• Non ci sono risultati.

Biological significance and prognostic role of PD-L1 expression in cutaneous melanoma

N/A
N/A
Protected

Academic year: 2021

Condividi "Biological significance and prognostic role of PD-L1 expression in cutaneous melanoma"

Copied!
69
0
0

Testo completo

(1)

1

TABLE OF CONTENT

INTRODUCTION 3

T CELL RECOGNITION AND IMMUNE CHECKPOINTS 5

PD-1/PD-L1 EXPRESSION AND BIOLOGICAL FUNCTION 8

PD1/PD-L1 INTERACTION: FUNCTIONAL IMPLICATION 10

PD-1/PD-L1 IN MALIGNANT DISEASES 12

PROGNOSTIC ROLE OF PD-L1 AND PD-1 IN MALIGNANCIES 16

CLINICAL TRIALS AND FUTURE DEVELOPMENTS 18

AIM OF THE STUDY 23

MATERIALS AND METHODS 25

RESULTS 32

PD-L1 expression in melanoma tissues 32

Correlation between PD-L1 expression, disease-free survival (DFS)

and overall survival (OS) 42

(2)

2

PDL-1+ A375 cells show a distinct gene profile with up-regulation

of genes regulating tumor growth and diffusion 49

PD-L1+ A375 cells show enhanced migration and invasion in vitro 51

PD-L1+ A375 cells show enhanced growth and diffusion in vivo 52

DISCUSSION 53

(3)

3 INTRODUCTION

Cellular transformation and tumor development result from an accumulation of mutational and epigenetic changes that alter normal cell growth and survival pathways (Suva

et al., 2013). Several lines of evidence indicate that tumorigenes is a multistep process that

potentially reflect the genetic alterations that drive the progression and metastasis. In 2000, Hanahan and Weinberg reported six biological capabilities (‘The hallmarks of cancer’) acquired during the multistep development of human tumors: sustaining proliferative signaling, evading growth suppressors, resisting cell death, enabling replicative immortality, inducing angiogenesis, and activating invasion and metastasis (Hanahan et al., 2000). In the last two decades, based on the emergence of new experimental data, two further hallmarks have been added to this list: reprogramming of energy metabolism and evading immune destruction (Hanahan et al., 2011).

This Copernican revolution had its foundation on the awareness that complete knowledge of cancer development cannot be achieved without recognizing the importance of the tumor microenvironment, a very important part of which is played by the immune system. A dynamic interplay exists between host and tumor, and the ability of the tumor to evade immune recognition (immune surveillance) often determines the clinical course of the disease (Hoenicke et al., 2012).

The concept of cancer immune surveillance is based on the hypothesis that the immune system can suppress the development or progression of spontaneous malignancies (Miska et al., 2012). Several data, first from animal models and later from studies in cancer patients, confirmed this original concept that the immune system can recognize and reject tumors, supporting the hypothesis that immune evasion by cancer cells plays an important role in the development and progression of tumors (Klein et al., 2005).

Based on this biological background, cancer immunotherapy focuses on the development of agents that can activate the immune system to recognize and kill tumor cells. However, until recently, all the efforts to therapeutically modulate the immune system were tempered by disappointing results from clinical trials with cancer vaccines and biochemotherapy regimens, as well as by the low response rates and high toxic effect associated with these two strategies (Atkins et al., 2008; Bajetta et al., 2006; Eton et al., 2002; Keilholz et al., 2005; Legha et al., 1996; Legha et al., 1998; O'Day et al., 2002; Ridolfi

(4)

4

et al., 2002). A third approach known as adoptive cell therapy has shown clinical benefit for

some patients although technical aspects and the complexity of the procedures have limited more widespread use (Zheng et al., 2013). Despite the initial lack of clinical success, extensive research over the past twenty years yielded the identification of innovative ways to manipulate the immune response to cancer.

The regulation of the immune system depends on an adequate control system in which a key role is played by cellular receptors that ensure the activation or inhibition of cells involved in the control of infections and tumors and development of autoimmunity. Some of these mechanisms are activating and dictate whether the response arises, while others play the role of powerful repressors. Antagonist antibodies acting on such repressors result in enhanced immune responses, a goal that may also be achieved with agonist antibodies acting on the activating receptors.

More recently, significant enthusiasm arose for a fourth immunotherapeutic strategy: the use of immunomodulatory monoclonal antibodies that directly enhance the function of components of the anti-tumor immune response such as T cells or block immunologic checkpoints that would otherwise restrain effective anti-tumor immunity. To date, the “immune checkpoints” not only normally function to control excessive immune activation but also appear to be a means by which tumors evade the immune system. Indeed, this strategy is based on the evidence that development of cancer is enabled by the dis-regulation and exploitation of otherwise physiological pathways that, under normal circumstances, down-modulate immune activation and maintain tolerance to self.

Immune recognition and elimination of malignant cells require a series of steps orchestrated by the innate and the adaptive response of the immune system. The majority of tumors have evolved mechanisms that allow for successful evasion of these immune responses. Recognition of these evasive processes led to the development of immunotherapeutic antibodies targeting the co-stimulatory and co-inhibitory receptors on T cells, with the goal of enhancement of T cell activation or reversal of tumor-induced T cell inhibition.

A series of therapeutic agents are under clinical development, and one of them, which is directed at the CTL-associated antigen 4 (CTLA-4) inhibitory receptor (Ipilimumab, Yervoy®), has been approved for the treatment of metastatic melanoma. The list of antagonist agents acting on repressors under development includes anti-CTLA-4,

(5)

5

anti-Programmed death-1 (PD-1), anti-PD-Ligand 1 (L1) or (B7-H1), anti-KIR (killer cell Ig-like receptor), and anti-TGF-β. Agonist antibodies currently being investigated in clinical trials target CD40, CD137 (4-1BB), CD134 (OX40), and glucocorticoid-induced TNF receptor (GITR) (Figure 1). Among these pathways an important role is covered by two immunologic checkpoints: the CTLA-4 and the PD-1/PD-L1 axis.

An emerging concept in cancer immunology is that inhibitory ligands such as PD-L1 are induced in response to immune attack, a mechanism termed ‘adaptive resistance’. This potential mechanism of immune resistance by tumors suggests that therapy directed at blocking interaction between PD-1 and PD-L1 might synergize with other treatments that enhance endogenous antitumor immunity.

Figure. 1. Agonist antibodies (on the left) and antagonist antibodies (on the right). Antagonist agents under development acting on repressors include anti-CTLA-4 and anti-Programmed death-1 (PD-1).

T CELL RECOGNITION AND IMMUNE CHECKPOINTS

Induction of immune response requires T cells to receive two sets of signals from antigen-presenting cells: the T cell receptor must recognize complexes of MHC with the antigen on the surface of an antigen-presenting cell (APC). T cells and the T cell receptor complex do not respond to antigen in solution, but even for the specific antigen they only respond to antigen-MHC complexes on the cell surface. This interaction is necessary, but not sufficient, for T cell activation. T cell activation also requires a co-stimulatory signal involving interaction of CD28 on the T cell with CD80 or CD86 (B7 family genes) on the

(6)

6

APC, which promotes T cell clonal expansion and cytokine secretion (Lafferty et al., 1980). CD28 activates a signal transduction pathway acting through phosphatidylinositide 3-kinases (PI-3K), lymphocyte-specific protein tyrosine kinase (Lck) and adaptor proteins GRB2 and GRB2-related adaptor protein (Grb-2/ITK) to provide its co-stimulatory signal for T cell activation. T cell-mediated immunity is regulated by balancing stimulatory and inhibitory signals that regulate the response (Vigano et al., 2012). In the absence of co-stimulatory molecules, the T cells enter an unresponsive state known as clonal anergy in which the T cells are incapable of providing antigen-specific immune responses (Schwartz, 1990). In physiological conditions, immune checkpoints are crucial for maintaining self-tolerance and for protecting tissues from the damage of the immune response to pathogenic infection. There is evidence that tumors resist immune attack by inducing tolerance towards tumor-specific T cells and by expressing ligands that bind inhibitory receptors such as immune checkpoints. Tumors can result in deregulation of the immune-checkpoint proteins and develop a mechanism for immune resistance, especially against T cells that are specific for tumor antigens (Pardoll, 2012).

