CTLA4

CTLA4 is the archetypal immune regulatory checkpoint, expressed exclusively on T cells. Although first recognized on cytotoxic CD8+ lymphocytes, its main effect is mediated through profound changes in CD4+ helper and regulatory T cells. The importance of this inhibitory pathway was demonstrated in studies of knockout mice, in which deletion of the CTLA4 gene resulted in systemic immune activation, culminating in expansive lymphoproliferation and death at ~ 3–4 weeks old [4]. The ligands for CTLA4, CD80 (B7.1) and CD86 (B7.2), are expressed primarily on antigen-presenting cells and other immune cells, including T cells themselves. CTLA4 is stored in cytoplasmic vesicles in resting T cells, and moves to the cell surface when T cells are activated by antigen binding to the T-cell receptor. CTLA4 then competes with the immune stimulatory receptor CD28 for binding of CD80 and CD86. The ligands CD80 and CD86 have a higher affinity for CTLA4 than for CD28, and CTLA4 therefore has the ability to dampen CD28-mediated immune activation [5], resulting in impaired co-stimulation. Additionally, CTLA4 binding activates intracellular phosphatases that interfere with downstream signaling cascades that are important for T-cell activation.

PD-1

PD-1 and its endogenous ligands form an inhibitory checkpoint at a different stage of the lymphocyte response to infection or inflammation in peripheral tissues. In healthy individuals, the PD-1 receptor is expressed on antigen-presenting cells, activated T cells and other immune cells in peripheral tissues, including natural killer and dendritic cells. PD-1 interactions may down-regulate signaling via the B-cell receptor and natural killer cells, although the majority of its inhibitory function has been established in the setting of T-cell responses [6]. The ligand PD-L1 (PDCD1LG1; CD274; B7-H1) has a broad expression profile, and is seen on the same cells as PD-1 but also on non-hematopoietic cells in non-lymphoid organs, such as heart, lungs and muscle, implying a role in regulation of effector T-cell responses in peripheral tissues. Expression of the ligand PD-L2 (PDCD1LG2; CD273; B7-DC) appears to be restricted to B cells, macrophages and dendritic cells. Initial stimulation of a T cell results in expression of PD-1, which persists when the cell migrates to the tissues. Within the tissues, inflammatory cytokines induce expression of its ligands. PD-1 engagement increases phosphorylation of Src homology 2 domain containing non-transmembrane PTP (SHP-2), a phosphatase that may be important in attenuating T cell receptor sign. In addition, PD-1 ligation inhibits T-cell proliferation by blocking cell-cycle progression without increasing cell death [7]. The end result is reversible suppression of T-cell activity and proliferation, known as T-cell exhaustion, which has also been demonstrated in the setting of chronic infection [8]. T-cell exhaustion typically occurs at a site of infection or inflammation to limit damage to surrounding tissues and prevent autoimmunity [9].

Subversion of immune checkpoints in lymphoid malignancy

Cancers may subvert either of these two inhibitory checkpoints to evade host anti-tumor immune responses. Malignant cells escape immune surveillance by up-regulating their expression of checkpoint receptor ligands, as well as by increased expression of these ligands within the tumor microenvironment in areas of dense T-cell infiltration. This leads to inhibition of effector T-cell function. CTLA4 contributes to the suppressive function of regulatory T cells, which play a role in T-cell anergy as well as in suppression of T- and B-cell responses [10, 11]. Regulatory T cells in the tumor microenvironment constitutively express CTLA4 [12]. Tumor engagement of the CTLA4 pathway may therefore promote the suppressive function of regulatory T cells and dampen the immune response in the microenvironment.

With the exception of chronic lymphocytic leukemia (CLL), PD-1 is rarely expressed on malignant lymphoid cells themselves, but is commonly found on tumor-infiltrating lymphocytes in the microenvironment, particularly in follicular lymphoma (FL) and Hodgkin lymphoma (HL) [13, 14]. Its ligands, particularly PD-L1, are more commonly expressed on lymphoid tumors [15]. Both PD-L1 and PD-L2 appear to be up-regulated in some B-cell lymphomas, including Hodgkin lymphoma and primary mediastinal B-cell lymphoma, due to amplification of the genomic locus encoding these two proteins on chromosome 9p as well as specific gene fusions that lead to PD-L1 and PD-L2 over-expression [16-18]. Classical Hodgkin lymphomas (cHLs) with normal 9p copy number may also express detectable PD-L1. An alternative mechanism of PD-L1 up-regulation in cHLs is Epstein–Barr virus infection (Fig. 1). Epstein–Barr virus infection promotes PD-L1 expression and PD-L1 promotor activity, implying a role for the PD-1 checkpoint in the pathogenesis of a variety of Epstein–Barr virus-related lymphoproliferative disorders [19]. PD-L1 expression may also be induced by interferon γ [7]. By expressing suppressive PD-1 ligands and engaging PD-1-bearing immune cells, tumors induce T-cell exhaustion and evade the host immune response. Immune checkpoints are therefore an attractive and important therapeutic target to harness the endogenous anti-tumor activity of the immune system.

