The clinical evidence for targeting human myeloid‐derived suppressor cells in cancer patients

Short abstract

Review on clinical trials that directly or indirectly target MDSCs in cancer patients, and future directions for the field.

Abstract

Myeloid‐derived suppressor cells (MDSCs) are a heterogeneous population of immature myeloid cells that represent a formidable obstacle to the successful treatment of cancer. Patients with high frequencies of MDSCs have significantly decreased progression‐free survival (PFS) and overall survival (OS). Whereas there is experimental evidence that the reduction of the number and/or suppressive function of MDSCs in mice improves the efficacy of anti‐cancer therapies, there is notably less evidence for this therapeutic strategy in human clinical trials. Here, we discuss currently available data concerning MDSCs from human clinical trials and explore the evidence that targeting MDSCs may improve the efficacy of cancer therapies.

Abbreviations

5FU

5‐fluorouracil

APL

acute promyelocytic leukemia

Arg1

arginase 1

ATRA

all trans retinoic acid

BPH

benign prostate hyperplasia

DC

dendritic cell

ED

erectile dysfunction

eMDSC

early‐stage myeloid‐derived suppressor cell

FDA

U.S. Food and Drug Administration

FOLFIRI

folinic acid, 5‐fluorouracil, and camptothecin‐11 combination

FOLFIRINOX

leucivirin, fluorouracil, oxaliplatin, and irinotecan combination

FOLFOX

folic acid, 5‐fluorouracil, and oxaliplatin combination

GemCape

gemcitabine and capecitabine combination

GIST

gastrointestinal stromal tumor

GSH

glutathione

HNSCC

head and neck squamous cell carcinoma

MDSC

myeloid‐derived suppressor cell

mo‐MDSC

monocytic myeloid‐derived suppressor cell

NRF2

NF (erythroid‐derived 2)‐like 2

NSCLC

nonsmall cell lung cancer

OS

overall survival

PD‐1

program cell death 1

PDE5

phosphodiesterase 5

PDGFR

platelet‐derived growth factor receptor

PFS

progression‐free survival

PMN‐MDSC

polymorphonuclear myeloid‐derived suppressor cell

PNET

pancreatic neuroendocrine tumor

RA

retinoic acid

RAR

retinoic acid receptor

RCC

renal cell carcinoma

ROS

reactive oxygen species

RTK

receptor tyrosine kinase

SBRT

stereotactic body radiotherapy

T_reg

regulatory T cell

VEGF

vascular endothelial growth factor

VEGFR

vascular endothelial growth factor receptor

Introduction

Therapies that target the immune system to eliminate tumors are improving the outlook for many cancer patients [1, 2]. Despite rapid advances in the field of immunotherapy, relatively few patients have long‐term, complete responses to these treatments [1, 2]. Tumors evade clearance by the immune system through a variety of mechanisms, including secretion of inhibitory factors, expression of inhibitory immune checkpoint molecules, and recruitment and expansion of T_regs and tumor‐associated macrophages [3]. One recently recognized mechanism of ongoing immunosuppression that may also limit the efficacy of immunotherapies is the recruitment and expansion of MDSCs [4, 5], which present a formidable obstacle to the successful treatment of many forms of cancer.

MDSCs are a heterogeneous population of immature myeloid cells that are expanded in tumor‐bearing hosts [4, 5, 6–7]. In addition to their impact in cancer, MDSCs play an important role in other disease processes, such as chronic infections, sepsis, autoimmunity, and chronic inflammatory conditions [5, 7, 8]. In general, MDSCs are subdivided into 2 populations: mo‐MDSC and PMN‐MDSC, based on their origin and cell surface markers [7]. Additionally, a population of MDSCs that do not express the markers that are associated with mo‐MDSCs or PMN‐MDSCs is known as eMDSCs [7]. MDSCs are difficult to study in human patients as a result of their low numbers and the complexity of the myeloid cell system. These difficulties have led to controversies in identifying MDSCs and differences in the defining surface markers of particular subpopulations across the field [7]. A consensus group has proposed the following guidelines for the characterization of human MDSCs; these cells are lineage (CD3, CD19, CD56)‐negative CD11b^PosCD33^PosHLA‐DR^Neg. Further characterization of MDSC subpopulations define human Mo‐MDSCs as CD14^PosCD15^NegCD66^Neg, PMN‐MDSCs are defined as CD14^NegCD15^PosCD66b^Pos, and eMDSCs are defined as CD14^NegCD15^NegCD66^Neg [7]. Some evidence suggests that the different subpopulations may be differentially up‐regulated depending on the specific type of tumor. For example, in RCC and pancreatic adenocarcinoma, PMN‐MDSCs are believed to be the dominant population of MDSCs, whereas in hepatocellular carcinoma patients, evidence suggests that mo‐MDSCs are the dominant subpopulation of MDSCs [9].