Many of the tumor-expressed targets for therapeutic antibodies are growth factor receptors that show increased expression during tumourigenesis. By blocking ligand binding and/or signalling through these receptors, monoclonal antibodies may serve to normalize growth rates, induce apoptosis and/or help sensitize tumors to chemotherapeutic agents. In addition, antibodies that target the tumor microenvironment and inhibit processes such as angiogenesis have shown therapeutic promise (Weiner et al., 2010). Preliminary clinical data show that antibody blockage of immune checkpoints can substantially enhance therapeutic antitumor immunity (Mellman et al., 2011; Pardoll, 2012). Unlike conventional antibodies used for the treatment of tumors, antibodies that block immune checkpoints do not bind directly to the tumor cells, but target lymphocyte receptors or their ligands, in order to modulate their antitumor activity. Several immunologic treatment options, such as the induction of an immune response or the administration of antibodies, have been investigated in melanoma and have shown interesting results (Weiner et al., 2010).

CTLA-4 and PD-1 are the two most investigated immune checkpoint receptors in melanoma and cancer immunotherapy (Table 1) (Keir et al., 2008). CTLA-4 is an inhibitory membrane receptor expressed exclusively on T cells, where it primarily regulates the amplitude of the early stages of T cell activation, counteracting the activity of the T cell

(7)

7

co-stimulatory receptor, CD28. The engagement of the T cell antigen receptor by itself is not sufficient for full T cells activation; a second co-stimulatory signal is required. This co-stimulation is mediated by engagement of CD28 on the T cell surface by members of the B7 family on APC (Lenschow et al., 1996). These co-stimulatory molecules are integral membrane proteins expressed on several cells with APC function. These molecules, including B7.1 (CD80) and B7.2 (CD86), bind to other ligands on T cells and provide the second signal for T -cells activation. Limited expression of B7 on APCs is a mechanism for maintenance of peripheral T cells tolerance, ensuring that T cells activation can only be stimulated by appropriate cells (Schwartz, 1992).

Interestingly, tumor cells do not express B7, and this contributes to their poor capacity to induce immune responses (Chen et al., 1992; Townsend et al., 1993). After activation, T cells express CTLA-4, a close homologue to CD28. CTLA-4 binds members of the B7 family with a much higher affinity than CD28 (Linsley et al., 1991). Accordingly, CTLA-4 expression on the surface of T cells decreases the activation of T cells by competing with CD28 for binding with CD80 and CD86. CTLA-4 exerts distinct effects on the two major subsets of CD4+ T cells: down-modulation of helper T cell activity and enhancement of regulatory T (Treg) cell immunosuppressive activity. The specific signaling pathways by which CTLA-4 blocks T cell activation are still under investigation, although a number of studies suggest that activation of the protein phosphatases, SHP2 (also known as PTPN11) and PP2A, is important in counteracting kinase signals that are induced by TCR and CD28. The central role of CTLA-4 for keeping T cell activation in check is dramatically demonstrated by the lethal systemic immune hyperactivation phenotype of CTLA-4-knockout mice. In experimental models, mice rapidly develop lymphoproliferative disease with multiorgan lymphocytic infiltration and tissue destruction, including particularly severe myocarditis and pancreatitis, and die at 3-4 weeks of age (Tai et al., 2007). The severe phenotype of mice lacking CTLA-4 implies a critical role for CTLA-4 in down-regulating T cell activation and maintaining immunologic homeostasis. In the absence of CTLA-4, T cells are activated, can spontaneously proliferate, and may mediate lethal tissue injury.

(8)

8

Table 1. Similarities and differences between CTLA-4 and PD-1

PD-1/PD-L1 EXPRESSION AND BIOLOGICAL FUNCTION

The PD-1 receptor is a 50-55 kDa type I transmembrane glycoprotein of the Ig superfamily, with an extracellular domain showing 21-33% sequence identity with CTLA-4, CD28 and the inducible co-stimulatory (ICOS) molecule, but distinct function and ligand specificity (Nishimura et al., 1999; Okazaki et al., 2007). PD-1 shows an extracellular IgV region, a transmembrane domain, and an intracellular tail. The cytoplasmic domain presents two tyrosine residues: one represents an immunoreceptor tyrosine-based inhibitory motif (ITIM), and the other an immunoreceptor tyrosine-based switch motif (ITSM) that may recruit phosphatases, similar to other negative regulators (Long, 1999; Sidorenko et al., 2003).

PD-1 expression is mostly detected at the membrane surface level, where it exerts its function of inhibitory receptor, however, PD-1 expression has also been observed in the cell cytoplasm (Raimondi et al., 2006). It is as yet unclear whether cytoplasmic PD-1 exerts biological functions or merely serves as a repository of the protein, allowing an immediate response as soon as its expression at the membrane is requested. PD-1 is expressed on CD4-/CD8-thymocytes in transition to CD4+/CD8+ stage and on mature T-and B cells upon activation, while it is not detectable on resting T cells. Thus, in comparison with CTLA-4, PD-1 has a slightly broader expression profile, also present on activated myeloid lineage cells such as monocytes, dendritic cells and NK cells (Freeman et al., 2000; Greenwald et al.,

(9)

9

2005). PD-1 is quickly up-regulated on T lymphocytes after exposure to cognate antigen and its expression is controlled by several cytokines including IFN-γ, IL-2, IL-7, IL-15, and IL-21 (Kinter et al., 2008). Upon antigen clearance, PD-1 expression wanes accordingly. In normal tissues, PD-1 signaling in T cells regulates immune responses to diminish damage, and counteracts the development of autoimmunity by promoting tolerance to self-antigens.

Most biological functions of PD-1 have been elucidated by generating PD-1-deficient mice with gene knockout technology. PD-1-knockout mice spontaneously develop phenotypes of lymphoproliferative autoimmune diseases, accompanied by a marked accumulation of inflammatory cells in affected organs, including CD4+ and CD8+ T cell subsets (Latchman et al., 2004). Autoimmune manifestations in PD-1 mice are different from those observed in CTLA-4 knockout mice, which die at 3-4 weeks of age from massive lymphocytic infiltration and tissue destruction in multiple organs (Tivol et al., 1995). The PD-1 related autoimmune phenomena are overall milder and less frequent than those observed following anti-CTLA-4 therapy. Table 1 summarizes the main differences between CTLA-4 and PD-1.

Two ligands for PD-1, designated PD-1 ligand 1 (PD-L1; B7-H1/CD274) and PD-1 ligand 2 (PD-L2; B7-DC/CD273), have been identified based on the similarity to other B7 superfamilies. PD-L1 and PD-L2 are type I transmembrane glycoproteins composed of IgC-and IgV-type extracellular domains that present 40% amino acid identity sequence (Dong et al., 1999; Latchman et al., 2001; Tseng et al., 2001). In contrast to the limited expression of PD-L2 on activated macrophages (Matsumoto et al., 2004), PD-L1 is more broadly expressed on immune and non-haematopoietic cells. Specifically, PD-L1 is constitutively expressed on T-and B cells, macrophages, and dendritic cells, and is upregulated upon stimulation by proinflammatory cytokines including IFN.

Despite the fact that PD-L1 mRNA has been reported in a wide range of human healthy tissues and cell types (with high expression in placenta, heart, lung and liver, low expression in spleen, lymph nodes and thymus, and no expression in the central nervous system), constitutive PD-L1 protein is much less ubiquitous in normal tissues. PD-L1 protein has also been reported to be expressed on parenchymal cells such as pancreatic islet cells, endothelial cells, muscle cells and trophoblasts. Discrepancies between mRNA and protein expression are most likely attributable to a series of post-transcriptional controls that have not yet been completely clarified (Saresella et al., 2012).

(10)

10

PD-L1-deficient mice do not develop spontaneous autoimmune diseases, mild-to-moderate levels of lymphocyte accumulation are evident in the kidneys, liver, and lung, with a predominant CD3+/CD8+ component exhibiting significantly decreased apoptosis (Yu et al., 2004). Although the mechanism underlying selective accumulation of CD8+ T cells in PD-L1 knockout mice remains to be clarified, these findings implicate a role for PD-L1 in the maintenance of T and cell homeostasis in peripheral organs. Overall, these data support the hypothesis that the expression of this ligand in non-lymphoid tissue cells can prevent immune-mediated tissue damage (Dong et al., 2003).

