From the time of its first description by Thomas Hodgkin to the latest classification of haematological neoplasms, classical Hodgkin Lymphoma (cHL) has always held a peculiar spot amongst haematological cancers with both its clinical and biological characteristics keeping it apart from other forms of lymphoma (1–3). Clinically, cHL is one of the most common cancer types being diagnosed in adolescents, with a characteristic incidence bimodal distribution having peaks in the young adult population and in subjects aged 55 and older. Moreover, it is characterised by good response rates to first-line therapy, with cure rates ranging from 70 to 90% with chemotherapy alone or its combination with radiotherapy, depending on stage and other risk factors, and with second line-therapy allowing for disease eradication in a good portion of relapsed/refractory patients (4).

From a pathological standpoint, cHL is composed of a minority of large mononuclear (Hodgkin) and multinucleated (Reed-Sternberg) neoplastic cells, scattered throughout a rich inflammatory microenvironment (ME) (1). The latter usually constitutes >95% of the tumour mass and consists of mature B and T lymphocytes, plasma cells, eosinophils, macrophages, mast cells, and, to a minor extent, of myeloid suppressor derived cells (MSDCs) and NK cells (5). The relative proportion of such populations and the distribution of each immune cell subset depend on a tight crosstalk between Hodgkin/Reed-Sternberg cells (HRSC) and the inflammatory background, allowing HRSC to shape the immune ME and producing biological niches that sustain their survival and shelter them from immune surveillance.

The overexpression of Programmed Death Ligand 1 (PD-L1) on the surface of HRSC is one of the hallmarks of cHL and is one of the main contributors to its ability to avoid immune-mediated disruption (6, 7).

From a physiological point of view, PD-L1 is a transmembrane protein belonging to the B7 family, usually expressed on the surface of immune cells, professional (such as dendritic cells, macrophages and B-lymphocytes) and nonprofessional (such as epithelial and endothelial cells) antigen-presenting cells (APCs) and many nonhematopoietic cells (8). Alongside PD-L2, PD-L1 exerts its role by binding to the PD-1 receptor expressed on the surface of T-lymphocytes, resulting in an inhibitory signal (9). After the binding of the PD-L proteins to the PD-1 receptor, the immunoreceptor tyrosine-based inhibitory and switch motifs (ITIM and ITSM, respectively) located in the cytosolic end of the receptor are autophosphorylated, allowing for the recruitment of Src homology region 2 containing phosphatase 2 (SHP2), which represents the main effector of the inhibitory signal triggered after the engagement of PD-1 (10–12). After it’s engagement to the cytosolic end of the PD-1 receptor, SHP2 undergoes a conformation change that allows for its activation (13). This phosphatase is then able to inhibit the phosphoinositide-3-kinase (PI3K) and the phospholipase C gamma 1 (PLCG1), interfering with the PI3K/Akt and Ras/MAPK/ERK pathways, which are crucially involved in T-cell activation and their survival after antigen recognition. SHP2 is also able to directly inhibit the function of the leukocyte-specific tyrosine kinase (LCK), which, in turn, is responsible for the phosphorylation of fundamental proteins involved in T-cell activity such as ZAP70, CD3ζ and PKCθ (14, 15). Moreover, an activated PD-1 is able to interfere with the activity of the casein kinase 2 (CK2) by reducing both its activity and its mRNA levels. This, in turn, allows for continued phosphatase activity by the phosphatase and tensin homolog (PTEN) protein, which is inhibited by CK2-mediated phosphorylation during T-cell activation, further interfering with the PI3K/Akt pathway and thus truncating the TCR downstream signals (16).

Furthermore, the inhibition of the PI3K/Akt pathway through the activation of the PD-1/PD-L1 axis induces metabolic changes, making in T-cells unable to exploit glycolysis, glutaminolysis and branched-chain amino acid metabolism for energy production, all of which are key aspects of the activated T-cell phenotype. Moreover, PD-1 derived signals induce an increase in fatty-acid oxidation through upregulation of key enzymes involved in lipid metabolism, a metabolic switch associated with cell longevity capability of reacting to antigenic stimuli ( Figure 1 ) (17).

