Dynamic host immunity and PD-L1/PD-1 blockade efficacy: developments after “IFN-γ from lymphocytes induces PD-L1 expression and promotes progression of ovarian cancer”

Abstract

In the article titled “IFN-γ from lymphocytes induces PD-L1 expression and promotes progression of ovarian cancer” in 2015, we showed that PD-L1 expression is induced by IFN-γ from lymphocytes in the tumour microenvironment. This article proposed that PD-L1 expression in cancer cells is not stable but varies among cases, or even within a case, which is influenced by the stromal infiltration of cytotoxic lymphocytes. Immune-checkpoint inhibitors, especially anti-PD-1/PD-L1 therapies, are now widely used to treat various types of cancer. Predictive biomarkers for the efficacy of immune-checkpoint inhibitors include PD-L1 expression, MSI/mismatch repair deficiency and high tumour mutation burden. However, clinical trials have proven that their use in ovarian cancer is still challenging. Reliable biomarkers and new treatment strategies may be sought by elucidating the complex immune microenvironment of ovarian cancer. Although the interaction between cytotoxic lymphocytes and PD-1/PD-L1 on tumour cells is at the centre of therapeutic targets, other immune checkpoints and various immunosuppressive cells also play important roles in ovarian cancer. Targeting these role players in combination with PD-1/PD-L1 blockade may be a promising therapeutic strategy.

Introduction

Seven years have passed since we published our article in the British Journal of Cancer (2015), titled “IFN-γ from lymphocytes induces PD-L1 expression and promotes progression of ovarian cancer” [1]. In this issue, celebrating the 75th anniversary of the British Journal of Cancer, we review what happened after the publication of our article.

In the early 2010s, researchers and physicians were excited to hear the encouraging reports of randomised controlled trials (RCTs) for nivolumab, one of the first available immune-checkpoint inhibitors used to treat various advanced-stage malignant tumours that were previously regarded as incurable [2, 3]. We previously reported a sustained complete response to nivolumab in some patients with platinum-resistant recurrent ovarian cancer [4]. PD-L1 is a ligand for PD-1, and PD-1 is a co-inhibitory molecule mainly expressed in T cells. The PD-L1/PD-1 interaction silences the major histocompatibility complex (MHC) to T-cell receptor (TCR) positive activation signal in T cells, resulting in the inhibition of excessive immune activation [5]. Insufficiency of this system leads to an autoimmune disease-like phenotype in mice [6]. Cancer cells take advantage of this system to evade the host’s immune system, which is otherwise able to eliminate cancer cells presenting non-self-antigens by cytotoxic T lymphocytes (CTLs). Therefore, blocking the PD-L1/PD-1 interaction results in a gain of functioning from otherwise exhausted CTLs and a renewed potential for killing cancer cells [7, 8]. Animal models and early RCTs of malignant tumours, including melanoma and lung cancers, have shown promising results 10 years ago [2, 3, 5].

At that stage, we as gynaecological cancer specialists were optimistic in our thinking that immune-checkpoint inhibitors would certainly help a fraction of patients with cases of ovarian cancer once deemed incurable and that we would manage to find well-founded biomarkers to select eligible patients in less than 10 years. We were partly right and partly wrong about these expectations. On the one hand, the Food and Drug Administration approved the use of PD-1 inhibitors in clinical practice for any patients with solid tumours proved to be MSI-high or carrying high TMB [9–11]. However, many patients do not respond to treatment with PD-1 inhibitors. Furthermore, while many patients with endometrial cancer have been shown to benefit from immune-checkpoint inhibitors, which might also be the case for many patients with cervical cancer soon, this is not the case for ovarian cancer [12–16]. First, very few patients with ovarian cancer are MSI-high or TMB-high [17]. The second reason is that even if these patients carry MSI-high status, the responses to anti-PD-1 therapy are unsatisfactory [7]. PD-L1 protein expression in cancer cells is regarded as an FDA-approved biomarker to predict response to anti-PD-1 therapy in non-small cell lung cancers and uterine cervical cancers [9, 18, 19], but not in ovarian cancer [4]. At the time of writing in 2022, we did not find a solid and reliable biomarker to select patients with ovarian cancer patients expected to truly benefit from immune-checkpoint inhibitor treatment.