PD1/PD-L1 INTERACTION: FUNCTIONAL IMPLICATION

PD-1/PD-L1 interaction inhibits T lymphocyte proliferation, survival, and effector functions, induces apoptosis of antigen-specific T cells, and promotes the differentiation of CD4+ T cells into Foxp3+ regulatory T cells. The mechanism by which PD-1 exerts the inhibitory effect has been partially explained. As soon as PD-L1 interacts with PD-1, there is a recruitment of Src homology region 2 domain-containing phosphatase-1 (SHP-1) and SHP-2, which dephosphorylate multiple members of the TCR signaling pathway. This abrogates downstream effects of T-cell activation, including cytokine production, cell-cycle progression, and the expression of survival proteins. Furthermore, the inhibition of RAS and PI3K/AK3 pathways cooperates to inhibit lymphocyte function and survival (Figure 2).

Figure 2. Upon interaction of PD-L1 with PD-1, there is a recruitment of

Src homology region

2domain-containing phosphatase-1 (SHP-1) and SHP-2, which dephosphorylate multiple members of the TCR signaling pathway. This abrogates downstream effects of T-cell activation, including cytokine production, cell-cycle progression, and the expression of survival proteins.

(11)

11

Considering that 1) PD-L1 is upregulated in hematopoietic and reticulo-endothelial cells, in response to proinflammatory cytokines such as IFN-γ, and 2) PD-1 is expressed to various degrees on activated T cells, it is reasonable to speculate that the co-expression of ligand and receptor in inflamed tissues mitigates against collateral tissue destruction by T cells at these sites. The PD-1/PD-L1 axis would therefore be important in moderating the inflammatory response and in turn potential autoimmune foci resulting from the dysregulation of the effector phase of the immune response. This mechanism is important in various physiological and pathological processes including feto-maternal tolerance, graft-vs-host disease, and various autoimmune diseases.

Another piece of the puzzle for understanding the central role of the PD-1/PD-L1 axis in the immune response derives from studies evaluating the developing of T cell exhaustion. While PD-1 is expressed on activated lymphocytes during an acute inflammatory process to limit tissue damage and, as soon as the exogenous stimulus is cleared, the PD-1 wanes accordingly, the persistent antigen exposure may prevent the down-regulation of PD-1. Since the immune response plays an important role in staving off cancer, mechanisms of immunosuppression hinder productive anti-tumor immunity. T cell exhaustion is one such mechanism. PD-1 has been identified as a marker of exhausted T cells in chronic disease states, including cancer, and blockade of PD-1/PD-L1 interactions has been shown to partially restore T cell function. The development of T cell exhaustion translates into a major immune resistance and promotes the immune evasion.

Thus, the PD-1/PD-L1 axis is of particular interest, in light of promising data demonstrating a restoration of host immunity against tumors, with the prospect of durable remissions. Indeed, remarkable clinical responses have been seen in several different malignancies including, but not limited to, melanoma, lung, kidney, and bladder cancers. Even so, determining which patients derive benefit from PD-1/PD-L1-directed immunotherapy remains an important clinical question, particularly in light of the autoimmune toxicity of these agents. An improved understanding of the host immune system and tumor microenvironment will better elucidate which patients derive benefit from these promising agents.

(12)

12 PD-1/PD-L1 IN MALIGNANT DISEASES

A lot of studies via immunoistochemistry, reveal that, PD-L1 is constitutively expressed in many human cancers (Dong et al., 2002). PD-L1 is commonly upregulated on many different tumor types including melanoma (Hino et al., 2010), ovarian cancer (Hamanishi et al., 2007), lung cancer (Konishi et al., 2004), clear cell renal carcinoma (Thompson et al., 2004), urothelial carcinoma (Nakanishi et al., 2007), squamous cell carcinomas of the head and neck (Strome et al., 2003) (Lyford-Pike et al., 2013), esophageal cancer (Ohigashi et al., 2005), cervical cancer (Karim et al., 2009) , breast carcinoma (Ghebeh et al., 2006), pancreatic cancer (Nomi et al., 2007), gastric cancer (Wu et al., 2006), Wilms tumor (Routh et al., 2008) and glioblastoma (Hirano et al., 2005) (Table 2).

Table 2. Expression of PD-L1 in human cancer tissues (Dong et al., 2002).

In addition to its expression on tumor cells, PD-L1 can be detected on cells located in the tumor microenvironment, and high levels of PD-L1 expression have been reported in TILs and tumor-associated macrophages (Cho et al., 2011; Hirano et al., 2005). However,

(13)

13

the last meaning of the PD-L1 expression on tumor cells and other cells in the microenvironment remains to be fully determined. But, the expression of PD-1 in TILs, has created an important rationale for mAb blockade of this pathway for cancer immunotherapy. PD-L1 expression on cancer cells may be an adaptive response to immune attack and induced by cytokines, or constitutive as a result of oncogenic processes, as occurs in BRAFV600 mutated melanoma. The role of oncogenes to drive immunosuppression has been suggested but not as yet clarified so far.

Interestingly, in melanoma models there is evidence that BRAFV600 mutation, along with the STAT3 signal, is essential for immune evasion by human melanomas. Recently, it has been reported that the activation of the MAPK signaling pathway in melanoma cells by oncogenic V600E BRAF leads to the production of inhibitory cytokines, interleukin 1 (IL-1)α/β (63) in melanocytes and melanoma cell lines to induce T cell suppression. Tumor-associated fibroblasts (TAFs) respond to IL-1 by upregulating an immunomodulatory transcriptional program resulting in the production of COX-2, PD-1 ligands, and chemokines, as well as in the amplification of IL-1 signaling (Figure 3).

Figure 3. The activation of the MAPK signaling pathway in melanoma cells by oncogenic BRAFV600E leads to the production of interleukin 1 (IL-1)α/β. Tumor associated fibroblasts (TAFs) respond to IL-1 by upregulating an immunomodulatory transcriptional program resulting in the production of COX-2, PD-1 ligands and chemokines as well as in the amplification of IL-1 signaling. Collectively, these factors and signaling circuitries suppress cytotoxic T lymphocyte (CTL) functions and ultimately promote tumor growth.

(14)

14

Collectively, these factors and signaling circuitries suppress cytotoxic T lymphocyte functions and ultimately promote tumor growth. These data highlight the hypothesis that both mechanisms (constitutive and induced) may concomitantly contribute to the PD-L1 expression on tumor cells and in other cells in the microenvironment, potentially fueling an immune tolerance state.

However, most of the data so far reported in literature are heterogeneous in terms of PD-L1 expression, variability in the assays and cell immunolocalization, as well as cut-off values for positive versus negative PD-L1 immunohistochemical expression. Furthermore, PD-L1 expression has been investigated in primary tumors and metastatic tissue samples, but little conflicting data is available with regard to the concordance/discordance of PD-L1 and PD-1 immunohistochemical expression in paired human tissues from primary melanomas and respective metastases. In addition, another role in this complicated puzzle is played by PD-1. Similar to exhausted virus antigen-specific T cells, which have been reported in chronic viral infections, the majority of TILs in melanoma express high levels of PD-1 compared with T cells from normal tissues and peripheral blood.

Previous studies on several human cancer types elegantly showed that overexpression of PD-L1 in tumor cells may allow immune evasion (Dong et al., 2002). When PD-L1 binds with PD-1 receptor on T-cells, T-cell function is compromised through induction of apoptosis, suppression of proliferation, and inhibition of T-cell cytokines release, such as IFN-γ, IL-4 and IL-2. Furthermore, the PD-1/PD-L1 interaction inhibits T lymphocyte proliferation, survival and cytokines secretion, promotes the differentiation of CD4positive/CD25-negative/Foxp3-negative T cells into Foxp3+ regulatory T cells (Treg), and induces apoptosis of tumor-specific T cells. Treg cells allow tumor cells to grow locally, and PD1/PD-L1 pathway on tumor cells causes immunosuppression by PD-L1-induced Treg cells.

The CD8+ TILs specific for the melanoma antigen MART-1 expresses significant levels of PD-1 compared with lower expression on MART-specific T cells from the peripheral blood of the same patients (Ahmadzadeh et al., 2009). PD-1 expression by TILs in melanoma correlated with an exhausted T cell phenotype and impaired effector function. The data reported underline again that the assessment of circulating lymphocytes does not necessarily reflect what happens in the tumor microenvironment, at the biological interface between immune system and cancer. Therefore, the evaluation of the immunological profile

(15)

15

in the circulation may be misleading in the assessment of the interaction between lymphocytes, APC, other cells in the microenvironment, and tumor cells. All the reported data suggest that the constitutively or the cytokine-induced PD-L1 expression would provide a selection advantage of cancer cells by inhibiting tumor-specific recognition and elimination by T cells.