TRAFFICKING AND MECHANISMS OF MDSC‐INDUCED SUPPRESSION

The accumulation of MDSCs in the tumor microenvironment and periphery is driven by tumor‐derived cytokines, chemokines, and growth factors [4, 7]. Once in the tumor microenvironment, MDSCs prevent a successful anti‐tumor immune response and promote a favorable microenvironment for tumor growth by several mechanisms. They include production of Arg1 and ROS, which decrease T cell recognition of and response to tumor cells [8]. Additionally, MDSCs inhibit T cell proliferation through direct cell contacts (Fas:Fas ligand), as well as through production of suppressive cytokines and molecules, including IL‐10, TGF‐β, and IDO [4, 5, 8, 10].

Since their discovery and documentation of their suppressive function in the tumor microenvironment, researchers and clinicians have sought to target MDSCs, as well as their suppressive functions, to improve the anti‐tumor immune response. Whereas this strategy has been promising in mice, there is little clinical evidence for using this approach in humans. Here, we discuss the results of clinical trials and studies in human cancer patients that directly or indirectly target MDSCs and review active and future clinical trials targeting MDSCs.

CLINICAL EVIDENCE FOR TARGETING MDSCs IN HUMANS

Most of the evidence for therapeutically targeting MDSCs in humans comes from secondary observations in patients undergoing standard or experimental therapies. In many cases, the initial observations were a result of off‐target effects, noted after experiments measuring MDSC numbers or function. However, evolving evidence suggests that some therapies have more direct effects on MDSC accumulation, maturation, and function. This review will focus on currently available (FDA‐approved) ( Table 1 ) therapies and summarize what is known about their direct and indirect effects on MDSCs.

Table 1.

FDA‐approved agents with reported effects on human MDSCs

Agent	FDA‐approved indications	Agent classification	General mechanism of action	Effects on human MDSCs	Effects on patient outcome in approved indication	Refs. for MDSC activity
ATRA	APL	Differentiating agent	Binds ERK1/2 and RARs; ↑GSH synthase; ↑GSH; ↓ROS	Reduces frequency of circulating total and mo‐MDSCs by differentiating cells into mature myeloid cells	↑OS; ↑PFS; ↑RR	[12, 14, 15]
Vitamin D	Secondary hypoparathy‐ and hyperparathy‐roidism	Differentiating agent	Inhibits ERK signaling and promotes apoptosis via up‐regulation of BAX	Reduces frequency of circulating CD34Pos MDSC by differentiating cells into mature myeloid cells	Unknown; currently in clinical trials for cancer treatment	[25, 26–27]
Sunitinib	RCC, GIST, and PNET	RTK inhibitor	Inhibits activity of RTKs, including PDGFR and VEGFR	Restricts MDSC development; reduces number and/or suppressive function of circulating total and PMN‐MDSCs	↑OS; ↑PFS; ↑RR (RCC and GIST)	[35, 37]
Bevacizumab	Colon and several other cancers	Angiogenesis inhibitor	Inhibits angiogenesis by inhibiting VEGF binding to VEGFR	Decreases circulating total and KITPos MDSC frequency	↑PFS; ↑OS (cervical and colon cancer)	[42, 43]
Tadalafil	ED and BPH	PDE5 inhibitor	NO and arginase inhibition; ↓Arg1 expression; ↓NOS2 expression; ↓ iNOS expression; catalyzes the hydrolysis of cGMP [7]; ↑ intracellular [cGMP]; ↑NO2; ↑penile BP	Reduces number and/or suppressive function of bone marrow and circulating total and mo‐MDSCs	Unknown; currently in clinical trials for cancer treatment	[52, 53, 56]
Ipilimumab	Melanoma	Check‐point inhibitor	CTLA‐4 antibody promotes immune activation.	Reduces frequency of circulating PMN‐MDSC	↑OS; ↓ROP; ↑RR; ↑disease control rate	[71, 72]
				Reduces the number of ARG1+ myeloid cells
Vemurafenib	Melanoma with BRAF V600E mutations	B‐raf/Mek inhibitor	Inhibits mutated B‐RAF	Indirect action; reduces frequency of circulating mo‐MDSC	↑OS; ↓ROP; ↑RR; ↑PFS	[75]
			↓IL‐6; ↓IL‐10; ↓VEGF production by the tumor

FDA‐APPROVED THERAPIES WITH DIRECT EFFECTS ON MDSCs

Differentiating agents

ATRA

ATRA is a vitamin A derivative. ATRA is the current FDA‐approved standard‐of‐care treatment for APL [11, 12]. In APL, ATRA terminally differentiates the immature myelocytic tumor cells, resulting in death of the tumor cells. Likewise, ATRA is thought to differentiate immature MDSCs into macrophages and DCs [13]. The mechanism of action is thought to be binding of RA to RARs that contribute to the transcriptional control of many genes containing RA response elements [11, 12, 14]. Furthermore, it is hypothesized that ATRA differentiates MDSCs through activation of ERK1/2, up‐regulation GSH synthase, and subsequent generation of GSH. This increase in GSH results in decreased production of ROS and subsequent cellular differentiation [13].