As stated before, the PD-1/PD-L axis plays a pivotal role in cHL’s ability to avoid immune recognition. One of the most characteristic genetic lesions, which is found in almost all cases of cHL, is the amplification of the 9p24.1 chromosomal region, containing the CD274 and the PDCD1LG2 genes, encoding, respectively, for the PD-L1 and PD-L2 proteins. Overexpression of these genes directly translates into a higher expression of both proteins on the surface of HRSC, and, consequently, direct inhibition of the T cells in the surrounding tumour ME. Moreover, the same genomic region includes the JAK2 gene, and its amplification leads to an overactivation of the JAK/STAT pathway, which, in turn, acts as a positive transcriptional regulator to the genes encoding for PD-L1 and PD-L2, further enhancing the HRSC ability to pathologically express these antigens on their surface (6, 7, 18).

Epstein-Barr Virus (EBV) positive cHL is rarely characterised by the 9p24.1 amplification, and, as such, is not able to exploit the genetic pathways detailed above to manipulate the PD-1/PD-L immune checkpoint. The latent membrane viral protein 1 (LMP1), produced as a consequence of EBV infection, is however able to increase transcription levels of the CD274 and PDCD1LG2 genes through stimulation of the AP-1 and JAK/STAT signalling pathways, therefore making EBV-infected HRS cells as proficient in immune evasion as their EBV-negative counterparts (19).

Pathological overexpression and dis-regulation of the PD-L1 and PD-L2 surface proteins, however, are not the only mechanisms that define cHL’s ability to manipulate the host’s immune system. Having a B-cell origin (20), HRSC derive from APCs and, as such, are expected to express both class I and class II Major Histocompatibility Complexes (MHC-I and MHC-II, respectively). More than half of cHL cases are characterised by the loss of surface expression of MHC-I due to mutations in the gene encoding for Beta-2 microglobulin (β2M), thus impinging antigen presentation to CD8+ T-cells (21, 22). The loss of MHC-I on HRSC is also associated with overexpression of HLA-G, a noncanonical MHC-I receptor that is able to bind to the inhibitory ILT2 and ILT4 surface proteins on other immune effector cells, as well as to CD8 on the surface of T and NK cells, triggering FAS-mediated apoptosis, a process also stimulated by the strong expression of FASL on HRSC ( Figure 2 ) (22, 23).

The nurturing of an immune-suppressive biological niche, whose members are directly involved in dynamic interactions among themselves and with HRSC throughout cell-to-cell contacts and soluble factors, is another key element in the biology of cHL (24). Through the release of many cytokines and chemokines, HRSC recruit circulating immune cells, reshaping their functions and promoting their local expansion and spatial distribution ( Figure 3 ). For these processes, the most relevant factors are IL5, IL7, IL8, IL9, CCL5, CCL17, CCL20 and CCL22, which have been variably associated with recruitment/activation of eosinophils, neutrophils, B and T lymphocytes, mast cells and specific macrophage subsets (25, 26). Beside these soluble factors, many surface receptors on HRSCs promote their growth/survival, through direct interaction with immune cell subsets and/or through binding of ME-derived cytokines. These include the activating receptors CD30 and CD40, and interleukin/chemokine receptors, such as IL7R, IL9R, IL13R, TACI and CCR5 (27–30). The engagement of such receptors activates downstream signalling pathways, which are pivotal for the pathogenesis of cHL. These include the PI3K/AKT pathway (interaction between CD30 and CD30-ligand on mast cells and eosinophils), the NF-κB pathway (interaction between CD40 and CD40-ligand on CD4+ T cells), and the JAK/STAT pathway (interaction of HRSC surface receptors with several ME-derived cytokines) (31–39).

Besides providing survival and proliferation stimuli, tumour ME is able to directly interfere with the host’s immune recognition of HRSC. In detail, the presence of regulatory T cells (Treg) among the inflammatory infiltrate exerts a major pathogenic role by secreting large amounts of immune inhibitory IL10 (40, 41). In cHL, Treg are recruited by chemoattractants of HRSC derivation, such as CCL17/TARC, CCL22, CCL5, IL-4, IL-5, IL-10 and IL-13. Treg infiltration is also facilitated by the over-expression of galectin-1, a beta-galactoside-binding protein that mediates cell-to-cell and cell-to-matrix interactions (42). Other mechanisms involved in Treg enrichment include HRSC-induced trans-differentiation of bystander T helper cells (43, 44) and Treg recruitment by cHL-associated accessory and myeloid cells. This is specifically mediated by M2 macrophages, mast cells and MDSCs. Moreover, the expression of the indoleamine 2,3-dioxygenase (IDO) enzyme by both HRSC and other elements of the tumour ME allows for the degradation of Tryptophan in the cHL biological niche, suppressing the activity of NK cells and forcing T cells to acquire a Treg phenotype (45). In addition, Kynurenine, which is the main Tryptophan metabolite, is able to inhibit antigen recognition-derived proliferation and to induce T-cell death (46, 47).