The understanding of the molecular basis in melanoma has become increasingly important with the development of therapies aimed to hit the specific targets involved in the progression of the tumor (Flaherty et al., 2010). Of particular relevance in melanoma is the signal transduction pathway known as the mitogen activated protein kinase (MAPK), which normally regulates cell growth, proliferation, and cell differentiation (Davies et al., 2002). The aberrant activation of MAPK is present in approximately 70-80% of primary melanomas, and mutations in proteins along the RAS-RAF-MEK-ERK (MAPK) signaling are thought to be mutually exclusive (coexistent NRAS and BRAF mutations are rarely found in the same patient). These mutations have been documented in all melanoma subtypes, including cutaneous melanoma [40-50% BRAF, 15-20% NRAS, 1-3% CKIT (the latter in chronically sun-exposed skin)], mucosal melanoma (5-9% BRAF, 7-12% NRAS and 20-25% CKIT), and uveal melanoma (GNAQ and GNA11 50%) (Curtin et al., 2005). However, kit mutation frequency depends on the anatomic location of the primary tumor. Somatic mutations in the BRAF gene have been identified as the most frequent and relevant in order to develop targeted molecular therapies in melanoma. BRAF is a serine-threonine kinase, and its non-selective and highly selective inhibitors have been developed in the last ten years. BRAFV600E mutations are found especially in younger individuals, and in melanomas that originate on skin which is not chronically sun-exposed (Curtin et al., 2005). Within BRAF mutations, 84% are represented by V600E mutation, 10% by V600K mutation, and the remaining 6% includes a variety of mutations such as V600D and V600R (Sullivan et al., 2013).

Genetic mutations PD-L1/PD-1 axis could be a potential point of contact between target molecules (BRAFV600), immune evasion and tumor growth. Therefore, therapeutic antibodies that block PD-1 in order prevent the inhibitory interaction between PD-1 and PD-L1 may partially circuit the cancer immune evasion and elicit the immune system to recognize and kill tumor cells.

(16)

16

PROGNOSTIC ROLE OF PD-L1 AND PD-1 IN MALIGNANCIES

Many studies reported the prognostic role of PD-L1 and PD-1 in different tumor types. Overall, the results published so far are heterogeneous in terms of patient’ selection criteria and methodology used for immunohistochemical analysis. In particular, difficulties in comparing results arise from the following considerations:

i) available data are retrospective;

ii) patients with different tumor stages have been included; iii) primary or metastatic sites have been included;

iv) different tumor histotypes have been compared within the same studies;

v) different types of tissues have been evaluated (frozen versus paraffin-embedded samples);

vi) monoclonal and/or polyclonal antibodies have been used;

vii) both cytoplasmic and/or membranous immunostaining have been evaluated to define positivity;

viii) different scores of PD-L1 and/or PD-1 positivity have been reported.

Thus, the prognostic role of PD-L1 and PD-1 in the context of different tumor types requires further investigation and only future studies will establish whether PD-L1 is an immune correlate and predictive marker for response to anti PD-1 antibody.

The role of PD-L1 in melanoma patients has been reported in three different studies. Hino and coll. evaluated the intensity of PD-L1 expression in melanoma specimens (59 primary tumors, 16 lymph nodes, and 4 in-transit metastases). By multivariate analysis, immunohistochemical PD-L1 expression correlated with Breslow thickness, disease free, and overall survival (Hino et al., 2010). Gadiot and colleagues evaluated paraffin-embedded tissues of 63 patients with stages III-IV melanoma diagnosed by the Netherlands Cancer Institute between 2000 and 2004. PD-L1 was analyzed in benign nevi (n=10), primary melanomas (n=43), satellite metastases (n=8), in transit-metastases (n=21), and distant organ metastases (n=22). Interestingly, the authors showed that 20% of the benign nevi samples were PD-L1-positive, while only 5% of primary melanomas were PD-L1 positive. Satellite metastases, in-transit metastases, lymph nodes and distant metastases were PD-L1 positive in

(17)

17

25%, 40%, 14%, and 18% of cases, respectively. PD-L1 was more expressed in satellite and in-transit metastases compared with lymph node, and distant metastases. In this study, PD-L1 did not correlate with disease free or overall survival (Gadiot et al., 2011).

Finally, Taube JM et al. demonstrated a correlation between PD-L1 and the presence of TILs in both nevi and malignant melanoma: 98% of PD-L1 positive melanomas were associated with TILs compared with only 28% of PD-L1 negative melanomas. Furthermore, PD-L1 positive melanocytes were almost always localized adjacent to TILs. Interestingly, a positive correlation between surface PD-L1 expression by tumor cells (≥ 5%) and overall survival in metastatic disease in 56 patients was reported. On the other hand, there was no significant difference in survival in relation to PD-L1 expression in primary melanoma (Taube et al., 2012).

There are important differences in the first two above referenced studies (Gadiot et

al., 2011; Hino et al., 2010). PD-L1 was evaluated with different immunostaining protocols.

In Hino’s study, it is unclear whether the authors tested their antibodies against isotype controls and PD-L1Fc protein blockade to ensure lack of background or unspecific signals. Another likely explanation for the conflicting results could be that small series of fewer than 100 melanoma patients from diverse origin were analyzed. The most noticeable difference between the Asian melanoma patient population (Hino et al., 2010) and the cohort population evaluated by Gadiot is the tumor type. In the latter, superficial spreading and nodular melanomas were predominantly diagnosed, whereas in the cohort of Hino et al, mainly patients with acral lentiginous melanomas were included. The latter have been shown to preferentially express c-kit amplifications or mutations, whereas the melanomas from the European cohort are more likely to express BRAF mutations. Finally, more women were included in Hino’s study. Women are known to have a better prognosis in melanoma and may have influenced the prognostic analysis of PD-L1.

Finally, Taube et al. suggest that patients with both PD-L1 expression and TILs in melanoma may have improved prognosis compared with the group with PD-L1 expression without TILs (Taube et al., 2012). This study seems to suggest that the evaluation of PD-L1 expression is important but not sufficient to understand the intricate interplay between melanoma and the immune system. PD-L1 overexpression in melanoma with TILS could be suggestive of an active and efficient immune response, and therefore may correlate with a better prognosis.

(18)

18

CLINICAL TRIALS AND FUTURE DEVELOPMENTS

Table 4 summarized the main prospective phase 1 studies evaluating anti PD-1/PD-L1 antibodies in cancer patients. Topalian and coll. enrolled 296 patients with advanced melanoma, non-small-cell lung cancer, castration-resistant prostate cancer, or renal-cell or colorectal cancer to receive BMS-936558 (anti-PD-1 antibody) at a dose of 0.1 to 10.0 mg/kg of body weight every 2 weeks for 8 week-treatment cycles for 12 cycles until progression or complete response (Topalian et al., 2012). Common treatment-related adverse events included fatigue, rash, diarrhea, pruritus, decreased appetite, and nausea. Grade 3 or 4 drug-related toxicities occurred in 14% of patients; moreover, there were three deaths from pulmonary toxicity. Drug-related serious adverse events occurred in 32 of 296 patients (11%). The spectrum, frequency, and severity of treatment related adverse events were similar across the dose levels tested. Drug-related adverse events of special interest (e.g., those with potential immune-related causes) included pneumonitis, vitiligo, colitis, hepatitis, hypophysitis, and thyroiditis. In this trial, objective responses were observed in 28% of melanoma patients. Interestingly, partial sustained, long term responses were observed at each cohort level (0.3 mg/kg to 10 mg/kg). At a dose of 3.0 mg/kg, objective responses were noted in 7 of 17 patients (41%). As measured by standard RECIST criteria, objective responses were long lasting, with response durations of 1 year or more in 13 out of 26 patients who had a response with 1 year or more of follow-up. In the same study immunohistochemical analysis was performed on pretreatment tumor specimens obtained from 42 patients. Objective response were reported in a subgroup of patients (9/25 patients: 36%) with PD-L1-positive tumors. These preliminary data would suggest a relationship between PD-L1 expression on tumor cells and objective response.

