Novel agents and immune invasion in Hodgkin lymphoma

A 19-year-old woman presents with a persistent dry cough and progressive dyspnea on exertion. Imaging reveals a 14-cm mediastinal mass, which is biopsied and found to be nodular sclerosis Hodgkin lymphoma (HL). Positron emission tomography (PET) demonstrates fluorodeoxyglucose (FDG)-avid adenopathy above and below the diaphragm, consistent with stage IIIA bulky disease. After 2 cycles of ABVD (doxorubicin, bleomycin, vinblastine, dacarbazine), restaging PET shows an overall reduction in disease burden but with focal areas of residual FDG uptake (Deauville 5). After discussing the PET findings and possible escalation to BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisone), she opts to continue treatment with ABVD. Unfortunately, a repeat PET scan after 2 additional cycles demonstrates clear progression of disease.

Although HL was initially described nearly 2 centuries ago by Thomas Hodgkin, the cell of origin remained elusive until the 1990s when molecular analysis of microdissected Reed-Sternberg (RS) cells revealed the presence of immunoglobulin heavy chain gene rearrangements.¹  Since that time, our understanding of the unique biology of HL has expanded rapidly, laying the groundwork for a growing number of effective targeted therapies.

The vast majority of RS cells arise from postgerminal center B cells that do not express typical B-cell markers on the cell surface. Instead, their immunophenotype is characterized by nearly universal expression of CD30, a marker of T-cell activation, and frequent expression of the myeloid marker CD15. Because CD30 expression is limited to a subset of activated lymphocytes and eosinophils, it was identified as an attractive target for HL therapy. Brentuximab vedotin (BV) is an anti-CD30 antibody drug conjugate that facilitates the targeted delivery of the microtubule-disrupting agent monomethyl auristatin E to the HL tumor microenvironment. In the pivotal study of BV, 102 heavily pretreated patients (median, 3.5 prior therapies; 89% relapsed after autologous stem cell transplantation [ASCT]) were treated with BV every 3 weeks for up to 16 cycles.²  The overall and complete response rates were 75% and 33%, respectively, with a median progression-free survival (PFS) of 9.3 months. Peripheral neuropathy occurred in 56% of patients, including 11% with grade 3 neuropathy; however, the majority of patients (80%) had resolution or improvement in their symptoms with discontinuation of BV. These results prompted a U.S. Food and Drug Administration approval for BV and led to a series of trials testing BV in earlier phases of treatment and in combination with other agents.

HL is also characterized by an unusual tissue architecture composed of rare RS cells surrounded by a complex network of lymphocytes, granulocytes, macrophages, and plasma cells that are unable to mount an effective antitumor immune response. RS cells support this unusual immune environment and evade immune detection through several mechanisms: cytokines and chemokines, altered antigen presentation, and genetic alterations in the programmed cell death 1 (PD-1) pathway.

RS cells secrete a diverse set of cytokines and chemokines that alters the composition and function of the cells in the surrounding microenvironment (eg, blunting the functional activity of effector T cells and shifting polarization toward a T regulatory phenotype).³ 

The expression of major histocompatibility MHC class I (MHC-I) and MHC-II receptors on RS cells is decreased or absent in the majority of patients with HL. Decreased expression of MHC-I may prevent CD8 T-cell recognition and has been linked with inferior PFS following front-line therapy.⁴ 

Nearly all patients with HL harbor genetic alterations in 9p24.1, an amplicon that contains the genes for both PD-1 ligands: PD-L1 and PD-L2.⁵  The amplicon also contains the JAK2 gene, and augmented JAK/STAT signaling generates additional production of PD-L1 and PD-L2 proteins. Together, these factors drive abundant expression of PD-1 ligands on RS cells that can bind to PD-1 on infiltrating T cells, leading to T-cell inactivation and immune evasion. Reverse signaling via PD-L1 may also promote RS cell survival and proliferation via activation of the MAPK pathway and increased mitochondrial oxygen consumption.⁶  Like MHC-I expression, the degree of 9p24.1 alteration appears to be a predictor of response to frontline therapy, with more severe 9p24.1 alterations predicting less favorable outcomes.⁵ 