The lymphocyte infiltrate is also rich in cells expressing high levels of lymphocyte-activation gene 3 (LAG-3), which, alongside PD-1 and TIM3, defines a subset of T-cells that have undergone activation and subsequent exhaustion due to chronic antigen stimulation. LAG3 is another contributor to the mainly tolerogenic attitude exerted by T cells in cHL due to its ability to bind MHC-II with greater affinity than CD4 leading to the inhibition of TCR signalling, proliferation and cytokine secretion by antigen specific T-cells (48–50). Interestingly, a second T-cell population that expresses CTLA-4 but is PD-1 and LAG-3 negative has been recently described to be in close contact with HRSC, together with the identification of strong expression of CD86 on the latter cell type, suggesting that cHL is able to exploit yet another known immune checkpoint to enhance its survival with limited crosstalk between the CD86/CTLA-4 and PD-1/PD-L1 axes, as they appear to engage on different T-cell populations ( Figure 2 ) (51).

Tumour infiltrating macrophages are also another key actor in the immunosuppressive inflammatory infiltrate that characterises cHL (52). These cells, under the stimulation of IFN-gamma produced in the surrounding environment, express high levels of PD-L1, and as such, they participate in the suppression of T-cell response both directly, by binding PD-1 on the surface of T-lymphocytes, and indirectly, by impeding the interaction between T cells and HRSC (53). Of note, the high levels of PD-L1 seen in cHL-associated macrophages may be partially due to trogocytosis of PD-L1-loaded HRSC membrane patches (54).

M2 macrophages are directly induced by HRSCs through TNF, IL-10, TGF-β, GM-CSF, IL-13 and CCR5 secretion (55, 56). Mast cells are attracted by HRSC-derived IL13 and promote Treg skewing via TGF-β. This growth factor also stimulates stromal remodelling by fibroblast proliferation and collagen production (57). Finally, MDSCs support immune suppression via TGF-β, IL-10 and peroxynitrite production and via arginase-mediated depletion of ME arginine, an immune-metabolite responsible of T cell proliferation and activation (58–62).

All the evidence discussed above demonstrates how a large percentage of cHL patients can benefit from CPI-based therapy in various phases of their disease history. The relatively high ORR seen with Nivo and Pembro monotherapy, obtained mainly in populations of patients that are canonically considered harder to treat, shows how this cancer can be intrinsically sensible to such therapies. However, the presence of non-responding patients, as well as the limited duration of responses hints at the existence of resistance mechanisms to checkpoint inhibition that can be present at baseline, in the first case, or that can be acquired during therapy, in the second one.

The mechanisms of CPI therapy resistance in cHL are, at this point in time, far from being completely understood. However, the utilisation of such drugs in the treatment of other malignancies predates their use in cHL, and, as such, evidence on how these solid tumours resist immune checkpoint blockade is becoming clearer and can help us glimpse into cHL’s own resistance mechanisms to checkpoint inhibition. The description of these mechanisms, unproven to directly take part in primary or acquired resistance to CPIs in the specific setting of cHL, is beyond the scope of this review and has been done elsewhere (98, 99). There are, however, some resistance mechanisms that once again set cHL apart from other malignancies and are worth noting.