Another study reported the activity and safety of BMS-936558 in patients with previously treated advanced melanoma (Taube et al., 2012). BMS-936558 was administered every two weeks at doses of 0.1 to 10 mg/kg during dose-escalation and/or cohort expansion. Patients received up to 12 cycles (4 doses/cycle) of treatment or until PD or CR. Ninety-five melanoma patients with Eastern Cooperative Oncology Group (ECOG) Performance Status (PS) ≤ 2 were treated with BMS-936558 at 0.1 (n=13), 0.3 (n=17), 1 (n=28), 3 (n=17), or 10 mg/kg (n=20). The majority of patients (60/95) had received interferon-alpha or IL-2 (prior anti-CTLA-4 excluded). Seven out of 95 patients previously received B-raf inhibitor therapy.

(19)

19

Sixty out of 95 patients had previously received more than 2 lines of treatment. The incidence of grade 3-4 related adverse events (Aes) was 19% and included gastrointestinal (4%), endocrine (2%), and hepatobiliary disorders (1%). There were no drug-related deaths. Clinical activity was observed at all dose levels, including patients with visceral or bone metastases. Of 20 patients with objective response, 12 had object response duration ≥1 year, and 6 patients were on study with objective response duration between 1.9 and 11.3 months. Several patients had prolonged stable disease.

Recently Hamid et al. evaluated the anti-PD-1 antibody lambrolizumab at a dose of 10 mg/kg of body weight every 2 or 3 weeks or 2 mg/kg every 3 weeks in 135 patients with advanced melanoma (Hamid et al., 2013). The confirmed response rate across all dose cohorts was 38%, with the highest confirmed response rate, 52%, observed in the cohort that received 10 mg/kg every 2 weeks. The response rate did not differ significantly between patients pretreated or not with ipilimumab. At a median follow-up of 11 months, among patients achieving a response to treatment, long term responses were reported in 81% of patients. The overall median progression-free survival among the 135 patients was longer than 7 months. Common adverse events treatment-related were fatigue, rash, pruritus, and diarrhea; most of the adverse events were low grade.

Finally, Brahmer et al. evaluated 207 patients (75 with non-small cell lung cancer, 55 with melanoma, 18 with colorectal cancer, 7 with gastric cancer, and 4 with breast cancer), using anti-PD-L1 antibody (BMS-936559) in a multicenter Phase I trial at multiple escalating dose (from 0.3 to 10 mg/kg) (Brahmer et al., 2012). Anti-PD-L1 antibody was administered every 14 days in 6 week cycles, for up to 16 cycles or until the patient had a complete response or confirmed disease progression. Grade 3-4 immune-related toxic effects occurred in 9% of patients. An objective response, complete or partial, was observed in 9/52 patients with melanoma (29% response rates at 3 mg/kg), 2/17 with renal cell cancer, 5/49 with non-small cell lung cancer (mostly non squamous subgroup) and 1/17 with ovarian cancer. Prolonged stabilization of disease was observed for 12-41% lasting at 24 weeks (Brahmer et al., 2012).

Overall, with anti PD-1 or anti PD-L1 antibodies, the incidence of serious (grade 3 to 4) adverse effects was at a range similar to that with CTLA-4-blocking antibodies, but most were less clinically significant with the exception of pneumonitis, which led to death as a result of toxicity. Blocking the ligand PD-L1 with the fully human IgG4 antibody

(20)

20

BMS-936559 resulted in a slightly lower frequency of objective tumor responses and also fewer adverse effects in phase I testing. These initial experiences, together with the current initial clinical testing of a series of other antibodies and blocking constructs to the PD-1/PD-L1 axis, provide an impetus for the continued clinical testing of these highly active immunotherapies for melanoma.

Based on experimental data, blockade of B7-H1 or PD-1 is not expected to stimulate

de novo immune responses but rather to enhance ongoing immune responses against tumor

antigens, since this axis is implicated in the effective phase and not in the priming phase of the immune response. A combinational strategy to elicit the immune response could improve the objective response rate as well as the duration of response. In patients with advanced disease it’s likely that the majority of TILs have an “exhausted” phenotype. The presence of inhibitory molecules would render these cells insensitive to the action of antiPD-1. On the basis of these considerations, preliminary results in mice models suggest that combination therapies may be helpful in restoring an activating phenotype in the tumor microenvironment (Mkrtichyan et al., 2012). In experimental models combining B7-H1/PD-1 blockade with cancer vaccines (Li et al., 2007; Pilon-Thomas et al., 2010; Rosenblatt et al., 2011; Slingluff, 2011), adoptive transfer of preactivated T cells, or T cell stimulation with anti-CD137 (Palazon et al., 2011) often provides dramatic synergistic antitumor effects, in some cases eradicating well-established tumors.

In addition, there is preclinical evidence that CTLA-4 and PD-1 could play complementary roles in regulating adaptive immunity. Curran et al. reported, in mice xenograft, that the combination of CTLA-4 and PD-1 blockade determines the accumulation of CTLA-4/PD-1 double-positive T effector cells within B16 melanoma cell lines (Curran et

al., 2010). These data suggest that T cells that would otherwise be functionally and

proliferatively repressed are instead able to continue expanding and carrying out effector functions. In the same model the combination of CTLA-4 and PD-1 blockade resulted more than twice as effective as either alone in promoting the rejection of B16 melanoma cell lines. Interestingly the further addition of PD-L1 antibody elevates the rate of tumor-free survival to 65% vs. 10% with CTLA-4 blockade alone (Curran et al., 2010).

Recently, on the basis of these observations, a phase 1 study to investigate the safety and efficacy of combined CTLA-4 and PD-1 blockade [with the use of ipilimumab (anti-CTLA4 antibody) and nivolumab (anti-PD-1 antibody), respectively] in patients with

(21)

21

advanced melanoma has been conducted (Wolchok et al., 2013). In this study the objective-response rate for all patients in the concurrent-regimen group was 40%. At the maximum tolerated doses (nivolumab at a dose of 1 mg/kg of body weight and ipilimumab at a dose of 3 mg/kg), 53% of patients had an objective response, all with tumor reduction of 80% or more. Grade 3 or 4 toxicities occurred in 53% of patients and were generally reversible and qualitatively similar to those reported with monotherapy (ipilimumab or nivolumab alone).

(22)

22

Table 4. Main prospective phase 1 studies evaluating anti PD-1/PD-L1 antibodies in cancer patients.

(23)

23 AIM OF THE STUDY

In its early stages melanoma can be cured by surgical resection, but once it progresses to the metastatic stage it remains an incurable disease. The finding of somatic mutations in the BRAF oncogene in ~40-50% of melanoma patients paved the way to the introduction of BRAF inhibitors (BRAFi) as a standard treatment in locally advanced or metastatic melanomapatients with BRAFV600 mutation. While clinical responses to BRAFi may be dramatic and some patients treated with BRAF and MEK inhibitors (MEKi) may stay in remission for years, the median duration of response is 7 months for patients treated with BRAFi alone and 11 months for patients treated with BRAFi and MEKi. For this reason, there is intense investigation into alternative or complementary therapeutic strategies, including novel immunomodulatory agents.

A dynamic interplay exists between host and tumor, and the ability of the tumor to evade immune recognition often determines the clinical course of the disease. Significant enthusiasm currently exists for a new immunotherapeutic strategy: the use of immunomodulatory monoclonal antibodies that directly enhance the function of components of the anti-tumor immune response such as T cells, or block immunologic checkpoints that would otherwise restrain effective anti-tumor immunity. This strategy is based on the evidence that development of cancer is facilitated by the dis-regulation and exploitation of otherwise physiological pathways that, under normal circumstances, down-regulate immune activation and maintain tolerance to self.

Among these pathways an important role is covered by the PD-1/PD-L1 axis. An emerging concept in cancer immunology is that inhibitory ligands such as PD-L1 are induced in response to immune attack, a mechanism termed “adaptive resistance”. This potential mechanism of immune resistance by tumors suggests that therapy directed at potentiating the antitumor T-cell response a tumor-specific level, by impairing the interaction of the inhibitory receptor PD-1 on T cells with PD-L1 expressed on tumor cells might synergize with other treatments that enhance endogenous antitumor immunity.