Before our article in 2015, it was generally considered that the expression of PD-L1 observed in surgical or biopsy specimens was solid and unchangeable. However, surprisingly, our study, along with others, has shown that it is dynamically changeable by interactions between host immune cells [1, 20, 21]. This indicates that PD-L1 expression in tumour cells can vary based on space and time. As a pathological feature is merely a snapshot of just a tiny fraction of the whole tumour taken on one occasion, we must imagine how it is different in other places with comparatively more or less immune cells being present and how it will be affected by the administered treatment, tumour cell evolution and varied immune reactions.

In this review, we briefly reviewed the contents of the findings shown in our article in 2015, and then reviewed how far we have come for the clinical application of the PD-L1/PD-1 blockade and the search for reliable biomarkers. We also discuss some pre-clinical and clinical studies elucidating the complex immune microenvironment in ovarian cancer, with the hope of illustrating the future perspectives to which direction we are going in when it comes to immune-checkpoint inhibitor treatment.

IFN-γ from lymphocytes induces PD-L1 expression

PD-L1 is a co-regulatory molecule expressed on the surface of various cell types, including immune and epithelial cells [7]. PD-1 is a receptor of PD-L1 that is mainly expressed in immune cells, including lymphocytes [22]. When PD-L1 binds to PD-1 on lymphocytes, an inhibitory co-signal is exerted on lymphocytes, resulting in the inhibition of activation [5]. Under normal circumstances, these “braking” systems are necessary to avoid further burdening of the immune system under conditions of injury or virus infection [7]. Knockout mice lacking PD-L1 present with various types of autoimmune diseases [6].

It is known that the immune system possesses the potential to eradicate cancer cells during the early initiation step of cancer development [23]. Cancer cells potentially harbour multiple mutations and therefore produce cancer antigens that are present on MHC molecules on both the tumour cell surface and antigen-presenting cells, including dendritic cells [24]. The antigen-presenting cells prime clonal T-cell expansion in secondary lymphoid organs and produce CTLs, which attack and eliminate tumour cells. The PD-L1/PD-1 system inhibits the excessive activation of T lymphocytes (Fig. 1), so it often happens that the speed of tumour cell elimination is exceeded by the rate of proliferation of tumour cells, which allows for clinical progression of the tumour [25]. Blocking PD-L1 or PD-1 in immunocompetent animal models results in tumour shrinkage [26].

Fig. 1. Unidirectional interaction of ovarian cancer cells and CTLs.

CTLs attacking ovarian cancer cells are inactivated by PD-1/PD-L1 binding.

However, the regulation of PD-L1 expression in the cancer microenvironment remained unclear. In 2015, we reported that the number of CD8 + cells in the cancer stroma was high in peritoneally disseminated tumours and was strongly correlated with PD-L1 expression in tumour cells in ovarian cancer clinical samples [1]. In most ovarian cancer cell lines, PD-L1 expression is low but can be induced by adding IFNs. In patients with ovarian cancer, the expression levels of PD-L1 in tumour cells vary among cases. In mouse models, depleting interferon-γ receptor 1 resulted in lower levels of PD-L1 expression in tumour cells, an increased number of tumour-infiltrating CD8 + lymphocytes or CTLs, inhibition of peritoneally disseminated tumour growth and longer survival. Injection of IFN-γ into subcutaneous tumours induced PD-L1 expression and promoted tumour growth, and PD-L1 depletion completely abrogated tumour growth caused by IFN-γ injection. Therefore, we concluded that IFN-γ secreted by CD8-positive lymphocytes upregulated PD-L1 in ovarian cancer cells and promoted tumour growth. These findings indicated that lymphocyte infiltration and IFN-γ status may be key to the response to anti-PD-1 or anti-PD-L1 therapy in ovarian cancer.

Using microarray data from clinical samples of patients with ovarian cancer, we found that the “IFN-γ signature score” was high in samples with high lymphocyte infiltration [1]. Tumours with a high IFN-γ signature are generally called “Immune-hot” tumours and are known to respond well to immune-checkpoint inhibitors. The development of testing strategies to select immune-hot patients for anti-PD-1/PD-L1 therapy, and treatment strategies to turn “immune-cold” tumours into immune-hot tumours, remains a hot topic [27].

In summary, before our paper in 2015, the function of PD-L1 on tumour cells against immune cells had been broadly investigated, but our paper elucidated the mechanism the other way around, i.e. how immune cells function against PD-L1 expression on tumour cells through IFN-γ (Fig. 2). We examined the interactions between host immunity and tumour cells.

Fig. 2. Bidirectional interaction of ovarian cancer cells and CTLs.