There have been at least 2 completed clinical trials using ATRA, in combination with immunotherapies, to target MDSCs in cancer patients. The first trial tested the safety and efficacy of a single dose of ATRA in combination with IL‐2 [15]. This phase I trial treated 18 metastatic RCC patients with 50, 100, or 150 mg/m²/d ATRA for 7 d, followed 7 d later by standard IL‐2 therapy. With the focusing on understanding the pharmacokinetics of ATRA, this trial showed that the greatest reduction in the number of circulating MDSCs occurred in patients for whom the circulating plasma levels of ATRA reached 150 ng/ml. The investigators measured the frequency of the total MDSC population (Lin^NegCD33^PosHLA‐DR^Neg) and did not focus on any of the specific subpopulations. Treatment with ATRA also resulted in a shift in the ratio of DCs and MDSCs. Interestingly, this study showed that subsequent IL‐2 treatment increased the number of circulating MDSCs, potentially abrogating the ATRA‐mediated changes in anti‐tumor immune responses. Outcomes were measured between wk 11 and 12 of treatment using Response Evaluation Criteria in Solid Tumors criteria. The investigators found that of the patients that completed the treatment phase, 1 patient had a complete response, 11 had stable disease, and 3 had progressed.

The second trial enrolled a total of 41 patients diagnosed with advanced‐stage small cell lung cancer. In this trial, ATRA was combined with a p53‐transduced DC vaccine [16]. The patients were divided into Arm A receiving vaccine alone, Arm B receiving vaccine and ATRA at a dose of 50 mg/m²/d, and Arm C receiving vaccine and ATRA at a higher dose of 150 mg/m²/d. Patients received a total of 4 vaccinations, at 4 wk intervals, and ATRA was given 1 d before, the day of, and the day after vaccination in Arms B and C. The investigators found that ATRA reduced the number or circulating, total MDSCs (Lin^NegCD33^PosHLA‐DR^Neg) and mo‐MDSCs (Lin^NegCD33^PosHLA‐DR^NegCD14^Pos) by 2‐fold in patients with plasma ATRA concentrations > 150 ng/ml. Although no patients in Arm A developed detectible, p53‐specific CD8⁺ T cells, 20% of patients in Arm B (P = 0.22) and 41.7% in Arm C (P = 0.012) developed an anti‐p53 immune response. Whereas ATRA was effective at reducing the number of MDSCs and improving the anti‐cancer immune response, as of writing this review, no outcome data have been reported from this trial.

ATRA has also been used to potentiate conventional chemotherapeutic agents, although MDSCs were not measured in these trials [11, 17, 18, 19–20]. One trial hypothesized that ATRA would increase the expression of the RAR isoform RAR‐β2 [21]. Loss of RAR‐β2 expression promotes proliferation and resistance to chemotherapeutics, and induction of RAR‐β2 expression by ATRA suppresses carcinogenesis [22]. This trial, involving 107 NSCLC patients found that ATRA increased the clinical response rate and significantly extended the overall and PFS of patients being treated with combination paclitaxel and cisplatin [21]. Patients received ATRA at a dose of 20 mg/m²/d for 1 wk before, as well as during, 2 cycles of standard paclitaxel and cisplatin treatment. Whereas MDSCs were not measured in this trial, the clinical results from this trial suggest that ATRA and other MDSC targeting agents have the potential to improve cancer patient outcomes.

Vitamin D

Similar to ATRA, vitamin D3 may differentiate MDSCs and improve the anti‐tumor immune response. Vitamin D signals and results in differentiation of immature cells using a similar mechanism as ATRA. Vitamin D3 binds to the vitamin D3 receptor, resulting in signaling through vitamin D response elements and regulating transcription of target genes [23]. In the case of myeloid cells and myeloid cell lines, vitamin D3 signaling and exposure to IL‐6 or TGF‐β can result in terminal differentiation [24, 25].

An initial phase‐1b dose escalation study found that treatment with 60 μg 25‐dihydroxyvitamin D3 (also known as calcifediol)/d for 6 wk decreased the number of circulating MDSCs in HNSCC patients by an average of ∼30% (P = 0.002) [26]. Patients included in this study had not been treated with any other cancer therapy for at least 3 wk before enrollment. No outcome data were reported in this study.

