Background
Lung cancer is the leading cause of cancer mortality worldwide, with 85% of patients diagnosed with non-small-cell lung cancer (NSCLC), many presenting at an advanced stage of the disease [
1,
2]. For decades, platinum-based chemotherapy was the only available systemic therapy for advanced NSCLC. However, the median survival of chemotherapy-treated patients was a modest 8–10 months [
3]. Immunotherapies, which harness the host immune response to treat cancers have recently generated great excitement in lung cancer treatment, but currently approved therapies, such as immune checkpoint blockades (ICB), are most effective in select patient populations which express high PD-L1 or harbor high tumor mutation burdens and infiltrating immune cells [
4,
5].
Adoptive cellular therapy (ACT) involves isolation and ex vivo expansion of cytotoxic immune cells, with or without genetic modification, for infusion into patients with cancer and may provide a new therapeutic option for patients who are not responsive to standard treatments [
6]. Currently, there are several ACTs proposed for cancer therapy such as tumor-infiltrating lymphocytes (TILs), chimeric antigen receptor (CAR)- or T cell receptor (TCR)-modified T cells and cytotoxic innate lymphocytes such as cytokine-induced killer (CIK) cells, γδ-T or natural killer T cells [
6‐
10]. Despite breakthrough success in targeting B cell leukemia and lymphoma [
11], clinical success for ACT in solid tumors is limited due to challenges in obtaining sufficient numbers of tumor-reactive T cells and the immunosuppressive tumor microenvironment [
12,
13]. The cell surface expressed programmed cell death 1 (PD-1) receptor has been implicated in tumor immune evasion of many cancer types through mediating inhibitory signals upon engagement of its ligand, PD-L1, expressed on tumors [
14‐
17]. Indeed, the success of ICB antibodies targeting this pathway has led to clinical reduction of tumor size and improvements in overall patient survival, but as mentioned above, the response rates remain low [
18‐
21].
Previously we have demonstrated that ex vivo expanded human peripheral blood T cells expressing CD3, without CD4, CD8, and NK T cell marker expression, termed double negative T (DNT) cells, have potent activity against lung cancer and leukemia cells in patient-derived xenograft (PDX) models [
22‐
25]. We have developed protocols allowing for large-scale ex vivo expansion of clinical grade DNT cells and demonstrated that allogeneic DNT cells expanded from healthy donors are able to target a broad range of cancer cells in a donor non-restricted manner in vitro and in PDX models. Moreover, infusion of allogeneic DNT cells did not induce a host-versus-graft reaction nor cause graft versus host disease [
24,
26]. These unique features of DNT cells make them different from conventional T cells and support their potential use as a new “off-the-shelf” ACT for cancers [
26]. Based on these findings, a first-in-human clinical trial using ex vivo expanded DNT cells from healthy donors to treat high risk acute myeloid leukemia has been initiated (NCT03027102).
Whereas the anti-cancer activity of DNT cells has been demonstrated, little is known about the presence of DNT cells in patient lung tumors and how immune checkpoint blockade may regulate them. Here, we show that DNT cells are found amongst TILs of lung cancer patients and express PD-1. We further demonstrate that DNT cell therapy can inhibit the growth of late-stage established lung cancers in xenograft models and that addition of anti-PD-1 therapy further augments DNT cell-mediated anti-tumor function and increases their infiltration into tumor xenografts. Together, these data support the use of DNT cells as adoptive cellular therapy for NSCLC either alone or in combination with anti-PD-1 and show, for the first time, that anti-PD-1 antibody can increase tumor infiltration of adoptively transferred DNT cells.
Discussion
Adoptive cellular therapy based on DNT cells emerges as a promising therapeutic option for hematological and lung malignancies [
22‐
26]. Here we show that adoptive transfer of DNT cells significantly inhibited growth of late-stage lung tumor xenografts and enhanced the survival of recipient mice. Moreover, we show that anti-PD-1 increased the accumulation of cytotoxic DNT cells within tumor xenografts. These results collectively demonstrate the potential of DNT cells to benefit NSCLC patients, particularly those receiving ICB treatment with limited response due to lack of TILs.