Interferon-γ secreted from activated CTLs induces PD-L1 expression on ovarian cancer cells. IFN interferon, IFGNR interferon-γ receptor.

Development of immune-checkpoint inhibitors and clinical application of PD-L1/PD-1 blockade in ovarian cancer

In 2012, two articles showing the efficacy of anti-PD-1 and anti-PD-L1 therapies in various cancers were published [2, 3]. This was a significant step forward in the treatment of cancer and showed that we had entered the era of immune-checkpoint inhibitors.

Along with the PD-1/PD-L1 system, CTLA-4 is another promising target for immune-checkpoint therapy. CTLA-4 is expressed mainly in lymphocytes and shares its ligands with CD28, a molecule that exerts activation co-signal [28]. Blocking CTLA-4 resulted in more CD28 being connected to the ligand, and lymphocytes were more activated. Anti-CTLA-4 antibody drugs are effective against some types of cancers. Combination therapy with anti-PD-1 and anti-CTLA-4 antibodies showed excellent efficacy in patients with melanoma [29]. Currently, therapies targeting PD-1/PD-L1 are being applied to broader types of cancer than those targeting CTLA-4. In this review, we focused on immune-checkpoint inhibitors targeting PD-1/PD-L1 applied to ovarian cancer.

We previously reported that PD-L1 expression in tumour cells is an unfavourable prognostic factor in ovarian cancer [30, 31]. Using murine models, we showed that PD-L1 on tumour cells is induced in malignant ascites and promotes peritoneal dissemination of ovarian cancer through the dysfunction of CTLs [26]. After these pre-clinical studies, we conducted a phase II clinical trial of nivolumab, an anti-PD-1 antibody, and reported that the drug showed partial or complete response in 15% of patients with platinum-resistant recurrent ovarian cancer [4]. The therapeutic effect was durable in patients with a complete response [32]. This trial was the first to apply anti-PD-1 therapy to ovarian cancer.

Nivolumab and pembrolizumab are two major anti-PD-1 drugs, which proved to be effective in various cancer types. Clinical trials of nivolumab and pembrolizumab for ovarian cancer are listed in Table 1 [4, 16, 33–40]. In ovarian cancer, primary tumours generally respond well to platinum-based chemotherapy [41]. Recurrent tumours retaining platinum sensitivity are treated with platinum-based chemotherapy followed by PARP-inhibitors [41]. Therefore, many clinical trials are designed for platinum-resistant recurrent cases (Table 1). Recently, we reported the results of the Phase III clinical trial of nivolumab for platinum-resistant ovarian cancer (NINJA Study) [16]. Unfortunately, anti-PD-1 therapy resulted in no improvement in overall survival and worse progression-free survival compared to cytotoxic monotherapy in patients with platinum-resistant ovarian cancer [16].

Table 1.

Clinical trial of anti-PD-1 therapy for ovarian cancer.

		Drug	Combination with	Phase	Ovarian cancer status	ORR (%)	Study name
Hamanishi [4]	2015	Nivolumab		II	Platinum resistant	15
Liu [33]	2019	Nivolumab	Bevacizumab	II	Relapsed	28.9
Zamarin [34]	2020	Nivolumab	Ipilimumab	II	Recurrent or persistent	12.2/31.4
Hamanishi [16]	2021	Nivolumab		III	Platinum resistant	a	NINJA
Varga [35]	2019	Pembrolizumab		II	PD-L1-positive advanced	11.5	KEYNOTE-028
Matulonis [36]	2019	Pembrolizumab		II	Recurrent	7.4/9.9	KEYNOTE-100
Lee [37]	2020	Pembrolizumab	PLD	II	Platinum resistant	26.1
Zsiros [38]	2021	Pembrolizumab	Bevacizumab, oral CPA	II	Recurrent	47.5
Walsh [39]	2021	Pembrolizumab	CDDP, GEM	II	Platinum resistant	60
Liao [40]	2021	Pembrolizumab	CBDCA	I/II	Platinum resistant	10.3

PLD pegylated liposomal doxorubicin, CPA cyclophosphamide, CDDP cisplatin, GEM gemcitabine, CBDCA carboplatin, ORR overall response rate.

^aMedian overall survival 10.2 vs 12.1 months (conventional, PLD or GEM monotherapy).

As for pembrolizumab, although fairly effective when administered in combination with cytotoxic drugs, anti-PD-1 monotherapy showed an overall response rate of around 10% [35, 36]. To our knowledge, a Phase III clinical trial of pembrolizumab for platinum-resistant ovarian cancer is not conducted.