Ipilimumab

Ipilimumab was the first checkpoint inhibitor studied in lymphoid malignancies. A dose-escalation study in 18 patients with relapsed B-cell lymphoma included 14 patients with follicular lymphoma (FL), three with diffuse large B-cell lymphoma (DLBCL) and one with mantle cell lymphoma. The drug was safe and tolerable, although diarrhea was noted in 56% of patients (grade 3 in 28%), probably a reflection of the autoimmune colitis that is now a well-recognized adverse event of the drug. The overall response rate was 11%, with one patient with DLBCL achieving a complete response of 31 months’ duration and another with FL achieving a partial response of 19 months’ duration. As a correlative endpoint, ipilimumab was shown to double T-cell proliferation in five of the 16 patients in whom it was evaluated [24].

Ipilimumab has also been studied in the context of relapsed malignancy after allogeneic stem cell transplant as a strategy to augment the graft-versus-tumor effect. A phase I study evaluated 29 patients, including 14 with Hodgkin disease, six with myeloma, two with CLL and one with non-Hodgkin lymphoma (NHL). There were no dose-limiting toxicities, and 15 patients were treated with the target dose of 3 mg·kg⁻¹, one of whom developed grade 4 pneumonitis possibly related to the drug. No patients developed grade 3 or 4 graft-versus-host disease after ipilimumab. Of the three patients who showed objective responses after ipilimumab, one had mantle cell lymphoma (partial response) and two had Hodgkin disease (complete responses) [25].

Checkpoint inhibition in lymphoid malignancies: how does it work?

Our understanding of the mechanism of checkpoint inhibitors is incomplete. Despite a rudimentary grasp of what immune checkpoint blockade accomplishes on a cellular and molecular level, the precise mechanisms of action remain unclear. At the most basic level, PD-1 and CTLA4 blocking antibodies function by releasing a brake on the immune system, thereby facilitating T-cell activation and proliferation, and both CD4+ CD8+ T cells appear to play a role. However, the unleashing of activated T cells probably represents a bird's eye view that disregards the intricate network of pathways that regulate the immune system in lymphoid neoplasms [31]. The difficulty in establishing a reliable biomarker reflects the uncertainty in this area. Tumors appear to up-regulate expression of PD-L1. However, substantial responses to nivolumab have been seen even in tumors that appear to be PD-L1-negative. The studies of PD-1 ligand expression have been complicated by different definitions of what constitutes positivity, heterogeneity in the assays used, expression of PD-L1 on tumor cells compared to immune cells, and inclusion of patients at various stages of disease who may have received differing doses of therapy. Overall, the lack of a reliably identifiable biomarker in lymphoid malignancies may reflect our over-simplification of the mechanisms involved.

It is also unclear to what degree features of tumor-infiltrating lymphocytes in the microenvironment are predictive of response, given the many complexities of the tumor–host relationship. This is particularly the case in hematologic malignancies, where lines are blurred because the tumor itself comprises immune cells. Malignant cells may evade the host immune response by up-regulation of PD-1 ligands and signaling through PD-1 on CD8+ T cells. The endogenous T-cell compartment is then unable to perform its function of recognizing peptide epitopes displayed on major histocompatibility complexes (MHCs) on the surface of malignant cells, a phenomenon known as adaptive immune resistance. Pre-existing CD8+ T cells located on the border of invasive tumor margins appear to express PD-1/PD-L1. In a study of 46 patients with metastatic melanoma, samples were obtained before and during anti-PD-1 therapy with pembrolizumab. Responders demonstrated proliferation of intratumoral CD8+ T cells that correlated directly with tumor size reduction. The pre-treatment specimens obtained from these patients showed more CD8+-, PD-1- and PD-L1 expressing cells at the tumor margin and inside the tumors. The authors contend that therapeutic PD-1 blockade therefore requires pre-existing CD8+ T cells in the tumor microenvironment (particularly at the border of invasion) that are negatively regulated by immune checkpoints [32]. The theory of silenced immune cells on the periphery of tumors becoming awakened by checkpoint inhibitors and then ‘rushing in’ appears to make sense intuitively in solid tumors, but the situation is less clear in hematologic malignancies, where the tumor and the immune system are one and the same, and T cells and malignant cells are typically inter-mixed at sites of disease.

