Reduced incidence of relapse after checkpoint inhibitors relative to brentuximab vedotin as salvage therapy before allogeneic stem cell transplantation for refractory/relapsed Hodgkin lymphoma: A retrospective analysis

Eighty-two patients were eligible for allogeneic transplantation (Allo-SCT) at our Institution between April 2014 and August 2022. Of these, 72 actually received Allo-SCT, n = 44 after checkpoint inhibitor (CPI) containing salvage therapy (CPI cohort) and n = 28 with brentuximab-vedotin (BV) containing therapy without CPI (BV cohort). With a median follow-up of 63 months, the CPI cohort had improved cumulative incidence of relapse (5% vs. 37%, p = 0.002) and progression-free survival (PFS) (79% and 56%, p = 0.049) relative to the BV group. By multivariable analysis, pretransplant CPI resulted in the only independent predictor of relapse (hazard ratio [HR]: 0.124, 95% confidence interval 0.270–0.570, p = 0.007) and a strong predictor for PFS (HR 0.43, 95% CI 0.18–1.02, p = 0.057). Our findings support the use of CPI over BV as the first-line salvage therapy for R/R HL patients relapsing after autologous stem cell transplantation when consolidation with Allo-SCT is planned. The prognosis of patients with relapsed/refractory Hodgkin lymphoma (R/R HL) has significantly improved with the introduction of drug-conjugated antibodies, such as brentuximab vedotin (BV), and checkpoint inhibitors (CPI), such as nivolumab and pembrolizumab. Even if both compounds have shown high overall response rates,1-3 chance of achieving complete remission (CR) remains low, resulting in a 5-year progression-free survival (PFS) of 22% for BV1 and a median PFS of 13–15 months for CPI.2, 3 In this scenario, allogeneic stem cell transplantation (Allo-SCT) remains the only curative option. The introduction of high-dose post-transplant cyclophosphamide (PT-Cy) as graft-versus-host disease (GVHD) prophylaxis markedly enhanced the long-term outcome of R/R HL relative to older studies,4, 5 achieving a 2-year overall survival (OS) of 70%–80% and PFS of 60%.6 Several centres have assessed Allo-SCT outcomes in patients treated with BV or CPI. Chen et al.7 found improved PFS due to lower cumulative incidence of relapse (CIR) in those receiving BV over chemotherapy. However, this observation was not confirmed by a large retrospective EBMT analysis.8 Similarly, De Philippis et al.9 reported superior PFS and reduced CIR of patients undergoing Haplo-SCT after CPI. Merryman et al.10 observed a 2-year PFS of 69% in CPI-treated patients, but noted a trend towards increased acute graft-versus-host disease (GVHD). On the contrary, Paul et al.11 found that pretransplant treatment with CPI resulted in an improved PFS relative to non-CPI (90% vs. 65%) without an increased risk of GVHD. Notably, none of these studies has directly compared outcomes in patients undergoing Allo-SCT after treatment with CPI versus BV. To address this gap, we conducted a retrospective analysis at our institution, comparing Allo-SCT outcomes in patients who received salvage regimens with BV without CPI versus those treated with CPI-based rescue strategies. For Patients and Methods, see Supporting Information. Between 2014 and 2022, 82 patients were evaluated for Allo-SCT at our centre for an R/R HL. Of these, 10 did not proceed to transplant because of progressive disease (n = 1 after BV, n = 1 after CPI, n = 5 after BV and CPI, n = 3 after chemotherapy only), while 72 received Allo-SCT. Patients' characteristics are summarized in Table 1: 28 received only BV without CPI (BV cohort) before Allo-SCT, while 44 were treated with CPI, either with (n = 28) or without (n = 16) prior BV exposure (CPI cohort). The two cohorts were comparable in terms of patients, disease and transplant characteristics. Median time from transplant of the last immunotherapy treatment was 97 days (interquartile range [IQR] 39–164) for BV and 62 days (IQR 42–133) for CPI. The median follow-up was 63.7 months for the BV group and 60.9 months for the CPI group. The median time to achieve unsupported ANC counts was 19 days in both the BV cohort (range 12–26) and the CPI cohort (range 12–47; p = 0.628). The median time to reach a platelet count above 20 000/μL