as a marker for degranulation. As expected, significant amounts of IFN-γ,TNF-α., and CD107a were detected (n = 5 for the group "with chemotherapy" and n = 11 for the group "without chemotherapy"). There was no statistically significant difference between chemotherapy and no chemotherapy group in both CD4+ and CD8+ populations (Fig. 4A). To confirm this, the production of IFN-γ was quantified using ELISA (n = 4 for the group "with chemotherapy" and n = 6 for the group "without chemotherapy"). The mean and SD of secreted IFN-γ with and without chemotherapy were 6989 ± 4017 pg/mL and 12,379 ± 10,328 pg/mL, respectively (Fig. 4B). Hence, we confirmed that systemic treatment did not change TIL functionality.

Discussion
In this study, we demonstrated successful TIL generation and expansion from various tissue types of patients with colon cancer types who received chemotherapy as little as 14 days before surgical resection. These TILs possessed phenotypes and functionalities similar to treatment naïve TILs. Exceptionally, we included metastatic cancers as well as primary cancers from multiple indications and found that functional TILs could be expanded from multiple indications in both primary and metastatic sites regardless of the status of chemotherapy prior to surgery. This suggests that patients with metastasized sites as well as patients who received chemotherapy prior to surgical resection may benefit from enrolling in TIL adoptive transfer therapy trials. However, it is important to note the clinical efficacy of such adoptive TIL transfers is yet to be determined and not guaranteed.
One of the crucial factors for the success of TIL therapy is TIL phenotype, especially the CD8+: CD4+ ratio and their ability to recognize and kill tumor cells. CD8+ T cells are known as cytotoxic T cells and play an important role in tumor killing; however, CD4+ TIL have a supportive role that is essential for antitumor properties [15,16]. The balance between CD8+ and CD4+ TILs could be a key factor in their persistence and their cytotoxic effects in the tumor microenvironment. It remains unclear in the prevailing literature as to which TILs are the most important against solid tumors. Krishna et al. showed that TILs with stemlike phenotype (CD39 CD69-CD8 +) were associated with cancer regression and CD39+ CD69+ TIL had poor persistence [12]; den Bulk and others indicated the importance of CD39+ CD103+ as neoantigen-specific cytotoxic T cells [17,18]. This will need to be further investigated with clinical trial samples and correlation with their clinical outcome.
To further develop TIL therapy, a study demonstrated that TIL generation in the presence of PD1 antibody (Nivolumab) or with 4-1BB antibody (Urelumab), and CD4+ T cell depletion improved TIL yield and tumor reactivity [19].Similarly, a recent study proved that the addition of Urelumab exhibited significant enhancement in cell expansion and elevated CD8+ TIL in the product [20] Additionally, the engineered immunocytokine, which is a fusion protein of IL-15 mutein and a PD-1 specific antibody (anti-hPD1-IL15m) demonstrated robustly enhanced the proliferation, activation, and cytotoxicity of CD8+ and CD4+ TILs from human primary cancers in vitro, whereas regulatory T cells were largely unaffected [21]. They also reported that anti-mPD1-IL15m demonstrated potent antitumor efficacy without exacerbating body weight loss in vivo mouse model [21].
Our study has some limitations. More work is warranted to determine the effect of chemotherapy on TIL generation. For example, procuring equal number of tissues in both groups (with and without chemotherapy) would enhance the power of this study. Despite this limitation, our study indicates that TILs can be generated and expanded from cancer patients who have received chemotherapy prior to surgical resection and that the TILs generated can exhibit immune characteristics similar to those generated from treatment-naive patients. These findings highlight a possibility to extend inclusion criteria to enroll cancer patients who underwent chemotherapy and patients who have unresectable primary cancer, but a resectable metastasized site for TIL therapy. Fig. 4 Analysis of TIL capability to produce cytokines A percentage of the secretion of CD107a, IFN-γ and TNF-α in the CD4+ T and CD8+ T cell populations. Values from the unstimulated cells were subtracted from the stimulated as background for this analysis (n = 5: with chemotherapy, n = 11: without chemotherapy). B Secreted IFN-γ was quantified by ELISA from the TILs with or without chemotherapy prior to surgery (n = 4, with chemotherapy; n = 6 without chemotherapy). The Wilcoxon rank sum test was performed. n. s.: not significant