A small clinical study involving 17 HNSCC patients showed that treatment with 1α,25‐dihydroxyvitamin D3 (also known as calcitriol) reduced the number of tumor‐infiltrating MDSCs (P = 0.006) [27, 28]. This study demonstrated that vitamin D3 treatment increased the number of mature tumor‐infiltrating DCs, activated CD4⁺ and CD8⁺ T cells, and improved the anti‐tumor immune response. Patients received 3 cycles consisting of 4 μg/d 1α,25‐dihydroxyvitamin D3 for 3 d, followed by no treatment for 4 d. The investigators went on to show that 1α,25‐dihydroxyvitamin D3 extended the median time to recurrence from 181 in the control group to 620 d in the vitamin D‐treated group (P = 0.048) [28]. Both of these clinical trials analyzed frequency of immunosuppressive CD34^Pos myeloid cells, a different population than is usually identified as MDSCs [26, 27–28]. Whereas CD34^Pos cells contain a population of immunosuppressive MDSCs, CD34 also identifies a large population of hematopoietic progenitor cells, and CD34 has generally been replaced with other markers [5, 7].

Tyrosine kinase and angiogenesis inhibitors

Sunitinib

Sunitinib is a small‐molecule, synthetic indoline‐based RTK inhibitor approved for treating RCC [29], PNET [30], and GISTs [31]. Sunitinib inhibits several RTKs, including VEGFR, PDGFR, fetal liver tyrosine kinase 3, and c‐Kit [32]. These RTKs drive tumor progression via the PI3K‐AKT pathway [33, 34]. Although no studies have shown a definitive mechanism for targeting of MDSCs by sunitinib, it has been hypothesized that it acts on MDSCs by blocking VEGFR, thereby preventing MDSCs from promoting angiogenesis [35]. Sunitinib may also function by blocking the signaling required for phosphorylation of STAT3 [36], a key intracellular signaling regulator in MDSCs [6, 37].

There have been at least 2 studies showing that sunitinib can decrease the number or suppressive function of MDSCs in metastatic renal cell cancer patients. The first study treated patients with a 6 wk course of 50 mg/d sunitinib monotherapy, comprised of 28 d of treatment and 14 d of rest [38]. The investigators found that sunitinib treatment decreased the frequency of total MDSCs (CD33^PosHLA‐DR^Neg) from an average of 5.42 to 2.28% (P = 0.002) and PMN‐MDSCs (CD33^PosHLA‐DR^NegCD14^NegCD15^Pos) from 5.49 to 2.21% (P < 0.001). However, in this study, the number of MDSCs or other immune parameters measured did not correlate with OS, PFS, or tumor burden. In another study, with the combination of sunitinib with SBRT, sunitinib significantly decreased the numbers of both circulating, total MDSCs (CD33^PosHLA‐DR^Neg/Lo; P = 0.0293) and mo‐MDSCs (CD33^PosHLA‐DR^Neg/LoCD14^PosCD16^Pos; P = 0.0483) at 6 to 30 d post‐SBRT [36]. This study also showed that sunitinib decreased the suppressive function of MDSCs by reducing their expression of Arg1 and phosphorylated STAT3. Whereas this study showed no statistically significant differences in patient outcomes, the investigators noted a trend toward improved OS, PFS, and cause‐specific survival in patients treated with both sunitinib and SBRT.

Bevacizumab

VEGF is a key mediator of tumor‐induced angiogenesis that promotes endothelial cell proliferation [39]. Importantly, VEGF also inhibits maturation of myeloid cells [40, 41–42]. VEGF‐mediated angiogenesis can be inhibited by the anti‐VEGF antibody bevacizumab. A study involving 19 metastatic colorectal patients showed that bevacizumab decreased circulating, immature, immunosuppressive myeloid cells (Lin^NegHLA‐DR^Neg) from 0.39 ± 0.30% to 0.27 ± 0.24% (P = 0.012) [43]. This study also showed that bevacizumab improved the stimulatory capacity of DCs isolated from treated patients. Both MDSCs and DCs were measured 14 d after bevacizumab administration. No OS or PFS data were published with this study.

A later study, involving 21 metastatic breast cancer patients, found that standard‐of‐care bevacizumab, in combination with standard‐of‐care paclitaxel, decreased circulating, immunosuppressive myeloid cells (Kit^PosCD11b^Pos) by 60–80% after 15 d of treatment (P ≤ 0.001) [44]. This is a different population of cells than is normally considered MDSCs, as the investigators did not measure the expression of HLA‐DR [7]. Of note, paclitaxel alone increased MDSCs by 2‐fold compared with baseline values; therefore, the addition of bevacizumab was able to overcome this effect and decrease MDSCs. However, no OS or PFS data were published with this study.

PDE5 inhibitors

Tadalafil

Tadalafil is a member of the PDE5 inhibitor class of drugs that also includes sildenafil and vardenafil. PDE5 inhibitors are approved to treat ED [45, 46], BPH [47], cardiac hypertrophy [48], and pulmonary hypertension [49] but have garnered increased attention for treating cancer patients [50]. PDE5 inhibitors have been shown to inhibit the suppressive function of MDSCs by decreasing the expression of IL‐4Rα while increasing intracellular cGMP concentrations [51]. IL‐4 signaling induces phosphorylation of STAT6, reinforcing the expression of Arg1 and other suppressive mechanisms [52]. There have been at least 3 clinical trials treating patients with tadalafil to target MDSCs.