When anti-PD1 monoclonal antibodies were first developed, the physiopathological basis that was supposed to be behind their efficacy was that the removal of the inhibitory signal coming from the PD-1/PD-L1 axis would rescue CD8+ T cells from their exhausted state, eliciting a direct killer response against the tumour. This has been demonstrated to be the case in different solid tumours (100), and, coherently, the loss of expression of MHC-I molecules on the surface of neoplastic cells is one of the described CPI resistance mechanisms (98, 99). As described above, the loss of MHC-I due to β2M inactivating mutations can be observed in roughly 70% of the cases of cHL. However, the high response rates to CPIs, points to the fact that the activity of these drug in cHL is not based on the direct release of killer T-cell inhibition but rather on the recognition of cancer epitopes exposed on MHC-II molecules by CD4+ T-cells. Indeed, the inflammatory ME appears to be enriched for Th1 polarised CD4+ T-cells, who express an effector memory phenotype (101), characterised, among other peculiarities, by a medium level of expression of PD-1 and a peculiar sensibility to PD-1 targeted therapeutic approaches ( Figure 3 ) (102). Coherently, the lack of expression of MHC-I was not associated with reduced CPI effectiveness in a recent report, whereas MHC-II expression was positively associated with better PFS (103). Another proposed mechanism of resistance to checkpoint inhibition is the reduced tumour mutational burden (TMB) seen in some cHL cases (104, 105). As high TMB was one of the first biomarkers associated with a high probability of response to CPI in solid tumours due to the higher production of neoantigens within the neoplastic cells, a low TMB in cHL, combined with the defective antigen presentation associated with the lack of MHC-I expression, might explain some cases of primary resistance to CPIs. In this context, combination therapy with chemotherapeutic drugs that can modify the tumour’s antigenic repertoire might become particularly appealing. The proposed role played by CD4+ T-cells in the response to CPIs seen in cHL has been recently challenged by the analysis of selected patients enrolled in the NIVAHL trial, for which tissue samples obtained at diagnosis and after the first days of Nivo monotherapy were directly compared. Herein, the authors report the complete disappearance of HRSC in up to 50% of the biopsies analysed after the start of therapy, showing no enrichment for T lymphocytes in the inflammatory infiltrate, but rather depicting a reduction of CD4+, LAG-3+ regulatory Th1 polarised T-cells, as well as of PD-1+ tumour associated macrophages. The clinical and pathological responses documented in this study were not significantly associated with MHC-I or MHC-II expression, as well as to clonal T-cell expansion or the increase in T-cell cytotoxic response (106).

While mechanisms of primary and secondary resistance to CPIs in cHL are still being formally assessed, it is tempting to speculate that the peculiar biological characteristics of HRSC and their ME described elsewhere in this paper might play a big role in the loss of an achieved response or the lack thereof seen in some cases. The overexpression of IDO and LAG-3 on T cells and other accompanying inflammatory cells, or the expression of molecules such as HLA-G on the surface of HRSC, represent paths of immune evasion alternative to the activation of the PD-1/PD-L1 axis, and their prevalence over the latter, either at diagnosis or after therapy with CPIs, might be a key element in primary and secondary resistance to PD-1 directed therapy.

The results from the clinical trials testing CPIs in cHL show how PD-1 targeted therapies are highly effective in this context and put both Pembro and Nivo in the spotlight as main players in the cHL’s treatment algorithm. These results show how the manipulation of immune evasion in the setting of cHL is a very effective therapeutic approach, and, as such, drugs that target some of the other physiopathological mechanisms exploited by HRSC to avoid host recognition could prove to be active and effective, both as standalone and as part of combination therapies.

Favezelimab is a humanised IgG4 monoclonal antibody that targets LAG3 ( Figure 4 ), whose activity is currently being explored in combination with Pembro in a phase 1/2 trial (NCT03598608). Notably, cohort 2 of this trial included cHL patients who relapsed after ASCT or were not eligible to undergo the transplant procedure and were refractory to anti-PD-1 therapy. Preliminary results of this trial were updated at the 2022 annual ASH conference. After a median follow-up of 19 months, median PFS was of 10.7 months, while median OS was of 25.7 months and the median duration of response was 19 months, with an ORR of 29% and a CR rate of 9%. Sixty-five percent of patients achieved a response duration of ≥12 months. The 12-month OS and PFS were 91% and 39%, respectively. Notably, 70% of responders had received anti-PD-1 compounds as their most recent line of therapy, suggesting that the efficacy of LAG3-directed therapy might be independent of its combination with PD-1 blockers (107).

Moreover, Camidanlumab-Tesirine (Cami) is a new anti-CD25 antibody-drug conjugate being tested in the setting of R/R cHL ( Figure 4 ) in a phase II study whose results have recently been reported in the 2022 EHA conference. The study is being conducted in a large cohort of heavily pretreated patients (n=115; median number of treatments=6), obtaining an ORR of 70.1% and a CRR of 33.3%, respectively, with a median duration of response of 13.7 months. The safety profile for Cami was notable for the development of Guillain-Barré Syndrome in around 7% of patients, an event that needs to be accounted for as the drug is being further tested. Due to the expression of IL-2Rα (CD25) on the surface of both HRSC and Treg cells, the activity of Cami might be attributed, at least partially, to the local depletion of T suppressor cell and the renewal of a competent anti-tumour immune response (108).

