Immunotherapy in lymphoma.

Abstract

A multitude of studies have investigated why the immune system fails to develop an effective response against cancer and how to mobilize a successful immune therapeutic attack against cancer cells. Lymphomas are tumours of the immune system that develop from the cells devoted to the body defence within permissive specialized and well-vascularized tissue microenvironments, such as bone marrow and secondary lymphoid organs. Malignant lymphoid cells entail a bidirectional dialogue with a host of nonmalignant immune cells of both innate and adaptive immunity. This cross-talk not only favours malignant cell growth, survival, and drug resistance but also prevents antitumour response through resistance to cytotoxicity, compromised cytotoxic cell activity, and altered tumour-identifying ligand expression or secretion.1 Improving the understanding of the mechanisms that underlie the deficiency of an adequate antitumour immune response is now allowing manipulation of the immune system for the purpose of lymphoma eradication. The first question to address is through which mechanisms malignant cells subvert normal cells of the immune system to protect themselves from attack. A number of different mechanisms have been ascertained. As an example, diffuse large B cell lymphoma (DLBCL) cells commonly fail to express cell-surface molecules necessary for the recognition of tumour cells by immune-effector cells. This may occur because of mutations and deletions that inactivate either the β2-Microglobulin gene, thus preventing the cell-surface expression of the HLA class-I (HLA-I) complex that is necessary for recognition by cytotoxic T lymphocytes (CTL) or the CD58 gene, which encodes a molecule involved in T and natural killer cell-mediated responses.2 Lymphoma cells divert the classical activation of the innate immune system and subvert the antitumour immune response exerted by T helper (Th) cells, CTL, and macrophages. The apposition of lymphoma cells with T cells within involved tissues induces a dysfunction of the T-cell immunological synapse, the complex molecular structure that represents the site where the T-cell receptor (TCR) is triggered by the peptide–HLA complex. This dysfunction has been described initially in B-cell chronic lymphocytic leukemia (CLL) and subsequently also in follicular lymphoma (FL) and DLBCL. It is due to inhibitory ligand-induced impairment of T-cell actin dynamics and prevents the lysis of tumour cells through CTL.3 Notably malignant lymphoid cells may also secrete cytokines such as IL-12 that induce T cell “exhaustion,” a reversible condition that occurs when T cells are exposed to prolonged stimulation with antigen and leads to a profound inability of T cells to respond to activation signals. Moreover, lymphomatous tissues are enriched with immune cell subsets that suppress an efficient immunological response against the tumour such as FoxP3 regulatory T cells (Tregs). Tregs suppress the proliferation and activity of both CD4+CD25– and CD8+ T cells, and their enrichment is contributed to by lymphoma cells that allow the preferential conversion of T helper cells into Tregs by proteins such as TGF-β or CCL22. A number of studies have suggested a critical role for monocyte/macrophage cells in chronic lymphoid malignancies. Their accumulation is favoured by the inflammatory microenvironment within infiltrated tissues that attracts monocytes/macrophages and cooperates with IL-10 to promote the M2 polarization of tumour associate macrophages. M2-polarized tumour associate macrophages promote tumour growth by stimulating angiogenesis and inducing an immunosuppressive Th2 response.4 A major achievement has been the understanding that the homeostasis of antigen-specific lymphocytes requires multiple different signals. Lymphocyte activation is triggered by specific antigen recognition through the TCR and costimulatory signals. Coinhibitory signals delivered through specialized receptors that function as immune checkpoints maintain self-tolerance and promote the resolution of inflammation during immune responses thereby preventing the development of autoimmunity. The molecular family of checkpoint inhibitors classically represented by the B7–CD28–CTLA-4 family has been enriched by the identification of the programmed cell death 1 (PD-1)/programmed death–ligands 1 and 2 (PD-L1 and PD-L2) immune checkpoint inhibitory receptor and ligands. The functional and biochemical properties of PD-1 are not limited to but have been best studied in T cells where PD-1, induced upon activation through the TCR and cytokine receptors, is necessary for the termination of the immune response. Programmed death–ligand 1 is constitutively expressed at low levels on both professional and nonprofessional antigen presenting cells as well as on a wide variety of