A small study involving a single multiple myeloma patient showed that combing tadalafil with lenolidomide, dexamethasone, and clendomycin produced a profound clinical response with a 90% reduction in disease burden [53]. A clinical response was not observed when the patient was previously put on the same combination without tadalafil. This response lasted ∼18 mo before the patient passed away from complications unrelated to multiple myeloma. This clinical response was associated with decreased expression of the suppressive markers IL‐4Rα, Arg1, and iNOS in bone marrow mo‐MDSCs (HLA‐DR^Neg/LoCD14^Pos).

These investigators followed this small study with a larger study enrolling 35 HNSCC patients, hypothesizing that tadalafil would improve anti‐tumor immune responses [54]. Patients were randomized into 10 or 20 mg/d groups and treated for 20 d before tumor resection. This study demonstrated that tadalafil decreased the number of circulating mo‐MDSCs (CD33^PosHLA‐DR^Neg/LoCD11b^PosIL‐4Rα^PosCD14^Pos; P = 0.046) and showed a trend toward a decrease in tumor‐infiltrating MDSCs (CD33^PosIL‐4Rα^Pos; P = 0.09). Additionally, the investigators showed that tadalafil decreased the number of circulating T_regs (P = 0.001) and increased the frequency of activated tumor‐infiltrating CD8⁺ T cells (P = 0.012) identified in paraffin‐embedded tumors. In this trial, experimental responses, including decreased MDSCs and T_regs and increased, active CD8⁺ T cells, were greatest in patients treated with the 10 mg/d dose. Some evidence suggests that higher doses of tadalafil (e.g., 20 mg/d) may have off‐target effects, inhibiting other PDEs, including PDE11, resulting in increased immune‐inhibitory cAMP concentrations [55, 56]. The investigators did not include OS or PFS clinical outcome data.

A similar trial in 32 patients diagnosed with HNSCC compared 20 mg/d tadalafil with placebo after 10 d of treatment. The majority of patients were then given radiation therapy [57]. This study found that tadalafil decreased circulating, total MDSCs (CD33^PosHLA‐DR^Neg/LoIL‐4Rα^PosCD14^Pos or CD15^Pos; P = 0.001). The investigators also showed that tadalafil decreased the expression of Arg1 and iNOS in the total MDSC population. Again, there was no OS or PFS data reported in this study.

Conventional chemotherapeutic agents

Gemcitabine and 5FU

Many trials have reported the off‐target effects of conventional systemic chemotherapeutic agents, such as gemcitabine, 5FU, and the 5FU prodrug capecitabine on MDSC frequency and function. Gemcitabine is a nucleoside analog that replaces cytidine in DNA and is approved to treat many forms of cancer. By incorporating into the DNA of rapidly dividing cells, gemcitabine treatment leads to apoptosis of cancer cells. Likewise, 5FU is a pyrimidine analog that causes cell death via inhibiting thymidylate synthase. Whereas there have been a significant number of studies using these drugs to target MDSCs in mice, the clinical evidence for their effects on human MDSCs is sparser [58, 59]. The mechanism by which these drugs affect MDSCs has also not been well characterized.

Gemcitabine and 5FU are often used in combinations with other drugs. Two studies showed efficacy in reducing the effects of MDSCs. The first of these used GemCape with the cancer vaccine V1001 containing low‐dose GM‐CSF to treat advanced pancreatic cancer patients [60]. The study involved 3 arms, healthy controls (n = 24), GemCape (n = 19), and GemCape with V1001 (n = 21). The investigators found that GemCape alone decreased circulating, total MDSCs (Lin^NegHLA‐DR^NegCD11b^Pos) in 8/19 patients, whereas GemCape with V1001 decreased circulating, total MDSCs (Lin^NegHLA‐DR^NegCD11b^Pos) in 18/21 patients. Decreased frequencies of total MDSCs in both groups were associated with decreased tumor‐secreted cytokines and correlated with changes in tumor size. No outcome data were included in this study.

Another study, involving 23 metastatic colorectal cancer patients, combined either FOLFIRI or FOLFOX [61]. In patients being treated with FOLFOX, the frequency of total MDSCs (CD33^PosHLA‐DR^NegCD11b^Pos) decreased throughout therapy by ∼52%, whereas the FOLFIRI patients showed an increase in total MDSCs (CD33^PosHLA‐DR^NegCD11b^Pos). Decreases in MDSC frequencies in the FOLFOX‐treated patients were associated with improved anti‐tumor immune responses, as measured by increased CD247 expression on T cells, and increased MDSC frequencies in the FOLFIRI‐treated patients were associated with decreased anti‐tumor immune responses. Outcome data have not been published from this study.