Tumor infiltrating CD8
+ and CD4
+ T cells remain an important predictor of patient outcomes and responsiveness to anti-PD-1 therapy, with recent discoveries highlighting a role for TCF7
+CD8
+ T cells in predicting responsiveness [
36‐
39]. However, the role of non-conventional T cells, such as DNT cells, in solid tumor remains largely unexplored. By examining lung cancer patients’ resected lung tissues we found that DNT cells were present within early stage lung adenocarcinoma (Fig.
1 a and b) and exhibited a predominate central/effector memory phenotype (Fig.
1d). Further, whereas we observed no significant difference between conventional CD4 and CD8 T cells in their infiltration, a lower frequency of DNT cells was found infiltrating tumors relative to adjacent and normal tissue, suggesting that the tumor microenvironment may be more hostile to DNT cells (Fig.
1b). Of note, whereas significantly higher frequencies of central memory CD4
+ and CD8
+ T cells were found in cancer tissue relative to adjacent and normal lung tissue (Fig.
1d and e), this was not the case for DNT cells (Fig.
1c). While our attempts to directly measure tumor infiltrating DNT cell cytotoxicity against lung cancer failed due to the limited number of DNT cells available for expansion, indirect evidence from DNT cells expanded amongst total TILs from pancreatic and glioma patient tumors showed intracellular IFNγ and TNFα expression upon stimulation by autologous tumor [
40,
41], suggesting that tumor infiltrating DNT cells are likely to be cytotoxic.
We have previously shown that allogeneic DNT cells do not induce host-versus-graft rejection nor cause graft-versus-host disease [
24,
26]. Consistent with the non-allogeneic nature of DNT cells, peripheral blood DNT cells derived from lung cancer patients exhibited similar cytotoxicity to that of healthy donor derived DNT cells against the same lung cancer cells (Fig.
1f). Additionally, a report showed that lung cancer patients have fewer circulating DNT cells in peripheral blood than healthy donors [
42] and that fewer DNT cells were expanded from lung cancer patients (Additional file
2: Figure S1B). Furthermore, we found that expanded DNT cells exerted a greater cytotoxicity against lung cancer in vitro compared to CD4 and CD8 T cells from the same donor (Additional file
2: Figure S2A). Together, these findings suggest that the use of healthy donor DNT cells is more practical and will make DNT-cell therapy more readily available.
Importantly, DNT cells, but not CD8 T cells, significantly inhibited late-stage H460 lung tumor growth in vivo (Fig.
2a and Additional file
2: Figure S2B) and prolonged survival of tumor bearing mice (Fig.
2b and d). In the case of slow growing patient-derived xenograft cell line, XDC137, DNT cell treatment limited the growth of the tumor for over 70 days of the observation period and DNT cells were found infiltrating the tumor at this time point (Fig.
2e), suggesting that adoptive transfer of DNT cells could lead to a long-lasting anti-tumor immunity. Interestingly, though adoptive cellular therapy shows promise in clinical trials, rarely do preclinical studies show complete tumor regression in xenograft models [
43‐
46]. Similarly, DNT cell therapy significantly inhibited tumor growth but did not eradicate late-stage lung cancer xenografts. This may be due to the lack of other components of the immune system in immunodeficient mice which may not support memory T cell formation or may be due to the immunosuppressive tumor microenvironment [
47].
Given the role of the tumor microenvironment in regulating T cells [
12,
13], we found that tumor-infiltrating DNT cells had a higher expression of PD-1 relative to adjacent and normal tissue (Fig.
3a). Consistent with this observation, DNT cells co-cultured with lung cancer cells increased PD-1 expression (Fig.
3e and Additional file
2: Figure S3B). Additionally, xenograft infiltrating DNT cells also showed higher PD-1 expression compared to pre-infusion cells (Fig.