nonhematopoietic cell types and its expression is also induced by proinflammatory cytokines. A key mechanism by which cancer cells limit the host immune response is the upregulation of PD-1 ligands and their ligation to PD-1 on tumour-specific CD8+ T cells. In the tumour microenvironment inflammatory mediators, among which interferon-γ produced by tumour-infiltrating T cells that are capable of recognizing tumour antigens is the most potent, can induce expression of PD-L1 and PD-L2 not only on cancer cells but also on other cell types such as macrophages, dendritic, and stromal cells. The expression of PD-1 ligands on tumour cells can have a genetic basis as best exemplified by classical Hodgkin lymphoma where chromosome 9p24.1/CD274(PD-L1)/PDCD1LG2(PD-L2) alterations have been shown to increase the abundance of PD-1 ligands.5 JAK2-STAT signalling further increases PD-1 ligand expression. Notably, Epstein-Barr virus (EBV) infection can increase expression of PD-1 ligands in EBV+ HL. Levels of PD-L1 expression in cancer cells are also regulated by epigenetic mechanisms. Lymphoma cells protect themselves from attack by the immune system through several different mechanisms. These observations have paved the way to the development of new therapeutic avenues of cancer immunotherapy. The use of genetically modified chimeric antigen receptor (CAR)-T-cells that overcome the mutational landscape of cancers and the blockade of immune checkpoint receptors that inhibit the immune response convincingly show that the possibility to eradicate lymphoid malignancies with immunotherapeutic approaches is more realistic than ever before. Two immune checkpoints have been targeted in lymphoma and antibodies that block signaling via these receptors have resulted in therapeutic benefit for patients. CTLA-4 is commonly expressed on T-cells after activation and specifically functions to down regulate T-cell activity through a variety of mechanisms. These mechanisms include preventing costimulation by outcompeting CD28 for its ligand B7 and also by inducing T-cell cycle arrest. Blocking CTLA-4 prevents down regulation of T-cell function and promotes the persistence and activation of intratumoural T-cells. Similarly, PD-1 is a negative regulator of T-cell activity and limits T-cell activation when stimulated by 1 of its 2 ligands, PD-L1 and PD-L2. Activation of this receptor typically down regulates T-cell activity and PD-1 is commonly expressed on activated and immunologically exhausted T-cells. Persistent stimulation via PD-1 results in an ineffective immune response and blocking signaling through PD-1 using an anti–PD-1 antibody allows for reactivation of suppressed or exhausted T-cells. Targeting these 2 immune checkpoints is therefore a rational therapeutic approach. Initial clinical studies utilized an anti–CTLA-4 antibody ipilimumab in B-cell lymphoma with a goal of enhancing antitumour T-cell responses. In an initial phase II clinical trial, 18 patients were treated and 2 responses were seen—1 patient with DLBCL had a complete response and 1 patient with FL had a partial response.6 Correlative studies showed that T-cell proliferation in response to recall antigens increased in most patients confirming activation of the immune system. Subsequent clinical trials were done using ipilimumab in patients with relapsed hematological malignancies post allogeneic stem cell transplant. In these studies, responses were seen in lymphoma patients including 2 complete responses in patients with HL and a partial response in patients with mantle cell lymphoma (MCL). Ipilimumab has subsequently been combined with nivolumab, an anti–PD-1 antibody, and the combination has shown significant clinical responses in HL. While targeting CTLA-4 has resulted in modest clinical benefit, clinical trials testing PD-1 blockade have shown significant promise in patients with lymphoma, particularly HL. Initial trials used an anti–PD-1 antibody pidilizumab, and while there is some controversy as to whether pidilizumab specifically targets PD-1, the initial trial of pidilizumab postautologous stem cell transplant did show a significant impact of this agent on progression free survival.7 A subsequent clinical trial of pidilizumab in combination with rituximab in FL showed a high complete response rate of 52% in the 32 patients treated. Additional clinical trials of other anti–PD-1 antibodies have been performed. Studies using nivolumab that targets PD-1 showed dramatic responses in patients with HL. In the initial phase I trial of nivolumab, the 23 patients in the HL cohort had an overall response rate of 87%.5 The responses seen in this trial have been durable and with longer-term follow-up, a significant proportion of patients have remained in remission. A subsequent confirmatory phase II study of nivolumab in HL patients confirmed an overall