Another clinical trial showed that atezolizumab, an anti-PD-L1 antibody, had no clinical benefit when used in the first-line setting in combination with cytotoxic drugs and bevacizumab [15]. The addition of atezolizumab did not significantly prolong progression-free survival [15]. Currently, the application of PD-1/PD-L1 blockade therapy in the general population of patients with ovarian cancer is not supported by evidence.

Seeking biomarkers for efficient PD-L1/PD-1 blockade

There was enthusiasm about applying immune-checkpoint inhibitors to treat various cancer types [42]. Some of the attempts were successful, whereas others were not. Some tumour types that were regarded as “immunologic” tended to respond well to immune-checkpoint inhibitors. Melanoma and renal cell carcinoma are examples. The clinical response rate was <30% in most tumour types. Significant efforts have been made to identify biomarkers for predicting the efficacy of anti-PD-1/PD-L1 therapy. First, we focused on PD-L1 expression in the tumours. Currently, in non-small cell lung cancer, uterine cervical cancer, gastric cancer, triple-negative breast cancer, and head and neck squamous cell carcinoma, PD-L1 expression is used as a clinical indicator to select patients eligible for anti-PD-1/PD-L1 therapy [43, 44]. However, PD-L1 expression has not been proven to be a good biomarker for many types of cancer, including ovarian cancer. We believe that this failure comes from the incorrect belief that the expression of PD-L1 is stable throughout the tumour across time and space. In contrast, PD-L1 expression in tumour cells can be dynamically changed upon attack by immune cells, and the distribution of immune cells varies considerably among cases, as discussed in a later chapter.

Another biomarker came from a trial involving various tumour types with MSI [9]. Cancers with a defective DNA mismatch repair (dMMR) system contain thousands of mutations most frequently located in monomorphic microsatellites and are thereby defined as having MSI [45]. MSI is a marker for dMMR. MSI/dMMR can be identified using immunohistochemistry to detect loss of MMR proteins and/or molecular tests to detect microsatellite alterations. MSI/dMMR is a good biomarker for a favourable response to immune-checkpoint inhibitors [9].

Tumour mutation burden is the latest clinically used biomarker to identify candidates for immune-checkpoint inhibitors after clinical next-generation sequencing tests [46]. Tumours harbouring numerous mutations tend to produce neoantigens, which means that the immune system can prime lymphocytes, and eventually, CTLs can attack tumour cells if the immune-inhibitory system does not work well [47]. A tumour mutation burden ≥175 mutations/exome is associated with improvement in the efficacy of pembrolizumab monotherapy and improved outcomes for pembrolizumab versus chemotherapy across tumour types [46]. TMB is reported to have clinical utility irrespective of tumour type, PD-L1 expression, or MSI status and can be used as a predictive biomarker for PD-1/PD-L1 targeted therapy. Through systematic analysis of mutational signatures, we have developed software to classify whole-exome sequenced tumours to subtypes significantly correlated with immune-checkpoint monotherapy efficacy [48]. Such attempts to implement new analysis tools to select patients benefiting from PD-1/PD-L1 targeted therapy are on the way.

Currently, immune-checkpoint inhibitors are generally used to treat:

1. Certain tumour types that are known to respond well to immune-checkpoint inhibitors;

2. Cases characterised by high MSI; and

3. Cases characterised by mutation burden

In gynaecology, cervical cancer may be the most immunogenic tumour type. It has been reported that cervical cancers carry a high tumour mutation burden at a higher rate. PD-1 blockade therapy, including cemiplimab or pembrolizumab, has been shown to be effective in phase III trials for recurrent cervical cancer [11, 12].

Genetic analysis of endometrial cancer has shown that there are four distinct biological subtypes of endometrial cancer, namely, POLE-mutant, MMR-deficient, p53-mutant, and p53 wild-type. POLE-mutant cases are also called ultra-mutated cases and harbour numerous mutations [49]. MMR-deficient cases harbour mutations in MLH1, MLH2, MSH6 or PMS2, resulting in MSI [50]. POLE-mutant cases generally show a favourable prognosis, while MMR-deficient cases often respond well to anti-PD-1 therapy [51, 52].

In contrast to cervical and endometrial cancers, ovarian cancer carries fewer mutations in most cases. Clinical biomarkers for the application of ICIs in ovarian cancer are unidentified. We examined the complex immune microenvironment of ovarian cancer to identify other immunotherapies or good combination drugs for use along with ICIs.