FDA‐APPROVED THERAPIES WITH INDIRECT OR VARIABLE EFFECT ON MDSCs

Immunotherapies

Ipilimumab

Immunotherapies function by improving immunologic anti‐tumor activity. Ipilimumab is a fully humanized, anti‐CTLA‐4 antibody approved for treating unresectable or metastatic melanomas [1] and as an adjuvant treatment for Stage III melanoma [62]. In 2 successful Phase III clinical trials, ipilimumab was the first treatment to show an absolute survival benefit in melanoma patients [1, 63]. CTLA‐4 is an inhibitory molecule on T cells that prevents T cell activation [64]. Ipilimumab blocks the inhibitory signals that T cells receive through CTLA‐4, resulting in an improved anti‐tumor immune response [64]. CTLA‐4 is expressed at high levels on T_regs, and ipilimumab treatment significantly reduces T_reg frequency in melanoma patients [65]. Whereas MDSCs can express the ligands for CTLA‐4 (CD80 and CD86) [66, 67], MDSCs are not known to express CTLA‐4, and effects of ipilimumab on MDSC frequency and function may be indirect.

As a standard of care treatment for advanced melanoma, several ipilimumab studies have correlated circulating MDSCs with clinical outcome and response rate [68, 69, 70–71]. Whereas the correlation between the frequency of MDSCs and clinical responses to ipilimumab and other immunotherapies is well accepted, the consequences of ipilimumab treatment on their frequency or suppressive function are more controversial. Examples of this include 2 recent studies showing different effects of ipilimumab on MDSCs in cancer patients. In the first study involving 8 melanoma patients treated with the 3 or 10 mg/kg dose of ipilimumab, the frequency of circulating PMN‐MDSCs (Lin^NegHLA‐DR^Neg/LoCD15^PosCD33^PosCD11b^Pos) was reduced by ∼1.5 fold [72]. The observed decrease in the frequency of circulating PMN‐MDSCs was accompanied by a decrease in Arg1 production. Alternatively, in a larger study involving 56 melanoma patients treated with the 3 mg/kg dose of ipilimumab, the frequency of total MDSCs (Lin^NegHLA‐DR^Neg/LoCD33^PosCD11b^Pos) was unchanged [73]. This trial concluded that the frequency of MDSCs was only correlated with severity of disease. Differences in patient populations and dosing, as well as the previously discussed differences in data analysis techniques, may explain the discordant results from these trials.

Tyrosine kinase inhibitors

Vemurafenib

Another important current therapy for the treatment of melanoma is the orally available small molecule tyrosine kinase inhibitor vemurafenib, which is a B‐Raf enzyme inhibitor approved for treating advanced melanoma patients whose tumors have the BRAF V600E mutation [74]. B‐Raf is a serine/threonine protein kinase in the MAPK/ERK signaling pathway that is up‐regulated in some melanomas. Mutations in B‐Raf can be found in 40–60% of melanomas [75]. In a study involving 41 melanoma patients, vemurafenib treatment decreased circulating mo‐MDSCs (Lin^NegHLA‐DR^NegCD33^PosCD14^Pos) by ∼21% in patients who exhibited a clinical response [76]. Whereas the frequency of PMN‐MDSCs (Lin^NegHLA‐DR^NegCD33^PosCD66b^PosCD16^Neg) decreased in some patients, the clinical response was strongly correlated with mo‐MDSCs, whereas there was no correlation between clinical response and the frequency of PMN‐MDSCs. This study suggests that vemurafenib has an indirect effect on MDSCs by decreasing the amount of IL‐6, IL‐10, and VEGF that is produced by melanoma cells. In this context, IL‐6 is important for MDSC function, as it maintains STAT3 phosphorylation and Arg1 expression [37, 77]. The authors suggest that vemurafenib may reduce immunosuppression in melanoma patients undergoing immunotherapy, but it is important to note that the severe hepatotoxicity was observed in a trial combining vemurafenib with ipilimumab that stopped the trial early for safety reasons [78].

AGENTS WITH POTENTIAL MDSC ACTIVITY IN HUMANS

New treatments targeting MDSCs in open or soon‐to‐be‐opened cancer clinical trials

There is a wide variety of therapies and combination therapies currently being tested or soon to be tested in human clinical trials that list targeting MDSCs as an outcome measure in the National Cancer Institute clinical trials database (www.cancer.gov/about‐cancer/treatment/clinical‐trials/search; October 19, 2016; Table 2 ). Three promising trials will be discussed here.

Table 2.