Based on nearly universal alterations in PD-1 signaling in HL, expansion cohorts of HL patients were included in phase 1 trials of nivolumab and pembrolizumab. Both drugs were well tolerated and achieved high response rates among heavily pretreated patients.^7,8  Pivotal phase 2 trials confirmed high overall response rates (69% for both nivolumab and pembrolizumab) among larger cohorts of relapsed or refractory (R/R) patients.^9,10  Notably, response rates for patients with high-risk features, such as primary refractory disease or BV-refractory disease, were similarly excellent in both trials. Immune-related adverse events (AEs), including thyroiditis, rash, colitis, and pneumonitis, occur with PD-1 blockade; fortunately, grade 3-4 AEs are infrequent. In the pivotal studies for PD-1 monoclonal antibodies (mAbs), only 4% to 6% of patients discontinued study treatment because of drug toxicity.^9,10  These results led to the U.S. Food and Drug Administration approval of nivolumab and pembrolizumab for patients with R/R HL. Both drugs are now being tested across all phases of treatment, with the goal of identifying the optimal timing for PD-1 blockade in HL therapy. Other drugs, like avelumab, which target the PD-1 synapse at the ligand level are in earlier stages of clinical development, but have also demonstrated activity in patients with R/R HL.¹¹  Although these drugs could facilitate enhanced antibody-dependent cell-mediated cytotoxicity, they leave intact interactions between PD-L2 and PD-1, which may be sufficient to maintain the immunosuppressive microenvironment in HL. Additional studies are necessary to determine the role of PD-L1 blockade in HL.

Although outcomes of frontline treatment with ABVD and BEACOPP are excellent, with cure rates exceeding 75% in advanced-stage patients, there is room for improvement, both in terms of reducing relapse rates and preventing short- and long-term treatment complications.¹²  Novel agents are being tested in the frontline setting in combination with standard treatment of fit patients or as part of less intensive regimens in older unfit patients. The largest such trial to date was the ECHELON-1 trial, which randomized patients with stage III/IV HL to 6 cycles of ABVD or 6 cycles of adriamycin, vinblastine, and dacarbazine (BV-AVD).¹³  The primary end point of the trial was modified PFS, a composite outcome that includes death, progression, or noncomplete response requiring subsequent anticancer therapy. After a median follow-up of 25 months, the 2-year modified PFS was significantly higher among patients treated with BV-AVD (82.1% vs 77.2%); however, there was no significant overall survival (OS) benefit. BV-AVD was associated with a lower rate of pulmonary toxicity but higher rates of peripheral neuropathy and febrile neutropenia, which led to an amendment mandating growth factor support for patients receiving BV-AVD. The adoption of BV-AVD has not been uniform, particularly in lower-risk patients, given the modest benefit, toxicity profile, and expense of therapy.

Numerous smaller trials are underway testing PD-1 mAbs as a component of frontline HL therapy. Based on encouraging preliminary data from these trials, a large North American Intergroup randomized phase 3 trial will compare nivolumab-AVD with BV-AVD in newly diagnosed advanced-stage HL patients. The trial includes a preplanned cost-effectiveness analysis that will be critical to assess the overall benefit according to prespecified risk groups.

Depending on the selection of front-line therapy, ∼10% to 15% of patients with early-stage HL and 15% to 30% of patients with advanced-stage HL will have primary refractory lymphoma or experience recurrence. The standard approach for R/R disease is salvage therapy, which has historically consisted of a combination of chemotherapy agents not used in front-line treatment. For fit patients who achieve a response, consolidation with ASCT is recommended, based on 2 randomized trials from the 1990s that showed an improvement in PFS with this approach.^14,15  Salvage regimens, such as ICE (ifosfamide, carboplatin, etoposide); DHAP (dexamethasone, cytarabine, and cisplatin); ESHAP (etoposide, methylprednisolone, cytarabine, cisplatin); and GVD (gemcitabine, vinorelbine, pegylated liposomal doxorubicin) have all been used with similar results.¹⁶  In recent years, BV and PD-1 mAbs have been incorporated into salvage regimens with the goal of increasing the complete response rate (CRR) on pretransplant FDG PET, which is a strong predictor of posttransplant outcomes.¹⁷ 