In addition to the immune cells in the tumor microenvironment, there are many unanswered questions about how much the malignant cell itself contributes to the mechanism of action of checkpoint inhibitors. These queries relate to longstanding uncertainty regarding the nature of the antigens involved in how the immune system distinguishes cancer cells from non-cancer cells. One predominant theory is that certain peptides arise as a consequence of tumor-specific mutations. These ‘neo-antigens’ are entirely absent from the normal human genome, and when the mutant peptide is presented by major histocompatibility complex molecules, they are recognized as tumor by the immune system. Therefore, tumors with more somatic mutations form neo-antigens more frequently that may be recognized by autologous T cells, and are therefore more susceptible to checkpoint inhibition. Solid tumors such as lung cancer and melanoma have the highest prevalence of somatic mutations, and indeed appear to be highly responsive to immune checkpoint blockade [33]. However, hematologic malignancies appear to fall somewhere in the middle in terms of the number of somatic mutations per megabase of coding DNA. Additionally, MHC expression is commonly lost in lymphoid malignancies, and this may interfere with neoantigen presentation. Does this imply that the formation of neo-antigens by lymphoid tumors and their subsequent presentation is less than in solid tumors? If so, theories of neo-antigenic immunogenicity and mutational burden may be less relevant in lymphoid malignancy than in solid tumors. These questions merit further investigation.

Checkpoint inhibition in various lymphoid histologies

Related to the many questions about the exact mechanism of action of checkpoint inhibition, there are also uncertainties as to what accounts for the striking differences in the response rate between different histologies of lymphoid tumors. Checkpoint inhibition is undoubtedly a promising strategy in cHL. On the other end of the spectrum, the apparent lack of responses seen in multiple myeloma patients is also striking, particularly because myeloma cells have been shown to express PD-L1 [34]. The biological basis for this difference is not understood, and there is little in the literature exploring this question. Are the differences due to less effective identification of the tumor in the first place, and, if so, is that caused by differences in tumor neo-antigens and mutational burden? Or is it due to substantial variation in the composition in the microenvironment amongst different lymphoid tumors?

There may be differences in the ways that the host immune system interacts with different lymphoid malignancies. Recent studies have revealed a role for T cells in affecting the immunogenicity of cancers early in their formation by selectively destroying malignant cells that express strong tumor-specific neo-antigens, leaving behind malignant cells that express weaker antigens (which are perhaps still tumor-specific mutant antigens) [35]. Second, the chronic T-cell response to a tumor may silence expression of certain neo-antigens through epigenetic mechanisms [36]. In this manner, differential tumor ‘editing’ by the immune system may lead to antigenic heterogeneity in different tumors, resulting in different responses to checkpoint blockade [33].

In addition to tumor immunoediting, host immunity may play a variety of roles in different tumor types that further confound the issue of disparate response rates in different lymphoid malignancies. Data suggest that there are two populations of PD-1-expressing T-cell sub-populations in follicular lymphoma, and the prevalence of the two subsets may affect patient outcome [37]. PD-1 is expressed on exhausted intra-tumoral effector T cells, but it is expressed more highly on follicular helper T cells. The native function of follicular helper T cells is rooted in their close interaction with B cells and their promotion of B-cell growth for germinal center formation: it is felt that they may ‘edit’ B cells. It is possible that follicular helper T cells may function in the host immune system to promote somatic hypermutation, affinity maturation and class-switch recombination, resulting in endogenously manufactured anti-tumor antibodies that help host B cells to fight malignant cells [38]. Therefore, PD-1 blockade in these cells may be inappropriate. The complex interplay between host B and T cells, as well as their interactions with tumor cells, is not well understood in the context of hematologic malignancy. Therefore, differences in the immune system itself in different tumor types may account for some of the varied response rates in different histologies.