5a). Collectively, our findings are consistent with the observation that tumor recognition and activation of T cells lead to upregulation of PD-1 [
17,
48] and suggest that expression of PD-1 on DNT cells is regulated in a similar manner. Interestingly, patient-derived tumor infiltrating DNT cells expressed a lower level of PD-1 than conventional CD4
+ and CD8
+ T cells (Fig.
3b). In line with this, we found that stimulation of CD4 T cells in vitro resulted in sustained PD-1 expression, which differs from what was observed for DNT and CD8 T cells (Fig.
3d). These findings show differences in PD-1 regulation between T cell subsets and suggest the possibility that DNT cells may be more resistant to tumor microenvironmental changes in vivo than conventional T cells.
Observations from patients responsive to ICB suggest that blocking PD-1 greatly increased the number and function of CD8
+ T cells infiltrating the tumor bed [
49]. Interestingly, we found that the addition of anti-PD-1 to DNT and lung cancer cell co-cultures only increased killing of PD-L1 over expressing cells but not the lung cell lines natively expressing PD-L1 (Additional file
2: Figure S7). Though initially surprising, this observation was consistent with results published by others using gamma/delta T cells and CIK cells [
44,
46] and suggests that tumor natively expressed PD-L1 may not have enough density to alter innate T cell function in vitro. Similar to observations in patients receiving ICB, we found that anti-PD-1 blockade also led to greater numbers of DNT cells within tumors (Fig.
5b and Additional file
2: Figure S5E), suggesting that DNT cells were regulated by the PD-1/PD-L1 pathway. Whether anti-PD-1 blockade increased the ability of DNT cells to migrate to xenografts or survive within xenografts was not directly explored, but given the role of PD-1 engagement in regulating T cell activation [
14] and apoptosis [
15,
16], and that DNT cells could infiltrate tumors in the absence of ICB (Fig.
2d and e), anti-PD-1 blockade may allow for continued DNT cell survival within tumors.
Tumor recognition by DNT cells was shown to be dependent on ligation of NKG2D and DNAM1 receptors by innate ligands preferentially expressed on malignant cells [
24,
25]. In addition to increasing the number of DNT cells within tumor xenografts, we found that anti-PD-1 treatment resulted in increased NKG2D
+ and DNAM1
+ DNT cells, capable of cytolytic granule secretion (Fig.
5). This increase in tumor-recognizing DNT cells within xenografts coincided with an increase in tumor necrosis (Fig.
4d and Additional file
2: Figure S5D), supporting direct engagement and lysis of lung cancer xenografts by DNT cells. Indeed, addition of anti-PD-1 to adoptively transferred DNT cells significantly enhanced DNT cell-mediated tumor inhibition and prolonged the survival of tumor bearing mice (Fig.
4 and Additional file
2: Figure S5). Taken together, these data support the notion that combination therapy of anti-PD-1 and DNT cells is beneficial to DNT cell therapy of solid tumors such as lung cancer.
Our results show that ex vivo expanded DNT cells can infiltrate and inhibit the growth of late-stage lung cancer in xenograft models. Given the similarity between DNT cells derived from lung cancer patients and healthy donors, the non-allogeneic “off-the-shelf” nature of DNT cells may be ideal for adoptive cell therapy in lung cancer. This contrasts other adoptive cellular therapy combination strategies that utilize autologous CIK
44, which are difficult to grow from patients, or antigen specific T cells [
43] which may be prone to resistance by tumor antigen loss [
5]. Given the innate recognition mechanisms utilized by DNT cells, which do not rely on traditional peptide-HLA recognition [
24], DNT cell therapy is less likely affected by the known primary or acquired resistances to ICB such as a low tumor mutation burden, lack of tumor reactive T cells [
4,
5] or loss of HLA [
50]. Further, as DNT cells show benefit from addition of ICB, DNT cell therapy can be used as an adjunct to patients already receiving immune checkpoint blockade and may be ideal for patients characterized as having “immune deserts”.