was 24 days (range 11–89) in the BV cohort and 22.5 days (range 11–47) in the CPI cohort (p = 0.758). Consistently, the 30-day cumulative incidence of ANC and platelet engraftment did not differ among the two groups (Figure S1). With a median follow-up of 63 months, 11 patients relapsed at a median of 239 days after Allo-SCT. Notably, nine relapses occurred in the BV cohort, while only two were observed in the CPI cohort. This corresponded to a 5-year CIR of 37% for BV and 5% for CPI (Figure 1A; Table 2, p = 0.002). Univariate analysis confirmed this finding (Table S1). Of note, by multivariable analysis (Table S2), pretransplant CPI resulted the only independent predictor of relapse (hazard ratio [HR] 0.124, 95% confidence interval 0.270–0.570, p = 0.007). A total of 15 patients died at a median of 7.2 months post-transplant: 7 (25%) in the BV cohort and 8 (18%) in the CPI cohort (Table S3). More deaths were due to disease relapse in the BV cohort, while the CPI group experienced more deaths related to post-transplant complications. Consequently, the 5-year non-relapse mortality (NRM) was 7% in the BV cohort and 16% in the CPI cohort (Figure 1B; Table 2, p = 0.305). By univariate and multivariate analyses, no variable was an independent predictor of NRM (Tables S1 and S2). The 6-month cumulative incidence of grade II–IV and III–IV acute GVHD was 32% and 4% for BV and 48% and 9% for CPI respectively (Figure S2A/B; Table 2). The 5-year cumulative incidence of moderate-to-severe chronic GVHD was 9% for the BV and 15% for the CPI group (Figure S2C; Table 2, p = 0.413). One-year cumulative incidence of invasive fungal disease, Epstein–Barr virus (EBV), human herpesvirus 6 (HHV6) and BK virus was comparable between the two cohorts (Table 2). In contrast, 1-year cumulative incidence of cytomegalovirus (CMV) reactivation was significantly lower in the CPI cohort (32%) than in the BV cohort (57%; p < 0.01) (Table 2; Figure S2D). In order to avoid confounding due to the introduction of letermovir prophylaxis in 2020, we conducted a subanalysis considering only patients transplanted in the pre-letermovir era (n = 58), confirming a significantly lower risk of CMV reactivation in CPI-treated patients than in BV-treated patients (42% vs. 64%; p = 0.035). Macrophage activation syndrome (MAS) was reported in six patients, occurring more frequently in the CPI cohort (5 patients, 11%) than in the BV cohort (1 patient, 4%), though the difference was not statistically significant (p = 0.244). Thrombotic microangiopathy (TAM) was observed in two patients, both in the CPI cohort, while one patient in the BV cohort developed veno-occlusive disease. There was no significant difference in OS or GVHD/relapse-free survival (GRFS) between the two cohorts. Five-year OS was 73% in the BV cohort and 81% in the CPI cohort, while 5-year GRFS was 52% versus 58% respectively (Table 2). Five-year PFS was significantly higher in patients who received CPI before Allo-SCT than in those treated only with BV (79% vs. 56%; p = 0.049) (Table 2; Figure 1C). In univariate and multivariable analyses (Tables S1 and S2), only pretransplant disease status remained an independent predictor of OS, while pretransplant CPI was a stronger predictor of PFS (HR 0.43, 95% CI 0.18–1.02, p = 0.057) compared to pre-transplant disease status. This analysis suggests that CPI as salvage therapy before Allo-SCT is associated with a lower risk of post-transplant relapse, improved PFS and a reduced incidence of CMV reactivation, supporting its role as a valuable pretransplant strategy for patients with R/R HL. Brentuximab vedotin and PD-1 inhibitors are widely used as salvage therapy for R/R HL, but no direct comparison between these strategies as a bridge to Allo-SCT has been reported so far. The KEYNOTE-204 trial12 showed that CPI significantly improved PFS compared to BV, but specific data for allo-transplant recipients were not reported. Moreover, Spinner et al.13 reported an improved PPS among HL patients relapsing post-autologous stem cell transplantation (ASCT) and treated either with BV or CPI, but no direct comparison between CPI and BV was made. De Philippis et al.9 found that CPI before haploidentical transplantation was