Active clinical trials targeting MDSCs in human cancer treatments

Agent(s)	Type of cancer(s) tested	Agent classification	General mechanism of action	Proposed effects of human MDSCs	Trial number
Arginine‐rich nutritional supplement with surgery	Colon cancer	Differentiating agent	Increases availability of arginine	Reduces MDSC‐suppressive function	NCT01885728
Capecitabine + bevacizumab	Recurrent glioblastoma	Cytotoxic agent + VEGF inhibitor	Kills rapidly dividing cells and blocks angiogenesis	Capecitabine ↓MDSC number	NCT02669173
				Bevacizumab ↓MDSC‐mediated angiogenesis and possibly MDSC recruitment to the tumor
Cisplatin + CKM + celecoxib + DC vaccine versus cisplatin + celecoxib + DC vaccine	Ovarian cancer	COX‐2‐selective inhibitor + arginase inhibitor + chemokine modulator + cytotoxic agent	Celecoxib prevents COX‐2 from making PGs out of arachidonic acid, which regulates inflammation; COX‐2 inhibitors also function as arginase inhibitors; CKM may alter cell signaling; cisplatin kills rapidly dividing cells.	Celecoxib ↓MDSC‐suppressive function	NCT02432378
				Cisplatin ↓MDSC frequency
DC vaccine with or without gemcitabine	Sarcoma	Vaccine + nucleoside and cytotoxic agent	Gemcitabine is a nucleoside and kills rapidly dividing cells.	Gemcitabine ↓MDSC frequency	NCT01803152
EP4 antagonist, AAT‐007 (RQ‐07; CJ‐023,423) + gemcitabine	Prostate cancer, NSCLC, and breast cancer	EP4 antagonist, anti‐inflammatory	EP4 antagonist, anti‐inflammatory	AAT‐007 and gemcitabine are proposed to ↓MDSC frequency.	NCT02538432
Ipilimumab and ATRA	Stage IV melanoma	Immune check‐point inhibitor + differentiating agent	Ipilimumab is an anti‐CTLA‐4 mAb that alters T cell activation and proliferation and reduces Treg function; ATRA, a vitamin, up‐regulates GSH synthase, leading to increased intracellular concentrations of GSH and decreased ROS production.	ATRA ↓MDSC frequency and ↓suppressive function; ipilimumab improves anti‐tumor immune response	NCT02403778
5FU, leucovorin, bevacizumab + anakinra	Metastatic colorectal cancer	Cytotoxic agent + VEGF inhibitor + IL‐1R antagonist	LV5FU2 consists of leucovorin and 5FU, which kill rapidly dividing cells; bevacizumab inhibits VEGF disrupting cell signaling; anakinra is an IL‐1R antagonist	Bevacizumab ↓MDSC‐mediated angiogenesis and possibly MDSC recruitment to the tumor	NCT02090101
				5FU ↓MDSC frequency
PF‐04136309 + FOLFIRINOX	Pancreatic adenocarcinoma	CCR2 antagonist + a cytotoxic agent	PF‐04136309 inhibits the inflammatory response and metastasis in some tumors.	PF‐04136309 proposed to ↓MDSC trafficking to tumor.	NCT01413022
			FOLFIRINOX consists of leucovorin, fluorouracil, oxaliplatin, and irinotecan, which kill rapidly dividing cells.
PD‐L1 or PD‐1 inhibitor	NSCLC	Immune check‐point inhibitor	PD‐1 inhibitors bind the PD‐1 receptor and block it from binding with PD‐L1 and PD‐2 ligand; PD‐L1 inhibitors block the ligand for the PD‐1 receptor and B7‐1; both PD‐L1 and PD‐1 inhibitors result in T cell proliferation activation, cytokine production, and cell adhesion.	Proposed to ↓MDSC‐suppressive function	NCT02827344
Omaveloxolone (RTA‐408) in combination with ipilimumab or nivolumab	Melanoma	NRF2 activation + immune check‐point inhibitor	In mice, RTA‐408 reduces tumor nitrotyrosine burden, inhibits the activity of MDSCs, and augments T cell anticancer activity.	RTA‐408 proposed to ↓MDSC‐suppressive function.	NCT02259231
			Ipilimumab is an anti‐CTLA‐4 mAb that alters T cell activation and proliferation and reduces Treg function; nivolumab is an anti‐PD‐1 inhibitor.
Tadalafil	Squamous cell carcinoma	PDE5 inhibitor	NO and arginase inhibitor	↓MDSC frequency and suppressive function	NCT01697800
			Tadalafil: ↓Arg1 expression; ↓NO2 expression; ↓ iNOS expression
Tadalafil + anti‐ MUC‐1 vaccine	HNSCC	PDE5 inhibitor + vaccine	NO and arginase inhibitor.	↓MDSC frequency and suppressive function	NCT02544880
			Tadalafil: ↓Arg1 expression; ↓ΝΟ2 expression; ↓ iNOS expression

CKM, Chemokine modulatory regime; COX‐2, cyclooxygenase 2; EP4, PGE2 receptor 4; LV5FU2, leucovorin and fluorouracil combination; MUC‐1, mucin 1; NO2, nitrogen dioxide; PD‐L1, PD‐1 ligand; ↓, down‐regulates/decreases; ↑, up‐regulates/increases.