Two trials have tested a sequential salvage strategy starting with BV monotherapy followed by combination chemotherapy (augmented ICE¹⁸  or investigator’s choice¹⁹ ) for BV noncomplete responders. With this approach, 27% to 43% of patients achieved a complete response (CR) with BV alone and proceeded directly to ASCT without more intensive multiagent salvage therapy. Many BV noncomplete responders achieved favorable outcomes with subsequent multiagent chemotherapy. In total, a BV-based sequential approach achieved a pre-ASCT PET CRR of 68% to 76% and resulted in excellent 2-year PFS and OS (Table 1). Other trials have investigated BV in combination with other agents. A phase 2 trial tested the combination of BV and bendamustine, a bifunctional alkylating agent with strong single-agent activity in R/R HL. The combination was well tolerated apart from frequent infusion-related reactions, which could be mitigated with high-dose steroid premedication. After 2 cycles of therapy, the overall response rate (ORR) and CRR were 93% and 74%, respectively.²⁰  The combination of BV and ESHAP also achieved impressive response rates (ORR, 91%; CR,70%) but appeared to be more toxic, with 5% of patients succumbing to treatment-related complications.²¹  Likewise, efforts to combine BV with gemcitabine-based regimens were halted after a trial of BV + GVD was stopped for excessive pulmonary toxicity.²² 

Based on the single-agent activity and unique toxicity profile of PD-1 mAbs, PD-1–based combinations are also being tested in the salvage setting. The largest trial to date examined the combination of nivolumab and BV in 62 patients and reported an ORR of 82% and a CRR of 61%. PFS was excellent, particularly for the 42 patients (68% of total) who proceeded directly to ASCT after 4 cycles of BV + nivolumab (21-month PFS, 97%).²³  Additional studies are testing PD-1 mAbs in other combinations. One trial uses a sequential strategy of nivolumab alone with escalation to nivolumab + ICE for patients who do not achieve a CR to nivolumab (NCT03016871). Two additional trials that combine pembrolizumab with ICE (NCT03077828) or GVD (NCT03618550) are ongoing and could provide further support for the inclusion of PD-1 mAbs in salvage regimens.

Thus far, no randomized trial has compared salvage regimens in HL, and cross-trial comparisons are challenging given the relatively small size of phase 2 trials, differences in response assessment, and subtle variations in trial populations. Although CRRs are important, ultimately PFS and OS are the most relevant but can be challenging to analyze because of differences in utilization of ASCT and maintenance therapy between trials. Prospective randomized studies are needed to define the optimal strategy for patients with R/R disease; however, in the meantime, selection of novel highly effective salvage regimens can be considered based on individual patient factors.

Despite high-dose chemotherapy and ASCT, many patients will relapse, particularly those who do not achieve a metabolic CR after salvage chemotherapy. With the goal of eliminating minimal residual disease and preventing subsequent relapse, consolidation or maintenance strategies have been developed using novel agents. The AETHERA trial randomized 329 high-risk R/R HL patients (primary refractory disease, relapse occurring <12 months from the end of frontline therapy, and/or extranodal disease at relapse) to 16 cycles of BV or placebo after ASCT. With a median follow-up of 5 years, the PFS was significantly higher among patients treated with BV (59%) compared with placebo (41%) (hazard ratio, 0.521; 95% confidence interval, 0.302-0.596). Peripheral neuropathy was the most common AE associated with maintenance BV, and it completely resolved (73%) or improved (17%) for the vast majority of patients following completion of therapy.²⁴ 

PD-1 mAbs may be particularly well suited for the posttransplant setting based on innate immune system activation, increased antigen presentation, and a relative increase in effector immune cells seen immediately after ASCT. To test this approach, a phase 2 trial treated 30 patients with R/R HL with 8 doses of pembrolizumab after ASCT. Therapy was well tolerated in the majority of patients, with only 3 patients experiencing grade 3-4 AEs and 4 patients discontinuing therapy because of toxicity. The trial met its primary end point with an 18-month PFS of 82% in 28 evaluable patients.²⁵  Other phase 2 trials are exploring PD-1–based maintenance strategies, including trials of nivolumab maintenance (NCT03436862) and nivolumab + BV maintenance (NCT03057795). Ultimately, however, randomized trials are needed to define the role of PD-1 mAb in this setting.

The patient achieved a CR (Deauville 2) following 2 cycles of bendamustine and BV. She underwent stem cell collection with granulocyte colony-stimulating factor, followed by high-dose chemotherapy with carmustine, cytarabine, etoposide, and melphalan, which was complicated by steroid-responsive pneumonitis. Prior to starting BV maintenance therapy, she was recommended to undergo proton beam therapy given the location and extent of her mediastinal disease. However, before her planning visit for radiotherapy, she developed tachycardia, dyspnea, and microcytic anemia and was found to have recurrent disease localized to the mediastinum.