Different studies reported the anti-PD-1 and anti-PD-L1–directed therapies showing significant clinical promises.

In agreement with the proposed function of the PD-1/PD-L1 axis in the induction and maintenance of peripheral tolerance, surface expression of PD-L1 in some tumors has been

(24)

24

reported to be an independent predictor of adverse clinical outcome. Furthermore, expression of the molecule appears to correlate with response to treatment, at least in some tumor models. However, the prognostic significance of PD-L1 expression in melanoma remains incompletely explored.

In our work, primarily, we focused our attention on the study of the role of PD-L1 and its expression in a cohort of metastatic melanoma patients (MMP) to assess and better defined its role in the prognosis of metastatic melanoma. Then by using in vitro and in vivo experimental setting, we also evaluated the genetic, phenotypic, morphologic and functional characteristics of a melanoma cell line expressing PD-L1.

(25)

25 MATERIALS AND METHODS

Patient characteristics

The study cohort comprised 81 consecutive cutaneous MMP referred to Papa Giovanni XXIII Hospital between July 1994 and April 2013. Data on treatment and survival were collected prospectively. When possible, paired samples of the primary lesion and of synchronous or asynchronous metastases were collected. Histopathological classification was confirmed by medical records, and/or review of pathology material. The study protocol was approved by the local Ethical Committee.

Immunohistochemistry for PD-L1

For immunohistochemical evaluation of PD-L1 on formalin-fixed paraffin-embedded (FFPE) specimens, we first tested selected melanoma cases and control tissues with three commercially available antibodies: a mouse monoclonal (clone 27A2, MBL, Woburn, MA, dilution 10 μg/mL); and two rabbit polyclonal antibodies (LSBio LifeSpan Biosciences, Inc. Seattle, WA, dilution 5 μg/mL and ab58810, Abcam, Cambridge, UK, 5 µg /ml). In addition, the mouse anti-human monoclonal antibody 5H1 (isotype mouse IgG1), kindly provided by Dr. L. Chen (New Haven, CT, USA) was used. We found that the MBL clone 27A2, used according to standard protocol and commonly employed for paraffin material, gave a high background that could be falsely interpreted as positive staining. The LSBio polyclonal antibody was associated with lower background but diffuse nuclear staining was observed. Only the 5H1 monoclonal antibody and the ab58810 rabbit polyclonal antibody fulfilled acceptable requirements for a reliable interpretation and were comparatively used to analyze the whole cohort of specimen (Flaherty et al., 2010; Garbe et

al., 2010). Isotype control staining for both selected antibodies was included. The

performance of PD-L1 immunostaining in remotely archived paraffin embedded tissues was similar to that observed in more recent specimens.

Interpretation of immunohistochemical analyses

Sections were initially evaluated and scored by a dermatopathologist (D.M.) and then independently reviewed by a second pathologist (L.C.), both blinded to clinical outcome.

(26)

26

Staining intensity was scored as follows: 0 (no staining), 1+ (weak), 2+ (moderate), or 3+ (strong). Samples were considered positive if ≥5% of the tumor cell population showed 2+ or 3+ membrane staining {Garbe, 2010 #117; (Hauschild et al., 2012); Salama, 2013 #118} (Topalian et al., 2012). In addition the tumor cell population was tabulated according to the absolute percentage of cells showing membrane staining in order to evaluate PD-L1 expression as a continuous variable. Cytoplasmic cellular staining had to unequivocally exceed background to be considered positive. In the final tabulation, cases with cytoplasmic staining only were plotted separately. In the majority of cases, comparison of the monoclonal and polyclonal antibodies yielded comparable results in the percentage of immunostained tumor cells. In cases with discrepancies of interpretation (n=7; 8.6%), a case was classified according to the results obtained using the 5H1 monoclonal antibody. The rabbit polyclonal ab58810 (Abcam, Cambridge, UK) and the mouse monoclonal (mAb) 5H1 [L. Chen (New Haven, CT, USA)] (1) were selected for evaluation of the cohort of melanoma lesions. Due to limited material, 8/81 cases were immunostained with the 5H1 mAb only. For two patients membrane immunostaining was not feasible for technical reasons.

DNA extraction from formalin-fixed paraffin-embedded tissues and B-RAF mutation detection

Genomic DNA was isolated from FFPE tissue samples (with at least 80% of neoplastic cells by light microscopy) using the Qiamp DNA FFPE tissue kit (Qiagen, Hilden, Germany). After amplification, the whole exon 15 of the BRAF gene was sequenced by an automated capillary sequencer (3500 DX Genetic Analyzer, Life Technologies) (Hamid et

al., 2013).

mRNA extraction and detection by qRT-PCR and Real Time

Total RNA was extracted from air dried deparaffinized sections by using TRIZOL (Life Technologies). Real Time PCR was performed on Qiagen RotorGene (Qiagen, Milan, Italy) and the level of PD-L1 mRNA calculated as the ratio of PD-L1 mRNA level and 18S housekeeping gene level for each sample. Primers, designed using Primer Blast (http://www.ncbi.nlm.nih.gov/tools/primer-blast/), were: set 1, 18S -FW (forward) 5'

(27)

27

5'-CCTCCAATGGATCCTCGTTA-3', and set 2, PD-L1-FW (forward)

5'-TACTGGCATTTGCTGAACGC 3', PD-L1-RE (reverse)

5'-CTTGTAGTCGGCACCACC. qRT-PCR to determine PD-L1 mRNA levels in cell lines was performed as previously described (Brahmer et al., 2012) and standardized over actin levels.

Cell lines

Cell lines used were: A375, COLO-38, FO-1, SKMEL5, MEL4778D and M14. Cells were cultured in RPMI-1640 added with 10% fetal calf serum (all reagents from Sigma, Milan, Italy, referred to as complete medium) and passaged every 2-3 days. For 3D cultures cells (15x103) were seeded in complete medium added with Matrigel™ (Corning, Milan, Italy) at a 1:3 final concentration. Cells were cultured for 5 days before further analyses. The PD-L1+ subpopulation in the A375 was separated by using immunomagnetic beads (Miltenyi Biotec, Calderara di Reno, Italy) or cell sorting using a FACSAria III (BD Biosciences, Milan, Italy).

Other antibodies for in vitro experiments

Antibodies used were: anti-PD-L1, -PD-L1-PE, -PD-L1-PE-Cy7, anti-CD56-PE and anti-CD74-FITC (all from eBioscience, San Diego, CA). Anti-HLA-Class II was a kind gift of Prof. F. Malavasi (Turin, Italy) and was highlighted using a PE-labeled goat anti-mouse IgG (Life Technologies, Monza, Italy). Anti-integrin-α1, - α2, - α3 and - α5 rabbit sera were a kind gift of Prof. G. Tarone (Turin, Italy) and were highlighted using an alexa488-labeled goat anti-rabbit Ig. Anti-caveolin1, -pERK1/2 and -ERK1/2 antibodies were from BD Biosciences (Milan, Italy), while anti-pJNK, -JNK, -pp38 and -p38 were from Cell Signaling Technology (Danvers, MA). A donkey anti-rabbit Ig (Santa Cruz Biotechnology, Santa Cruz CA) and a goat anti-mouse IgG (Perkin Elmer, Monza, Italy) horse radish peroxidase (HRP)-conjugated antibodies were used as secondary antibodies in western blot experiments.

Induction of PD-L1/CD274 expression by lentiviral technology

Lentiviral particles containing the genetic material for PD-L1 (Origene, Rockville, MD) were generated as described (Crawford et al., 2007). A375 cells were infected with

(28)

28

PD-L1 lentiviral particles and expression of the molecule monitored 48 h later by flow cytometry. To obtain stable PD-L1+ clones, cells were sorted with a BD FACSAria III (BD Biosciences) gating on GFP+ population and directly seeded in 96w plates.

Xenograft models

PD-L1- and PD-L1+ A375 cells (107) were injected subcutaneously in the presence of Matrigel into the right and left flanks, respectively, of 6- to 8-week-old male NOD/SCID mice (Charles River, Calco, Italy). After two weeks mice were sacrificed and tumor size measured. Lesions were then partly fixed and processed for histopathological studies and partly dissociated for cytofluorimetric analyses and culture.

Flow cytometry

Stained cells were analyzed using a FACSCantoII (BD Biosciences), equipped with the Diva7 software, acquiring at least 10,000 events per sample.