A complex immune microenvironment possibly affects ICI efficacy in ovarian cancer and other potent targets to enhance ICI efficacy

Controlling lymphocyte infiltration in ovarian cancer

Since ICI efficacy is unsatisfactory in patients with ovarian cancer, the mechanisms of PD-L1 expression on cancer cells have been explored [1]. PD-L1 is generally expressed when cells are exposed to interferon (IFNs). IFN-γ plays an important role in both innate and adaptive immunities. IFN-γ is mainly produced by lymphocytes in the cancer microenvironment. We thus explored the number and types of tumour-infiltrating lymphocytes. High infiltration of CD4+ and CD8+ lymphocytes is associated with a significantly better prognosis in ovarian cancer [31].

Tertiary lymphoid structures (TLS) are transient ectopic lymphoid aggregates that are occasionally observed in cancer. The coexistence of CD8+ T cells and B cell lineages in the tumour microenvironment significantly improve the prognosis of ovarian cancer patients and is correlated with the presence of TLS [53]. CXCL13 gene expression was correlated with TLS presence and infiltration of T and B cells. CXCL13 expression predominantly coincided with CD4+ T cells in TLS and CD8+ T cells in TILs, and shifted from CD4+ T cells to CD21 + follicular DCs as TLS matured [53]. TLS is associated with favourable survival outcomes in endometrial cancer [54]. TLS might be the source of tumour-infiltrating lymphocytes and, as a result, a source of IFN-γ.

Cancer-associated fibroblasts (CAFs) constitute an important part of the tumour microenvironment. CAFs present a pathologically activated phenotype that enables them to influence tumour progression, dissemination, and response to therapy through the remodelling of the extracellular matrix (ECM) and signalling to cancer, endothelial and immune cells [55]. Presenilin 1 was highly expressed in CAFs of ovarian cancer. Presenilin 1 silencing significantly promoted CD8+ CTL proliferation and penetration in murine ovarian models, resulting in tumour regression and growth inhibition [56].

Other immune checkpoints

It has been reported that inhibitory immune checkpoints other than PD-L1 affect tumour behaviour [57]. V-domain Ig suppressor of T-cell activation (VISTA) is a novel inhibitory immune-checkpoint protein [57]. VISTA selectively engages and suppresses T cells at an acidic pH, similar to that found in tumour microenvironments [58]. We have previously reported that, in tumour cells, VISTA suppresses T-cell proliferation and cytokine production in vitro and decreases tumour-infiltrating CD8+ T cells in vivo. The anti-VISTA antibody prolonged the survival of tumour-bearing mice [59].

B7-H3 is another immune-checkpoint protein belonging to the B7 family [60]. B7-H3 is highly expressed in cancer cells but seems to have immunostimulatory and immunoinhibitory effects on T cells through two different receptors [61]. We reported that B7-H3-high ovarian cancer cases showed shorter overall survival and knockout of B7-H3 in a murine ovarian cancer model prolonged mouse survival [62].

Other immunosuppressive checkpoint molecules that have been identified in ovarian cancer include T-cell immunoglobulin and mucin domain and T-cell immunoreceptor with Ig and ITIM domains [63–65].

Immunosuppressive cells

Tumour cells evade host immunity in two ways, i.e. (1) expression of immune-checkpoint molecules, including PD-L1, and (2) recruitment of immunosuppressive cells [66]. Immunosuppressive cells include myeloid-derived suppressor cells (MDSC), tumour-associated macrophages (TAM) and regulatory T cells (Tregs). These immunosuppressive cells play important roles in tumour immune escape by attenuating the proliferation and activation of lymphocytes [67, 68].