Historically, patients who are ineligible for or relapse following ASCT have experienced poor outcomes, with a median OS of ∼2 years.²⁶  However, with the development of new highly active agents, the prognosis for chemotherapy-resistant disease has improved dramatically. Initial trials of BV and PD-1 mAbs in multiply relapsed patients demonstrated high ORRs in patients with multiple R/R disease; however, CRRs were low (∼33% for BV and 20-25% for PD-1 mAbs), and the vast majority of patients eventually relapsed following these treatments.^2,9,10  In the pivotal phase 2 trial of BV, the median duration of response was 6.7 months. Notably, the subset of patients achieving a CR (34 of 102 patients) achieved more durable remissions (median duration of response, 20.5 months). Among 27 complete responders who did not receive consolidative transplantation or radiation, 9 patients (9% of the total study) remained in a CR after >5 years of follow-up, suggesting that BV may be curative in this small subgroup.²⁷  Depth of response also appears to be a predictor of response duration for PD-1 inhibitors. In the phase 2 trial of nivolumab, patients achieving a CR had more durable remissions (median, 20.3 months) than did patients with a partial response (median, 15.1 months). However, the majority of responders relapsed within 2 years of starting treatment, suggesting that PD-1 blockade alone is also unlikely to be curative for most patients.⁹ 

As discussed above, investigators are moving BV and PD-1 mAbs to the first- and second-line setting with the goal of increasing the cure rate of initial treatments. Novel agents, particularly PD-1 mAbs, are also being tested as combination partners for patients with multiply relapsed disease with the goal of achieving deeper and more durable remissions. Based on successful dual checkpoint inhibition in other malignancies, investigators tested the combination of nivolumab and ipilimumab in 31 patients with R/R HL. There was a substantial increase in the frequency and severity of immune-related AEs compared with PD-1 monotherapy without a clear benefit in efficacy (ORR, 74%; CRR, 19%; Table 2).²⁸  In another trial, the triplet combination of BV + nivolumab + ipilimumab was tested in 22 patients and achieved ORR and CRR of 82% and 68%, respectively. This trial also reported higher rates of grade 3 AEs and dose-limiting toxicities compared with PD-1 monotherapy.²⁹  Longer follow-up is necessary for both trials to determine whether more durable remissions might warrant the added toxicity of either combination approach. The bispecific tetravalent antibody AFM-13, which targets CD30 on RS cells and CD16A on natural killer cells, has modest single-agent activity in HL and was also combined with pembrolizumab in a phase 1b trial. The combination resulted in frequent infusion reactions, including grade 3-4 reactions in 13% of patients, but it demonstrated an encouraging efficacy signal with an ORR of 87% and a CRR of 39% among 29 evaluable patients.³⁰  Many other PD-1–based combination trials are ongoing, including trials that target other checkpoint receptors, like LAG3 (NCT03598608, NCT02061761), and cell signaling targets, like JAK2 (NCT 03681561) and Bruton tyrosine kinase (NCT02940301, NCT02362035).

Multiple other novel agents are currently in clinical trials with promising preliminary results. ADCT-301, a CD25-directed antibody conjugated to pyrrolobenzodiazepine, achieved an ORR of 88% and a CRR of 43% in 37 heavily pretreated patients treated with the target dose of 45 mg/kg.³¹  The drug was associated with Guillain-Barré syndrome in 7.5% of all patients treated (n = 67), as well as frequent reversible liver function test abnormalities, skin toxicity, and edema/effusions. Additional correlative studies are planned for future ADCT-301 trials to better understand the mechanism of Guillain-Barré syndrome, which is currently unknown. Several small molecule inhibitors have also been tested in the R/R setting with more modest efficacy signals. Two clinical trials have tested JAK2 inhibitors (ruxolitinib³²  and SB1518³³ ) in R/R HL based on frequent copy gain and amplification of the JAK2 gene, which is located on chromosome 9p24.1. Both trials demonstrated an ORR < 20% in HL patients, which could reflect the poor specificity of the 2 inhibitors for the JAK2 isoform. In a phase 2 trial of 38 patients, lenalidomide had limited activity, with an ORR of 19% and median PFS of 4 months.³⁴  After encouraging early data, a phase 2 trial treated 129 patients with the histone deacetylase inhibitor panobinostat and reported disappointing ORR and CRR of 23% and 4%, respectively.³⁵  Based on encouraging preclinical data, the phosphatidylinositol 3-kinase inhibitor idelalisib was tested in patients with R/R HL, but it was associated with an ORR of only 20% among 25 heavily pretreated patients.³⁶  Finally, the mechanistic target of rapamycin inhibitor everolimus achieved higher response rates (ORR, 42%; CRR, 8%) and a longer PFS (median, 9 months) in a phase 2 study of 57 heavily pretreated patients.³⁷  Although small molecule inhibitors have activity in some patients with HL, novel combinations may ultimately be needed to achieve meaningful responses.