Western blot

Cells were lysed as previously described (Thompson et al., 2004; Yamazaki et al., 2002). Proteins were resolved by SDS-PAGE and transferred to nitrocellulose membranes (Life Sciences, Bio-Rad, Segrate, Italy). Image acquisition and densitometric analyses were performed using ImageQuant LAS4000 and TL Version 7.0 software (GE Healthcare, Milan, Italy). The band quantification of phosphorylated proteins is made by normalizing over the corresponding non-phosphorylated protein; total proteins are quantified on actin.

Confocal microscopy

Unconjugated anti-PD-L1 and Alexa488-labeled goat anti-mouse antibodies were used to determine PD-L1 expression. Alexa-568-conjugated phalloidin (Life Technologies) and DAPI (4,6 diamidino-2-phenylindole) were used to counterstain. Slides were mounted in SlowFade Gold reagent (Life Technologies). Cells were analyzed using a TCS SP5 laser scanning confocal microscope with four lasers (Leica Microsystems, Milan, Italy). Image

(29)

29

acquisition and processing is detailed in SMM were acquired with the LAS AF software (version Lite 2.4) and processed with Adobe Photoshop (Adobe Systems, San Jose, CA, USA). Pixel intensity and raw integrated density analyses were performed using ImageJ

(http://rsbweb.nih.gov/ij/).

Wound healing assay

A375 cells were seeded in 6-well plates (5x105/well) and incubated in complete medium overnight. Cells were then treated with 10 µg of Mitomycin C (Sigma) for 20 minutes. Wounds were made with a 200 µl tip and the wells were washed several times to remove all non-adherent cells. Wound repair was documented at 24, 48 and 72 h using a DMI 3000 B optical microscope (Leica Microsystems), equipped with a DCF 310 FX digital camera and LAS Version 3.8 software.

Invasion assay

A375 cells were seeded in 100 µl of Matrigel in the upper chamber of a Boyden chamber (2x104/well, Corning, Milan, Italy). RPMI + 20% FCS was added in the lower chamber as attractant. At the end of the incubation period (72h at 37°C), Matrigel in the upper chamber was wiped off with a cotton swab. The cells that had traversed Matrigel and spread on the lower surface of the membrane were fixed in 4% formaldehyde (Sigma) and stained with haematoxylin and eosin. Slides were analyzed using a DMI 3000 B optical microscope (Leica Microsystems). At least 5 different fields/well were photographed and cells counted by three independent observers.

Gene expression profiling and analysis

RNA was extracted using RNeasy Plus Mini kit (Qiagen) and quality checked with the Agilent 2100 Bioanalyzer (Agilent Technologies Italia, Cernusco sul Naviglio, Italy). For each sample RNA (300 ng) was amplified and biotinylated using Illumina® TotalPrep RNA Amplification Kit (Life Technologies) and quantified using a NanoDrop ND-8000 spectrophotometer (Thermo Scientific, Milan, Italy). cRNA (750 ng) was hybridized to HumanHT-12 v4 Expression BeadChips (Illumina, Milan, Italy) using Whole-Genome Expression Direct Hybridisation kit (Illumina) and scanned with iScan System (Illumina).

(30)

30

Analysis of the scanned images and determination of the detection call for each probe set of the array was performed using the GenomeStudio Gene Expression Module Software (Illumina). Raw data were exported applying a standard quantile normalization. Expression values were background subtracted and filtered against a p-value threshold of 0.01. Clustergram analysis of the whole probe-set was performed using the clustergram tool of the of MATLAB bioinformatics package (MathWorks, Natick, MA) using a Pearson correlation coefficient as metric. Differentially expressed lists (derived by comparing mean signals from three RNA replicates for each cell line) were built on probes which: (i) passed the p-value threshold an all experiments, and (ii) whose minimum fold change was larger than the one in absolute value (|logFc|>1). Venn diagrams were then created using Venny free on-line tools (http://bioinfogp.cnb.csic.es/tools/venny/) to visualize intersections between class comparison results and to select the sequences of interest. For enrichment analysis, differentially expressed genes were classified according to their gene ontology (GO) annotations and cellular pathway association according to Kyoto Encyclopedia of Genes and Genomes (KEGG) database, using tools freely available online. These analyses used a cumulative hypergeometric distribution and a cut-off on Bonferroni-corrected p-value (0.05).

Statistical analysis

Disease-free survival (DFS) was calculated as the time from first melanoma diagnosis to first recurrence or death. Overall survival (OS) was calculated as the time from first melanoma diagnosis to death. Patients who have neither recurred nor died at the end of the study were censored at the time of the last follow-up. Patients who were not reported as having died at the time of the analysis will be censored at the date they were last known to be alive. Survival curves were estimated with the Kaplan-Meier method. Cox proportional-hazards models were used for univariate and multivariate analyses. Results are expressed as hazard ratios (HR) with 95% confidence intervals (95%CI). PD-L1 was tested as a continuous or as a dichotomous variable using the 5% cut-off reported in literature (Garbe et al., 2010; Hauschild et al., 2012; Salama et al., 2013; Townsend et al., 1993). In the current study we applied the recursive partitioning analysis (RPA), a statistical method that allowed us to identify 17.5% as the cut-off that best predicted prognosis in our series.

(31)

31

Chi-square test and t-test were used to analyze the association between PD-L1 status and other clinical and pathological features. Statistical analyses were carried out using SAS version 9.1 (SAS Institute, Cary, NC) and R language environment for statistical computing (open source, www.r-project.org). The inter-rater agreement was test by Cohen’s Kappa. The kappa-statistic measure of agreement is scaled to be 0 when the amount of agreement is what would is expected to be observed by chance and 1 when there is perfect agreement. For intermediate values, the measurement scale made by Landis and Koch was used for interpretations. P values less than 0.05 were considered significant.

(32)

32 RESULTS

PD-L1 expression in melanoma tissues

PD-L1 expression was studied by IHC in 81 consecutive, well characterized MMP. PD-L1 expression was considered either as a continuous or discontinuous variable using the 5% published (Dong et al., 2002) or the 17.5% cut-off, determined here by recursive partitioning analysis (RPA), yielding comparable results. In 73/81 MMP (90%) both ab58810 polyclonal (Abcam) and 5H1 monoclonal anti-PD-L1 antibodies were tested, showing a high concordance rate (Cohen’s Kappa 72%, p<0.0001). Representative images of PD-L1 immunostaining in melanoma tissues are illustrated in (Figure 4-7)

Figure 4. PD-L1 immunohistochemical expression using the mouse anti-human B7-H1 mAb (clone 5H1). (A) Staining of placental tissue, used as positive control. (B) PD-L1-negative melanoma. (C–I) Examples of PD-L1 staining in melanoma tissues. Scattered PD-L1+ melanoma cells (C); focal and patchy PD-L1 staining (D); scattered melanoma cells showing PD-L1 expression and melanin granules (E); roundish aggregate of melanoma cells with moderate PD-L1 membrane expression (F); strong PD-L1 immunostaing in melanoma cells easily recognizable from intra and extracellular granular melanin depositions (G); (H–I) staining of melanoma metastases. Note that cytoplasm of melanoma cells is pigmented due to abundant melanin deposition (original magnification ×40, scale bar 50 μm).

(33)

33

Figure 5. Reactivity of polyclonal Abcam PD-L1 (D) vs. 5H1 PD-L1 (E) antibodies on serial melanoma sections. A) Haematoxylin & eosin stain; B) isotype control using purified normal rabbit IgG1 for polyclonal Abcam PD-L1 antibody; C) isotype control using purified normal mouse IgG1 for monoclonal 5H1 PD-L1 antibody. (Original magnification x40, scale bar 50 μm).

Figure 6. Representative PD-L1+ melanoma specimens (monoclonal 5H1 PD-L1 antibody) showing different intensity of staining, graded as A) weak staining; B) moderate, focally strong, staining; and C) diffusely strong staining. (Original magnification x40, scale bar 50 μm).

(34)

34

Figure 7. Representative staining patterns with monoclonal 5H1 PD-L1 (upper row) vs. polyclonal Abcam PD-L1 antibodies (lower row) on serial sections from melanoma specimens. Note that the two antibodies differ for cellular localization: preferential (granular) cell membrane staining (monoclonal 5H1 antibody) vs. cell membrane and endomembrane system (cell surface enhancement), with variable cytoplasm reactivity and occasional low to moderate background (polyclonal Abcam antibody). (Original magnification x40, scale bar 50 μm).