response rate of 68%.8 These responses have remained durable and with a median follow-up of 14 months, the median duration of response has not yet been reached. The response rates in other lymphomas have been lower but responses have been seen in patients with FL, DLBCL, and T-cell non-Hodgkin lymphoma.9 In general, responses in these histologies have been in the 20% to 40% range and responses have not been as durable. Similar results have been seen with pembrolizumab, a further antibody targeting interactions between PD-1 and its ligands. Data with this agent in classical HL have confirmed high response rates and good tolerability. In an initial phase I clinical trial, the overall response rate for the 31 patients treated was 58% and these responses have also been durable with longer-term follow-up.10 A subsequent phase II trial of pembrolizumab, including patients who progressed after autologous stem cell transplant as well as patients ineligible for a transplant, confirmed a high response rate of 68% with the responses varying between 64% and 74% depending on the subset analyzed. Correlative studies from the Hodgkin cohort in both the pembrolizumab and nivolumab studies have confirmed high expression of PD-1 ligands on the malignant cells and the majority of the overexpression is due to amplification or copy number gain at chromosome 9p24.1. The significant clinical results of particularly pembrolizumab and nivolumab in HL and other lymphomas have led to a plethora of combination studies. An initial study combining ipilimumab and nivolumab, so as to have dual immune checkpoint blockade, resulted in a 74% response rate in a cohort of 31 HL patients. Of note, however, responses have been significantly lower even with dual checkpoint blockade in other lymphoma histologies. Similarly, combination studies of immune checkpoint therapy with antibody drug conjugates such as brentuximab vedotin have been conducted. Recent reports of the results of combining brentuximab vedotin with nivolumab showed an overall response rate of 90% in HL patients in first relapse and 100% in patients who had progressed posttransplant. Further studies using multiple combinations in various lymphoma histologies are currently in progress. Clearly, targeting immune checkpoints is a promising approach to optimizing the antitumour immune response in both Hodgkin and non-Hodgkin lymphoma. A further option to overcome tumour resistance is to adoptively transfer large numbers of expanded or genetically reprogramed T-cells specific for a tumour expressed antigen. Approaches include the use of tumour infiltrating lymphocytes, cells transfected with TCR specific for tumour antigens in the context of the patient's HLA, or chimeric antigen receptor gene-modified T cells (CAR-T). Chimeric antigen receptor gene-modified T cells specific for the B cell lineage antigens CD19, CD20, and CD22 have shown the greatest promise in patients with acute lymphoblastic leukemia, NHL, and CLL. For NHL, the most encouraging antitumour activity has been with second-generation CAR constructs that have used a murine scFv specific for CD19 and linked to either the 41BB or CD28 costimulatory domains and a CD3zeta signaling sequence. Typically, cells are collected from the patient via leukapheresis and then either separated into subsets (CD4 and CD8 cells) or bulk lymphocytes directly stimulated through CD3 and CD28 and transfected with either a lentiviral or retroviral construct containing the CAR gene. After a 7- to 14-day period of cell growth in vitro, the cells are given back to the patient at a variety of doses, usually 2 to 4 days after the patient has received a cycle of lymphodepleting chemotherapy. Lymphodepletion appears important to allow upregulation of homeostatic regulatory cytokines (IL-7 and IL-15) that improve the expansion and persistence of the adoptively transferred cells and may provide some degree of transient tumour control. In most studies, low-dose or high-dose cyclophosphamide alone or in combination with fludarabine has been used. The CAR-T cells are a living therapy that may replicate and exponentially expand upon interaction with normal or abnormal cells expressing the target antigen. This can lead to very rapid tumour regression, including tumour lysis syndrome as well as toxicity resulting from cytokine release syndrome (CRS). Cytokine release syndrome appears to be caused by the interaction of CAR-T cells and antigen expressing tumour cells as well as interaction with the patient's immune system and release of inflammatory mediators including IL-6, TNF-alpha, and gamma interferon and manifest as fever, hypotension, capillary leak, and coagulopathy. This may require ICU-level care. For grading, see Lee et al.11 In most studies, an unexplained neurotoxicity with delirium and somnolence or even seizures and/or