Given the success of chimeric antigen receptor (CAR) T-cell therapy in diffuse large B-cell lymphoma, trials of this approach are ongoing in HL. In the first anti-CD30 CAR T-cell trial, which did not use lymphodepleting chemotherapy, ORRs and persistence of engineered T cells were disappointing. However, early results from a subsequent trial that incorporated preinfusion cyclophosphamide and fludarabine were encouraging, with 4 of 9 patients achieving ongoing responses with short follow-up.³⁸  Thus far, AEs associated with CAR T cells in HL appear less problematic than in aggressive B-cell non-HL. Cytokine release syndrome has been mild, and neurotoxicity has not been seen, although a self-limited maculopapular rash is common.

EBV-directed cytotoxic T cells (CTLs) directed against LMP-1 and LMP-2 have activity in EBV⁺ HL. CTLs with enhanced activity against LMP-1/2 using adenoviral vector–transduced dendritic cells and EBV-transformed B-lymphoblastoid cell lines were tested in patients with EBV-associated HL and non-HL. Thirteen of 21 patients with active disease responded to therapy, with 11 CRs.³⁹  In a recent small study, Bollard and colleagues attempted to counteract transforming growth factor-β–associated immune suppression in the tumor microenvironment by engineering LMP-specific CTLs that express dominant-negative transforming growth factor-β receptor 2.⁴⁰  Four of 7 patients treated with these CTLs (in the absence of lymphodepleting chemotherapy) achieved clinical responses, and 2 patients remain in CR beyond 4 years. Other cellular therapy approaches, including autologous T cells directed against multiple tumor-associated antigens, including WT1, PRAME, and SURVIVIN, are being studied in relapsed HL.

The role and timing of allogeneic stem cell transplantation (allo-SCT) for patients who relapse following ASCT is not clearly defined in the era of immune checkpoint blockade. Allo-SCT can achieve long-term remissions for many patients, but it is also associated with high rates of nonrelapse mortality. A meta-analysis including 42 studies and 1850 patients demonstrated that allo-SCT outcomes have improved over time. Studies accruing patients in 2000 or later had 5% to 10% lower nonrelapse mortality and a 15% to 20% improvement in relapse-free survival and OS.⁴¹  Several studies have suggested that allo-SCT after PD-1 blockade may be associated with increased early immune complications, including several cases of fatal acute graft-versus-host disease.⁴²  For such patients, caution is advised. Consensus recommendations⁴³  include discontinuation of checkpoint inhibitor therapy ≥6 weeks before transplantation and using posttransplant cyclophosphamide-based graft-versus-host disease prophylaxis, which may reduce the risk of early immune complications. Based on several reports that suggest that PD-1 mAbs may sensitize patients to subsequent therapies,^44,45  delaying allo-SCT (and its attendant toxicities) may be a reasonable approach for patients responding to novel agents.

The patient was treated with nivolumab, 240 mg every 2 weeks, with rapid improvement in fatigue and tachycardia. After 2 cycles, she will undergo restaging and proceed to proton beam radiotherapy if her disease is responding. Thereafter, we will resume nivolumab with a plan to treat for 6 months beyond CR. If she does not achieve a CR, we will investigate clinical trials with or without consolidation with allo-SCT.

Recent advances in HL, including the approval of BV and PD-1 mAbs, have improved outcomes for patients with R/R disease. Ongoing efforts to develop novel drugs and treatment combinations should reduce and may eventually eliminate the need for transplantation. In the upfront setting, identifying the minority of patients who do not experience long-term disease control with initial chemotherapy is critical. In addition, less toxic therapies are needed for elderly patients or those with comorbidities that preclude combination chemotherapy. Supported by extensive advances in our understanding of the biology of HL and identification of prognostic and predictive biomarkers, the field continues to move forward at a rapid pace.

Ann LaCasce, 450 Brookline Ave, Boston, MA 02215; e-mail: ann_lacasce@dfci.harvard.edu.