Complete clinico-pathological information of the antecedent primary melanoma was available for 81 patients (Table 5).

Table 5. Patient demographics and clinical characteristics of the antecedent primary melanoma according to membrane IHC PD-L1 status (clone 5H1 PD-L1 monoclonal antibody, cut-off 5%).

PD-L1- (n=51) PD-L1+ (n=30) p N° patients % N° patients % Gender Female Male 30 21 58.8 41.2 15 15 50 50 .4431

(35)

35 Age, years Median Mean Q1-Q3 57 57.8 50-70 57.5 58.6 48-75 .8387 Breslow thickness, mm Median Mean Q1-Q3 3.5 4.6 2.1-6.2 4.4 5.4 2.8-7.1 .4674 Histotype SSM (ref.) NM ALM Other Not applicable 30 10 4 2 5* 58.8 19.7 7.8 3.9 9.8 20 7 1 0 2* 67 23 3 0 7 .5630 Mitotic rate Mitoses absent Mitoses present 4 40 8 78 1 24 3 80 .4364

(36)

36 Not known 7 14 5 17 Lymph Nodes Positive Negative Not known 28 5 18 55 10 35 20 1 9 67 3 30 .2407 Ulceration Absent Present Not known 15 30 6 29 59 12 7 20 3 23 67 10 .5118 Therapy No Yes 8 43 16 84 9 21 30 70 .1290 Adjuvant treatment Yes 12 24 6 20 .7139 I line treatment Yes 39 76 20 67 .3411 Type of I line BRAF inhibitor Chemotherapy Ipilimumab 17 21 1 44 54 3 10 9 1 50 45 5 .7618 II line treatment .3320

(37)

37 Yes 19 37 8 27 Type of II line BRAF inhibitor Chemotherapy Ipilimumab 2 11 6 11 58 32 2 4 2 25 50 25 .1781

*Seven patients, 3 PD-L1+ and 4 PD-L1- respectively, had unknown primary melanoma. SSM: Superficial spreading melanoma; NM: Nodular melanoma, ALM: Acral lentiginous melanoma

No meaningful associations were highlighted between PD-L1 expression and the degree of tumor infiltrating lymphocytes, scored according to the current classification [brisk, non-brisk and absent (Tables 6,7)]. These results were consistent when PD-L1 was considered as continuous variable or with the 17.5% cut-off determined by RPA.

Table 6. Correlation between IHC PD-L1 membrane expression and tumor infiltrating lymphocytes (TILs) in the training patients’ cohort of 52 primary melanomas (clone 5H1 PD-L1 monoclonal antibody, cut-off 5%). PD-L1 TILs p Frequency Percentage Row percentage Column percentage 0* 1* 2* Total PD-L1- 8 15.38 25.81 57.14 17 32.69 54.84 62.96 6 11.54 19.35 54.55 31 59.62 0.8723 PD-L1+ 6 11.54 28.57 42.86 10 19.23 47.62 37.04 5 9.62 23.81 45.45 21 40.38

(38)

38 PD-L1 TILs p Frequency Percentage Row percentage Column percentage 0* 1* 2* Total Total 14 26.92 27 51.92 11 21.15 52 100.00 Frequency Missing = 1

0*: TILs absent; 1*: TILs non brisk; 2*: TILs brisk

Table 7. Correlation between PD-L1 membrane expression and tumor infiltrating lymphocytes (TILs) in the independent patients’ cohort of 17 primary melanomas (clone 5H1 PD-L1 monoclonal antibody, cut-off 5%). PD-L1 TILs p Frequency Percentage Row Percentage Column Percentage 0* 1* 2* Total PD-L1- 4 23.53 33.33 80.00 6 35.29 50.00 66.67 2 11.76 16.67 66.67 12 70.59 0.8674 PD-L1+ 1 5.88 20.00 20.00 3 17.65 60.00 33.33 1 5.88 20.00 33.33 5 29.41 Total 5 29.41 9 52.94 3 17.65 17 100.00 Frequency Missing = 1

(39)

39

The frequencies of PD-L1 membrane immunohistochemical positivity in paired primary and metastatic tumor samples are reported in Table 8. Overall, 40.3% of metastatic melanoma samples were positive to PDL-1 (PD-L1+), as compared to 14% of primary melanomas (p=0.001, Table 9A,B). The PD-L1 positivity barely correlated with BRAF mutation, although did not reach statistical significance (p=0.051, Table 10).

Table 8. Frequencies of PD-L1 membrane immunohistochemical positivity in paired primary and metastatic tumor samples (clone 5H1 PD-L1 monoclonal antibody, cut-off 5%).

N Row Percentage PD-L1- Melanoma Metastasis PD-L1+ Melanoma Metastasis TOTAL PD-L1- primary melanoma 34 66.7 17 33.3 51 PD-L1+ primary melanoma 0 00.0 5 100.00 5 TOTAL 34 60.7 22 39.3 56

(40)

40

Table 9 A. PD-L1 immunohistochemical expression in metastatic melanoma samples (clone 5H1 PD-L1 monoclonal antibody, cut-off 5%).

PD-L1 melanoma metastatis N %

Negative (<5%) 43 59.7

Positive (5%) 29 40.3

Total 72 100

Table 9B. PD-L1 immunohistochemical expression in primary melanomas (clone 5H1 PD-L1 monoclonal antibody, cut-off 5%).

PD-L1 Primary melanoma N %

Negative (<5%) 56 86

Positive (5%) 9 14

(41)

41

Table 10. Frequencies of PD-L1 membrane expression in BRAF mutated versus BRAF wild type melanomas (clone 5H1 PD-L1 monoclonal antibody, cut-off 5%).

N Percentage PD-L1- membrane PD-L1+ membrane TOTAL p BRAF WT 23 45.1 7 23.3 30 0.0516 BRAF MUT 28 54.9 23 76.7 51 TOTAL 51 30 81

Lastly, PD-L1 mRNA levels did not correlate with membrane protein expression in 34 samples analyzed (Cohen’s Kappa <3%, p=0.4273, Table 11 and Fig. 8), suggesting that protein expression must be used for diagnostic and prognostic purposes.

Table 11. Frequencies of mRNA and protein PD-L1 membrane expression in melanoma tissues from 34 patients (clone 5H1 PD-L1 monoclonal antibody, cut-off 5%).

N Row percentage Column percentage PD-L1 mRNA Negative PD-L1 mRNA Positive TOTAL PD-L1 protein Negative 12 48.00 75.00 13 52.00 72.22 25 100.00 73.53 PD-L1 protein Positive 4 44.44 25.00 5 55.56 27.78 9 100.00 26.47 TOTAL 16 47.06 100.00 18 52.94 100.00 34 100.00 100.00

Riferimenti

Documenti correlati

In relazione al secondo approccio, ricerche di tipo longi- tudinale hanno evidenziato come i principali predittori della violenza tra partner coincidano essenzialmente con

This is explained in detail in Section 1 where we point out how it opens the way to the defini- tion by Nash-Williams [ NW65 ] of the concept of better-quasi-order (bqo): a

In occasione della Giornata Mondiale di lotta contro l’AIDS, AMREF ricorda alla comunità internazionale – e al Governo italiano in particolare – che è tempo di spezzare il

Finché un giorno la magistratura scoperchia il pentolone dei grandi eventi trasformati in emergenze nazionali per meglio distribuire montagne di quattrini in deroga a leggi e

In the next section we will demonstrate that this enhancement of the radiatively-induced operators can be significant, particularly in the case of Majorana fermion DM, and lead

(a) Sequences of I-V curves collected at the same position on the surface of a 100 nm Pt/300 nm SiO 2 /Si sample, using a conductive atomic force microscopy (CAFM) connected to

Zarnqoshnm hr oqnctbdc ax Z qdZkl ne Zsnlhb sqZmrhshnmr sgZs qdrtkshm Z rdqhdr neZarnqoshnm khmdr eqnl hnmhydc dkdldmsr rdd&amp;

«Mi ha sconcertato,» gli disse, «che, mentre tutti i miei cavalieri sono partiti alla svelta in cerca della principes- sa, il mio miglior cavaliere sia rimasto nel suo castello come