associated with improved 2-year PFS due to a lower CIR, though the non-CPI cohort included patients exposed to chemotherapy and/or BV. Desai et al.14 found that PPS was significantly higher in patients receiving CPI compared to chemotherapy alone, whereas BV-treated patients did not show a significant advantage over the no-CPI/no-BV cohort. Consistently, the John Hopkins group11 first described an improved PFS and more recently an advantage in terms of OS for patients receiving CPI relative to the no-CPI cohort, due to reduced relapse incidence.15 Our findings align with and extend these reports, showing that salvage therapy with CPI, compared to BV, was associated with improved 5-year PFS, driven by a lower CIR. Another interesting finding is that the reduced incidence of relapse in the CPI cohort occurred regardless of the timing of CPI administration before Allo-SCT. Among the 44 patients treated with CPI, 26 received PD-1 inhibitors as their last therapy before Allo-SCT, while 18 received either BV or chemotherapy. These results are consistent with recent studies suggesting that CPI may have a chemosensitizing effect.16 The immunomodulatory effects of CPI have raised concerns about an increased risk of post-transplant complications, including acute GVHD and NRM.17, 18 In our analysis, we did not observe a significant difference in terms of acute GVHD and NRM between the CPI and BV cohort, even if both acute GVHD and NRM were more frequent in patients receiving CPI. It is possible that the small sample size did not allow to identify a significant difference in terms of GVHD and NRM, and the lack of difference in OS between the two cohorts may be due to the higher NRM in the CPI group. A recent report from the Société Francophone de Greffe de Moelle et de Thérapie Cellulaire19 found an increased risk of acute GVHD in patients undergoing Allo-SCT within 60 days of their last CPI infusion. In our study, the median interval between the last CPI dose and Allo-SCT was 62 days which may explain the absence of increased toxicity in the CPI cohort. Our study has several limitations, primarily its retrospective design and small sample size. The lack of randomization resulted in two inherently different cohorts; however, their similar pretransplant characteristics allowed for meaningful statistical comparisons. Another key limitation is that most patients in the CPI cohort had prior BV exposure, as Italian regulations previously required BV treatment before CPI could be administered. Interestingly, we observed a reduced CMV reactivation in the cohort receiving CPI relative to BV, potentially due to a post-transplant immune activation induced by CPI, similar to pembrolizumab-related increase of CD4+ and CD8+ activity against JC virus infection.20 In conclusion, our findings support the use of CPI over BV as the first-line salvage therapy for R/R HL patients relapsing after ASCT when consolidation with Allo-SCT is planned. Additionally, for patients with an unsatisfactory response to PD-1 inhibitors, rechallenging with chemotherapy rather than BV may be an option, as CPI treatment appears to enhance chemosensitivity. Study conception and design: SB and JM; collection and interpretation of data: JM, CP, DT, BS, DM and CDP; statistical analysis: LG, JM; manuscript drafting: JM and CP; manuscript editing: SB, AS and LG; approval to submit: all. We wish to acknowledge Luca Giacomelli and Aashni Shah for editorial and linguistic assistance, which was supported by internal funds. The publication fee for this work was covered by the Italian Ministry of Health's 'Ricerca Corrente' funding to the IRCCS Humanitas Research Hospital. None. The authors have no conflict of interests and no sources of support to disclose. Informed consent was obtained from all subjects involved in the study. The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of the Humanitas Cancer Institute on 18 February 2025 (study n. 07/25). No material from other sources was used. This is not necessary for retrospective study. The data presented in this study are available on the Zenodo platform, doi: 10.5281/zenodo.14976720. Figure S1. Figure S2. Data S1. Tables S1–S3. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.