MDSCs are immature myeloid cells that migrate to peripheral tissues without differentiating into macrophages or dendritic cells [69]. MDSCs are classified into two types: monocytic-MDSCs (M-MDSCs) and polymorphonuclear MDSCs (PMN-MDSCs or granulocytic-MDSCs) [66]. Both types of MDSCs suppress cytotoxic T lymphocytes (CTLs) by producing iNOS and arginase. Increased MDSC counts in the peripheral blood of patients with cancer. We have previously found that an increased MDSC count in the tumour microenvironment is associated with poor prognosis in patients with ovarian cancer [70]. In clinical settings, there are many reports showing negative correlations between MDSC counts and survival in patients with cancer [71]. For example, circulating MDSCs negatively impact survival and are inversely correlated with the presence of functional antigen-specific T cells in patients with advanced melanoma [72]. Reducing the number of tumour-infiltrating MDSCs inhibited tumour growth in ovarian cancer mouse models [70]. We found that anti-angiogenic therapy-induced hypoxia leads to the secretion of GM-CSF, which recruits MDSC to the tumour microenvironment [73]. The anti-VEGF drug bevacizumab is a key drug used for ovarian cancer treatment [74–76]. MDSC may be a potential target for overcoming resistance to anti-angiogenic therapy in ovarian cancer. In another study, we found that the activation of the NF-κB pathway resulted in the expression of CXCL1/2, a chemokine and ligand for CXCR2. The CXCL1/2-CXCR2 axis induces MDSC migration in ovarian cancer [77]. Hypoxia-related gene expression is closely associated with the NF-κB pathway [78]. NF-κB is also a key molecule involved in the induction of PD-L1 during chemotherapy in ovarian cancer [79]. Epithelial-mesenchymal transition (EMT) is a key process in ovarian cancer metastasis, and Snail is a key molecule implicated in EMT [80]. We previously reported that Snail induces CXCL1 and CXCL2 through the NF-κB pathway, possibly through direct binding to their promoters [77].

Tumour-associated macrophages (TAM) are major role players in the immunosuppressive tumour microenvironment. Macrophages are differentiated cells of the mononuclear phagocytic lineage and are classified as activated (M1) or alternatively activated (M2) based on the polarisation of their activation and secreting of cytokines [81]. In the tumour microenvironment, M2 macrophages gain an immunosuppressive phenotype [82, 83]. In a murine ovarian cancer model, a CXCR4 antagonist enhanced anti-tumour responses in combination with anti-PD-1 therapy [84]. The CXCR4 antagonist downregulated the expression of CXCL12 and CXCR4 and promoted macrophage polarisation from M2 to M1 [84]. In another murine ovarian cancer model, blocking macrophage function using a colony-stimulating factor 1(CSF-1) receptor kinase inhibitor resulted in reduced infiltration of M2 macrophages and decreased ascites volume [85].

Regulatory T cells (Tregs) are CD4 + lymphocytes with immunosuppressive functions. FoxP3 is a central molecule expressed in Tregs, which are associated with higher stages of ovarian cancer, but not with survival [86].

Taken together, the recruitment of immunosuppressive cells, other than immune checkpoints, fosters immunosuppression. Targeting immunosuppressive cells in combination with immune-checkpoint inhibitors may be a promising strategy for the treatment of ovarian cancer.

In the last decade, by applying immune-checkpoint inhibitors to gynecologic cancers, we have found that some cases respond well to immunotherapy while others do not. Ovarian cancer seems to be one of the most challenging cancer types to treat with immunotherapy. From our study using clinical samples and murine models in 2015, we found that PD-L1 status is not stable throughout the tumour but varies according to space and time [1]. In particular, PD-L1 status varies with tumour-infiltrating lymphocyte status. We started exploring the diverse and complex immune microenvironment, lymphocytes, and IFN-γ in the centre, but with many other important side-players. The major players in the immune tumour microenvironment are illustrated in Fig. 3. Using new drugs to target by-players in combination may increase the efficacy of PD-1/PD-L1 therapies.

Fig. 3. Multi-directional interaction of ovarian cancer microenvironment.

CTLs are inactivated through signal from checkpoint molecules and immunosuppressive cells. CTLs can be supplied from TLS. Immunosuppressive cells include TAM, MDSC and CAF. Checkpoint molecules include PD-L1, VISTA, B7-H3, PD-1, TIGIT and TIM-3. TLS tertiary lymphoid structure, TAM tumour-associated macrophage, MDSC myeloid-derived suppressor cell, CAF cancer-associated fibroblast, VISTA V-domain Ig suppressor of T-cell activation, TIGIT T-cell immunoreceptor with Ig and ITIM domains, TIM-3 T-cell immunoglobulin and mucin domain.

Author contributions

KA drafted the manuscript. JH, NM and MM revised the manuscript. MM supervised the study.

Funding

The authors received no specific funding for this work.

Data availability

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.

Competing interests

JH received grants from Ono Pharmaceutical, Sumitomo Dainippon Pharma, KinoPharma and MSD outside the submitted work. NM is an outside director of Takara Bio Inc. NM received a grant and lecture fees from AstraZeneca and received lecture fees from Takeda Pharmaceutical outside the submitted work. The remaining authors declare no competing interests.

Ethics approval and consent to participate

None.

Consent for publication

None.