Evidence-Based Minireview: What is the optimal timing of anti–PD-1 antibodies in relapsed classical Hodgkin lymphoma?

The programmed cell death (PD-1) pathway is integral to classical HL tumorigenesis and immune evasion. The malignant Hodgkin Reed-Sternberg cell contains genetic alterations of the PD-L1/PD-L2 locus on chromosome 9p24.1, which predicts increased responsiveness to PD-1 blockade.¹  Anti–PD-1 antibodies exploit HL reliance on this pathway. Monotherapy with nivolumab or pembrolizumab has been clinically approved in relapsed or refractory (rel/ref) HL based on early-phase clinical trials demonstrating efficacy and tolerability in this setting.^2-4  We aim to describe the optimal timing and use of PD-1 inhibitors in rel/ref HL.

Early-phase studies of pembrolizumab and nivolumab were conducted primarily in patients who experienced both BV and autologous stem cell transplantation (ASCT) failure after a median of 4 to 5 lines of therapy.^2-4  The phase 2 Checkmate-205 trial assessed nivolumab in 3 cohorts of patients with relapse following ASCT (n = 243).²  The 2 cohorts with failure of both BV and ASCT (n = 180) had an overall response rate (ORR) of ∼70% but rare complete responses (CRs) (12% to 13% in cohorts B and C). Progression-free survival (PFS) was nearly double in patients experiencing CR compared with partial response (PR) or stable disease (SD) (22, 15, and 11 months, respectively). Therapy was well tolerated with minimal impact on blood cell counts; 4% had immune-related adverse events (AEs) leading to discontinuation including pneumonitis (n = 2) and autoimmune hepatitis (n = 1).

Pembrolizumab in this setting demonstrated similar responses and AEs.^3,4  The phase 2 KEYNOTE-087 enrolled 210 patients with rel/ref HL into 3 cohorts: (1) failure of BV and ASCT, (2) BV failure and ASCT ineligible, and (3) BV naive with ASCT failure.³  Cohort 1 demonstrated an ORR of 74% and CR of 15%, however, responses were similar across all cohorts, and there was no major difference by type or number of prior lines of therapy. The median overall survival (OS) was not reached, with only 4 deaths in the study time period; 9-month OS and PFS rates were 97.5% and 63.4%, respectively, for the entire cohort (Table 1).

There are limited data in the pretransplant setting as the nivolumab trials were largely performed after transplant failure. Cohort 2 of KEYNOTE-087 included patients with BV failure who did not achieve adequate response to proceed to ASCT (n = 81).³  The ORR was 64.2% and CR rate was 20%, again indicating little difference in response by prior therapy. Patients with primary refractory disease performed remarkably well with an ORR of 80%.

There are comparatively less data for PD-1 inhibition in the BV-naive setting. Sixty-three patients in cohort A of the Checkmate-205 study were BV naive and had an ORR of 65% and a CR rate of 29%.²  Similarly, 35 BV-naive patients in cohort C of KEYNOTE-087 had an ORR of 71.4% and a CR of 20%.³  Based on these results, a phase 3 analysis of pembrolizumab vs BV in rel/ref HL regardless of prior ASCT status is ongoing (NCT02684292). Overall, these data indicate that PD-1 therapy is highly effective in the rel/ref setting with similar response rates regardless of prior therapy.

A retrospective international study of 39 patients, 79% of whom had relapsed HL, treated with PD-1 blockade prior to allogeneic transplantation revealed low relapse rates after transplant, though graft-versus-host disease (GVHD) occurred commonly.⁵  Patients received PD-1 blockade at a median of 62 days prior to transplant with a range of 7 to 260 days. Thirty-six percent and 26% of patients achieved CR and PR, respectively. Among 31 patients with HL, the ORR was 74%. There was no association with graft type or donor source. At 1 year, rates of acute GVHD were 44% and 23% for grade 2-3 and grade 3-4 events, respectively. Chronic GVHD occurred in 41% of patients. One-year PFS was 76% and OS was 89% with 4 treatment-related deaths including 3 from acute GVHD and 1 from hepatic sinusoidal obstruction syndrome. The cumulative incidence of relapse at 1 year was 14%, and nonrelapse mortality was 11%. Correlative studies performed on blood samples demonstrated that compared with controls, patients who received PD-1 inhibitors had lower numbers of both PD-1⁺ T cells and decreased ratios of regulatory T cells compared with CD4⁺ and CD8⁺ cells.