Checkpoint inhibitors: timing and combinations

Aside from the questions related to the precise mechanism of checkpoint inhibition and the different responses amongst lymphoid malignancies, many questions exist regarding when in the disease course to use checkpoint blockade and whether it should be used as part of combination therapy. There have not yet been any studies of checkpoint inhibitors in the frontline setting, and therefore our experience is limited to patients with relapsed or refractory disease. It is unknown whether responses will differ between newly diagnosed and relapsed patients. In cHL in particular, the question arises as to how much improvement on the excellent frontline success rates with cytotoxic chemotherapy may be achieved. As a tumor becomes more chemo-resistant, it acquires more mutations and may display higher neo-antigenic immunogenicity than a recently diagnosed, less mutated tumor. It is also unclear whether the increase in mutations in tumors from chemo-resistant patients may include mutations that result in higher expression of PD-1/PD-L1, something that may be seen less frequently in chemo-naive patients. If it is the case that recurrent tumors are more antigenic and express PD-L1 more highly, the response rates observed in the relapsed or refractory population may be reflective of a population in whom checkpoint blockade works better than it would in the frontline. On the other hand, one may argue that a patient who has been through multiple rounds of cytotoxic chemotherapy has an immune system that is depleted. In this sense, checkpoint blockade may be more effective in a patient who is therapy-naive, i.e. a patient whose more functional immune system is ripe for activation. However, this theory is challenged by the outstanding responses seen in immuno-suppressed post-transplant patients [16, 27].

Aside from these biological hypotheses, it is necessary to consider whether immune therapy should precede chemotherapy to avoid the increased toxicity of standard chemotherapy, or whether it should follow chemotherapy to address minimal residual disease and avoid relapse. It appears to make intuitive sense to deplete the malignant cell that is subverting the immune system in the first place, and then follow this with immune therapy. Conventional cancer therapies may led to tumor cell death and release antigens to activate T cells, priming an ‘immunogenic’ tumor microenvironment in which subsequent checkpoint blockade is more effective [39]. Clinical trials in development are likely to address these important questions regarding sequential or combined use of checkpoint inhibitors and conventional chemotherapy.

If not given sequentially, should immune therapy be given in combination, either as CTLA4 or PD-1 blocking agents together or as combinations of checkpoint inhibitors with other agents? In metastatic melanoma, there appears to be benefit in terms of response rate to the combination of ipilimumab and nivolumab, albeit at the cost of increased treatment-related toxicity [40]. It is unclear whether the benefits and risks will be similar with combined checkpoint blockade in lymphoid neoplasms. Combining immune checkpoint inhibitors with immuno-modulators and targeted agents is an area of active investigation, as small molecules directed at signaling pathways involved in malignancy affect immune responses. Lenalidomide is an immune modulator that affects the tumor microenvironment. Preclinical studies in multiple myeloma have suggested various mechanisms by which the drug may stimulate antitumor immunity. A recent report suggests that lenalidomide promotes sustained proteasomal degradation of the substrate proteins Ikaros (IKZF1) and Aiolos (IKZF3) in a cereblon-dependent manner, resulting in down-regulation of c-MYC and interferon regulatory factor 4 and ultimately tumor-cell apoptosis [41, 42]. Degradation of Ikaros and Aiolos also appears to potentiate production of interleukin-2 by T cells [43]. Lenalidomide may also promote T-cell proliferation by decreasing T-cell expression of PD-1 and decreasing the number of regulatory T cells in the microenvironment [44]. Further support for this mechanism comes from a trial of rituximab and lenalidomide in relapsed indolent NHL. Correlative studies of immune cell subsets demonstrated increased CD8+ T cells and increased PD-1-positive cells in patients receiving lenalidomide, implying activation of CD8+ intra-tumoral lymphocytes. There also appeared to be fewer suppressive CD4+ regulatory T cells [45]. Preclinical data also support the combination of checkpoint inhibitor therapy and the Bruton tyrosine kinase inhibitor ibrutinib. This drug inhibits Bruton tyrosine kinase but also blocks an enzyme known as interleukin-2-inducible T cell kinase, which is essential in Th2 cells. The combination of anti-PD-L1 antibody and ibrutinib demonstrated enhanced antitumor T-cell immune responses in mouse models of lymphoma, and led to suppression of tumor growth [46]. These studies open the door for clinical studies of combinations of immunotherapy with immunomodulators such as lenalidomide, or targeted agents, including Bruton tyrosine kinase inhibitors, phosphoinositide 3-kinase inhibitors and others.