Avoidance of the host’s immune system is considered one of the modern hallmarks of cancer (109), and is a key element in the physiopathology of cHL. HRSC achieve immune evasion by building an immune-suppressive ME and through the exploitation of the physiological immune checkpoints. In this context, the PD-1/PD-L1 axis plays a central role, through the overexpression of PD-L1 both on the surface of HRSC and in different non-neoplastic cell types throughout the tumour ME.

After the success of CPIs in the treatment of non-haematological cancers, the efficacy of both Nivo and Pembro has been effectively tested in cHL, achieving impressive results even in canonically harder to treat patients. Interestingly, while in the setting of solid tumours the anticancer activity of PD-1 directed CPIs is mediated by the restoration of killer T-cell function, the usual lack of MCH-I expression in cHL makes this mechanism unlikely, turning our attention to CD4⁺ T-cells and the expression of MHC-II on the surface of HRSC, proven to be a key element in CPI response both by clinical and pathological data. On the other hand, data coming from pathological analysis of biopsies taken only a few days after starting CPI therapy in the context of the NIVAHL trial is suggesting that, at least in the first line setting, it’s not a T-cell response, but rather the loss of pro-survival factors due to depletion in Tregs and macrophages, that is not the driving force for the observed response. Notably, these data contrast quite sharply with the T-cell responses previously described in the relapsed/refractory cHL setting following successful anti-PD-1 therapy (110).

As with many other effective therapeutic tools in the history of cancer, our efforts are now focused on bringing CPIs further up into the cHL’s treatment algorithm, and to improve their efficacy by making them part of combination therapies. The recently published results from the three phase 2 trials mentioned above show how the combination of anti-PD-1 monoclonal antibodies with BV or with chemotherapy schemes such as ICE or GVD as first salvage therapy could improve our cure rates even in primary refractory patients. Moreover, these combinations bring to light the impressive synergistic effect between CPIs and certain chemotherapeutic compounds, associated with the off target immunomodulating effects of agents such as Gemcitabine, Doxorubicin and Vinorelbine and their ability to trigger immunogenic cell death, which makes cancer cells more easily targetable by immune effector cells. The synergy between CPIs and chemotherapy, however, goes both ways, as the exposure to anti-PD-1 seems to render cHL more susceptible to the cytotoxic effects of both chemotherapy and allogenic cellular products in the context of alloHSCT. Further studies are needed to identify the optimal therapeutic approach regarding the specific chemotherapeutic compounds to utilise, the best treatment schedules and the molecular mechanisms underlying these synergisms.

Many ongoing trials are testing varying combinations of CPIs and chemotherapy with a sequential approach or by concurrent administration both in the frontline and relapsed/refractory setting.

As with many other promising agents introduced before them, CPIs are not able to achieve a response in the totality of treated patients, and, especially when used in later lines of therapy and as single agents, responses can be partial and often limited in time, which points to the subsistence of primary and acquired mechanisms of resistance to checkpoint blockade. The existence of alternative strategies of immune escape in the physiopathology of cHL is another fascinating area to explore. While the majority of cHLs seems to largely depend on the aberrant activation of the PD-1/PD-L1 axis to avoid host recognition, the cases that exhibit primary resistance CPIs might be reliant on the subversion of other immune-checkpoints, while in the context of secondary resistance these mechanisms, that may not be prevalent at diagnosis, might take the spotlight away from the PD-1/PD-L1 axis and restore effective immune evasion. Some of these secondary mechanisms are currently being targeted by experimental therapies, which could prove to be effective both following progression while on CPI therapy or in combination with PD-1 blockers. Hopefully, while our experience with these novel agents grows, some clinical, molecular or pathological markers can be identified to properly select the patients that will benefit the most from CPI-based therapies, separating them from the patients that can be effectively cured with a classical chemotherapeutic approach, and from those who wouldn’t benefit from either therapy, delineating a subset of patients that might represent one of the future’s hard-to-solve unmet clinical need.

AC designed the study, reviewed the literature and wrote the manuscript, FS wrote pathological paragraphs and provided suggestions, MP and LT revised the manuscript and provided suggestions, AV designed the study, reviewed the literature and revised the manuscript. All authors contributed to the article and approved the submitted version.