stroke like phenomena can occur in a subset of patients. Fortunately, even if significant, these neurotoxicities are usually fully reversible. Moderate-to-severe CRS can be treated by anti–IL-6 protein or receptor antibodies (tocilizumab or siltuximab) often in combination with a short course of dexamethasone. Tumour responses are usually quite rapid and occur in the first 1 to 2 months. Hospitalization is often required for supportive care, but approximately 30% of patients in the Seattle series were treated entirely as outpatients. Kochenderfer at the NCI has reported the longest outcomes for patients with NHL.12 This approach uses a murine scFv and a CD28 costimulatory domain that is transfected into bulk-stimulated T cells using a retroviral vector. Using this approach, cells are produced rapidly (<7 days), but may be comprised of widely variable T cell subsets. In the initial report, they treated 15 patients (9 with DLBCL, 2 indolent NHL, and 4 CLL) with 8 CR and 4 PR. In patients with DLBCL, 4/7 evaluable patients had a CR and 3 were ongoing between 9 and 22 months. The CRS and neurotoxicity was the most significant toxicity and 1 patient without known CRS had sudden death at day 16 post–CAR-T cells. Patients had extensive prior therapy, and the achievement of CR's in this population represents a significant finding. In addition, this was feasible in patients who had relapsed following allogeneic transplantation, without any apparent increase in graft-vs-host disease or graft rejection. Kite pharmaceuticals is planning to commercialize this approach (KTE-C19) in relapsed/refractory DLBCL and reported early results from the pivotal ZUMA-1 at ASH 2016.13 In 62 patients with DLBCL (n = 58) or transformed FL or PBMCL (n = 11), the overall response rate was 79% with a 52% initial CR rate to a targeted dose of 2 × 106 cells/kg after low-dose Cy/Fu conditioning. At 3 months, the ORR was 44% with a CR rate of 39%. Data have not yet been released for 6- and 9-month follow-up, but this compares favorably to historical data for patients with relapsed/refractory DLBCL (8% CR from the SCHOLAR-1 study). Grade 3 or greater CRS was observed in 18% and neurotoxicity in 34% of patients. Two patients died of related side effects. The Seattle group has taken a different approach by creating a more uniform CAR-T cell product by selecting CD4 and CD8 T cell populations and administering a CAR-T cell product containing CD4:CD8 cells in a defined 1:1 ratio in 32 patients with NHL.14 In this phase I/II trial, we observed that the dose of CAR-T cells was important as well as the type/intensity of lymphodepletion. Patients treated with Cy alone had a high incidence of immunologic rejection of the modified T cells that limited the duration of response. The addition of fludarabine decreased this immune rejection and resulted in more robust CAR-T cell expansion, which required a 1 log reduction in the administered T cell dose to limit toxicity. By using a defined ratio of cells, we were able to identify dose response and dose toxicity relationships. The overall response rate was 63% with a 33% CR rate. However, using Cy/Flu lymphodepletion the ORR was 72% with a 50% CR and with a longer duration of remission. This was correlated with greater CAR-T cell expansion and persistence. We were able to observe complete remission of disease in the blood, bone marrow, nodal and extra nodal tissues, and CNS. Severe CRS was observed in 13% and severe neurotoxicity in 28%, and 2 patients died of toxicity (both treated above the eventual MTD). Importantly, measurement of serum biomarkers 1 day after CAR-T cell infusion was correlated with the development of severe CRS or neurotoxicity suggesting an avenue for early intervention. Juno Therapeutics has developed this approach further (JCAR-17) in an ongoing multicenter phase I/II trial that also reported encouraging results at ASH 2016 with 82% OR and 73% CR in the first 11 evaluable patients with DLBCL.15 Of interest, 1 patient with a parenchymal brain lesion also responded with a CR and without CRS or CNS toxicity. At this time, it is difficult to compare these very different CAR-T cell products, but both approaches appear to be generating response rates and CR rates that are vastly superior to the current standards of care in this difficult to treat patient population (relapsed refractory DLBCL). The duration of remissions and the causes of disease relapse will be important to the field. However, this is likely just the beginning of a new treatment era because combination therapy with CAR-T cells and checkpoint inhibitors are a logical extension and clinical trials are ongoing. Stephen Ansell: Research funding from Bristol Myers Squibb and Merck. Federico Caligaris-Cappio: No relevant conflict of interest to declare. David Maloney: Research funding from Juno Therapeutics.