Similarly, 31 patients (29 HL) who received PD-1 blockade after allogeneic transplant experienced high response rates, though severe GHVD was a frequent complication.⁶  The vast majority of patients received reduced-intensity conditioning and peripheral blood stem cells. Approximately one-half had matched-related donors and 13% received haploidentical transplants. With a median follow-up of 428 days, ORR was 77% with 15 CR, 8 PR, and 11 progressive disease; 21 (68%) were alive at last follow-up. Seventeen (55%) developed GHVD after a median of 102 doses of anti–PD-1 antibodies, and 8 patients died as a complication of GVHD. The majority of patients developed grade III/IV GVHD, which was poorly responsive to therapy. Two of 17 patients experienced complete resolution of GHVD and 14 required >1 therapy.

PD-1 antibodies are being studied in many other settings, including as consolidation post-ASCT,⁷  as salvage prior to ASCT combined with chemotherapy, and frontline combined with chemotherapy.⁸  A small phase 2 study of consolidation pembrolizumab (n = 30) for 8 cycles starting 3 weeks post-ASCT revealed an 18-month PFS of 82%, and 100% OS, but with increased immune-related AEs (40% grade 1 or higher). These results met the primary end point in achieving >60% PFS, supporting future randomized comparisons against BV. In the frontline setting, nivolumab was studied as a single-agent window then concurrent with doxorubicin, vinblastine, dacarbazine (AVD) in 51 patients with advanced-stage HL.⁸  At the end of monotherapy, the ORR was 69% with 18% achieving CR. Follow-up is limited, but the reported 9-month modified PFS was 92%.

Despite high activity and durability in some patients, anti–PD-1 antibodies are associated with a pattern of continued relapse, with a median duration of response of 1 to 2 years, and therefore are not considered to be curative as single agents in most patients. Our current approach to anti–PD-1 antibodies in HL is to reserve their use for patients refractory to both high-dose chemotherapy and BV in the pre-ASCT setting, and after BV failure post-ASCT. Patients with single site or minimal progression on anti–PD-1 antibodies may be considered for treatment beyond progression if they have not experienced significant immune-related AEs. Given relapse after discontinuation of PD-1 therapy, we typically continue treatment in patients deriving clinical benefit in the absence of significant toxicity.

We consider allogeneic transplant cautiously in patients achieving a CR to PD-1 blockade who have relapsed after ASCT. Further studies are needed to identify pretransplant risk factors for the development of GVHD, such as time from therapy to stem cell infusion as well as the optimal prophylaxis to reduce GVHD. Some centers are using posttransplant cyclophosphamide as one such strategy. After transplant, patient selection and close monitoring is key, as PD blockade is associated with increased incidence of severe and treatment refractory GVHD, and existing therapies to treat subsequent GVHD are suboptimal.

The ideal timing of PD-1 blockade in the future will depend upon additional data from ongoing and future randomized studies, including a randomized comparison in the frontline setting of a nivolumab-AVD vs BV-AVD. In the relapsed setting, PD-1 therapy combined with BV NCT02572167 and prior to ASCT NCT03016871 is being actively investigated. Future trials assessing whether pembrolizumab or BV will be the optimal posttransplant consolidation strategy are also warranted. As BV is increasingly incorporated into frontline treatment, anti–PD-1 antibodies will likely experience increased usage earlier in the relapsed course. Multiple trials have demonstrated similar response rates with anti–PD-1 antibodies regardless of prior therapy, therefore they will likely remain an option for patients early or late in their treatment course.

1. Anti–PD-1 antibody therapy with nivolumab or pembrolizumab is recommended following failure of BV and autologous transplant in patients with relapsed classical HL (GRADE A).

2. Anti–PD-1 antibody therapy with pembrolizumab may be considered prior to BV failure in patients who are contraindicated to brentuximab given similar response rates in a limited number of patients (GRADE C).

3. Caution should be exercised in patients in the periallogeneic setting to weigh the benefits of high response rates with the increased risk of GVHD (GRADE C).

Pamela Blair Allen, Department of Hematology and Oncology, Winship Cancer Institute of Emory University, Atlanta, GA 30322; e-mail: pallen5@emory.edu.