One of the immunosuppressive processes used by MDSCs is the production of ROS [4]. Inhibition of ROS production or neutralization of ROS has been shown to reduce the immunosuppressive function of human MDSCs in vitro [79]. Recent efforts to characterize the ROS production and oxidative stress pathways in MDSCs have identified the regulatory transcription factor NRF2 as a key regulator of ROS and antioxidant production [80]. NRF2 controls the transcription of several antioxidant genes that decrease the amount of ROS generated by MDSCs [80]. The triterpenoid class of drugs, including the first‐generation triterpenoid 2‐cyano‐3,12‐dioxooleana‐1,9(11)‐dien‐28‐oic acid methyl ester (RTA‐402), activates NRF2 [81]. A trial using RTA‐402 did not alter the number of circulating MDSCs in advanced pancreatic cancer patients; however RTA‐402 treatment did increase the frequency of T cells responding to tetanus toxoid and phytohemagglutinin [81]. These studies led to the development of the second‐generation triterpenoid drug, omaveloxolone (RTA‐408). RTA‐408 activates NRF2, resulting in decreased ROS‐mediated immunosuppression. In a trial recruiting melanoma patients (NCT02259231), RTA‐408 will be combined with ipilimumab or the anti‐ PD‐1 antibody nivolumab.

Another promising strategy for targeting MDSCs is inhibition of their trafficking to the tumor site. MDSCs can be attracted to and accumulate in tumors by several different mechanisms [4]. A number of chemokines and their receptors promote MDSC accumulation and immunosuppression within the tumor microenvironment in human cancer patients [4]. For example, CXCL12, produced by malignant ovarian tumors, is strongly associated with the accumulation of CXCR4^Pos mo‐MDSCs (CD11b^PosHLA‐DR^NegCD33^PosCD14^Pos) in the tumor microenvironment in these patients [82]. Whereas in patients with HNSCC, CXCL8 (IL‐8) has been shown to recruit PMN‐MDSCs [83]. CXCL8 was targeted with the anti‐CXCL8 antibody HuMax‐IL‐8 in a recently completed clinical trial involving patients diagnosed with unresectable, locally advanced malignant solid tumors (NCT02536469). This trial did not list quantification of frequency or function of MDSCs as an outcome, and the results of this study were not available at the time of publication.

One mechanism being explored to inhibit the chemokine–chemokine receptor‐mediated accumulation of MDSCs is to target CCR2 [84], which binds to the ligands CCL2, CCL8, and CCL16, resulting in chemotaxis toward the source of the ligand. These chemokines can be produced by cancer cells and may lead to accumulation of MDSCs within the tumor microenvironment [84, 85]. In mice, CCR2 antagonism significantly improves anti‐tumor immune responses [85]. CCR2 is expressed at higher levels on mo‐MDSCs and tumor‐associated macrophages than PMN‐MDSCs and is an important regulator of trafficking of MDSCs to tumor sites [86, 87]. In a trial involving pancreatic adenocarcinoma patients (NCT01413022), investigators will treat patients with the CCR2 antagonist PF‐04136309, in combination with the cytotoxic cocktail FOLFIRINOX. Interestingly, PMN‐MDSCs are the dominant MDSC population in pancreatic adenocarcinoma patients [88]. Through blocking the chemokine receptor CCR2, this trial aims to decrease the ability of MDSCs to migrate to the tumor and promote a productive anti‐tumor immune response.

Our group is currently interested in expanding on the previous trials using ATRA to differentiate MDSCs [15, 16] by combining ATRA with ipilimumab to treat Stage IV melanoma patients (NCT02403778). In previous trials, ATRA significantly decreased the frequency of circulating, total MDSC (Lin^NegCD33^PosHLA‐DR^Neg) and mo‐MDSCs (Lin^NegCD33^PosHLA‐DR^NegCD14^Pos), resulting in improved anti‐tumor immune responses. This new trial will recruit a total of 48 melanoma patients into 2 arms, comparing the combination of ATRA and ipilimumab with ipilimumab alone. As the frequency of circulating MDSCs correlates with clinical responses to ipilimumab [68], this trial aims to improve the efficacy of ipilimumab by decreasing the frequency and suppressive function of MDSCs. The 3 MDSC populations (PMN, mo, and e) will be measured over the course of treatment and will be correlated with anti‐tumor immune responses and clinical outcomes.

CONCLUDING REMARKS

MDSCs present a significant barrier to the successful treatment of many types of cancer. The targeting of MDSCs in cancer has garnered increased attention as a potential mechanism to improve the anti‐cancer immune response. Challenges in identifying human MDSCs and the differences between mouse and human MDSCs have, until recently, prevented the successful elimination of MDSCs from cancer patients. Here, we have summarized what is known about MDSCs in human cancer clinical trials with the hope that the targeting of MDSCs may further enhance the advances made in immunotherapeutic cancer therapies.

AUTHORSHIP

R.P.T., D.D., K.R.J., and M.D.M. wrote the review.

DISCLOSURES

The authors declare no conflicts of interest.