Bone Marrow Transplantation (2017) 52, 1463–1464 © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved 0268-3369/17 www.nature.com/bmt LETTER TO THE EDITOR Association between GvHD and nivolumab in the FDA adverse event reporting system Bone Marrow Transplantation (2017) 52, 1463–1464; doi:10.1038/ were searched including bleomycin, brentuximab vedotin, car- bmt.2017.158; published online 31 July 2017 mustine, chlorambucil, cyclophosphamide, dacarbazine, doxoru- bicin, liposomal doxorubicin, tositumomab, lomustine, mechlorethamine, mercaptopurine, methacycline, nivolumab, The US Food and Drug Administration (FDA) approved nivolumab procarbazine, thiotepa, triethylene melamine, vinblastine and for treatment of relapsed or refractory classical Hodgkin’s vincristine. Among them, we further extracted the subjects whose lymphoma (HL) in May 2016. Nivolumab is supposed to inhibit indications of a primary suspected drug for an adverse event were immune checkpoint programmed cell death 1 (PD-1) and to HL. The data were analyzed using R software (version 3.1.2; R release the brakes on tumor-suppressed lymphocytes.1 In patients Development Core Team, Vienna, Austria). treated with allogeneic hematopoietic stem cell transplantation During the study period, we identified 2720 subjects with HL (allo-HSCT), GvHD could be unleashed through the same reported for an adverse event in the FAERS database. Among 19 mechanism theoretically. Several recent reports have assessed drugs that we searched, no adverse drug reaction was reported the possible association between nivolumab and the onset of for tositumomab, mercaptopurine, methacycline, thiotepa and GvHD for patients treated with allo-HSCT.2–6 In May 2016, the US triethylene melamine, and another 14 drugs which were included FDA made an amendment of the warning and precaution in the in the analyses (Table 1). Among them, 35 (1.29%) cases prescriber’s information concerning complications of allo-HSCT developed GvHD. Of the 72 (2.64%) nivolumab-treated cases, 9 that patients should be followed for hyperacute GvHD, severe developed GvHD (12.5% (95% confidence interval (CI) 5.88–22.4)). acute GvHD, steroid-requiring febrile syndrome, hepatic veno- Of the 2648 HL cases not treated with nivolumab, 26 developed occlusive disease and other immune-mediated adverse reactions. GvHD (0.98% (95% CI 0.64–1.44)). Only cases treated with However, no comparative study with or without nivolumab that cyclophosphamide had a significant point estimate with 10 GvHD assessed the link has published so far. To examine the association cases (3.73% (95% CI 1.80–6.75)). When all of drugs other than between nivolumab and GvHD, we performed a comparative nivolumab were used as a reference, the crude odds ratio for study using the FDA Adverse Event Reporting System (FAERS). GvHD in nivolumab-treated cases was 14.35 (95% CI 5.68–33.3)). The FAERS database was searched using the webpage of the In this study, we found a signal for an increased risk of GvHD Pharmapendium (https://www.pharmapendium.com) in February after nivolumab treatment in HL patients compared with other 2017. Institutional review board approval was not required chemotherapeutic drugs using the FAERS database. There have because the FAERS is an unlinkable anonymized database open been a limited number of studies concerning the association to the public. We extracted reports registered between November between the use of nivolumab and GvHD, and the FAERS database 1997 and September 2016 for drugs approved for HL; 19 drugs has enabled us to investigate the possible association using Table 1. Proportion and odds ratio of GvHD cases in drugs used for Hodgkin’s lymphoma Drug Number of GvHD cases Number of subjects Proportion of GvHD Odds ratio Point estimate (%) 95% CI Point estimate (%) 95% CI Nivolumab 9 72 12.50 (5.88–22.41) 14.35 (5.68–33.29) Other drugs 26 2648 0.98 (0.64–1.44) Reference Bleomycin 0 265 0.00 (0.00–1.38) Brentuximab vedotin 15 917 1.64 (0.92–2.68) Carmustine 0 27 0.00 (0.00–12.77) Chlorambucil 0 7 0.00 (0.00–40.96) Cyclophosphamide 10 268 3.73 (1.80–6.75) Dacarbazine 0 116 0.00 (0.00–3.13) Doxorubicin 1 396 0.25 (0.01–1.40) Liposomal doxorubicin 1 219 0.46 (0.01–2.52) Lomustine 0 1 0.00 (0.00–97.50) Mechlorethamine 0 59 0.00 (0.00–6.06) Procarbazine 0 205 0.00 (0.00–1.78) Vinblastine 0 64 0.00 (0.00–5.60) Vincristine 0 115 0.00 (0.00–3.16) Total 35 2720 1.29 (0.90–1.79) Abbreviations: CI = confidence interval; GvHD = graft versus host disease. Nineteen drugs approved for treatment of Hodgkin’s lymphoma were searched including bleomycin, brentuximab vedotin, carmustine, chlorambucil, cyclophosphamide, dacarbazine, doxorubicin, liposomal doxorubicin, tositumomab, lomustine, mechlorethamine, mercaptopurine, methacycline, nivolumab, procarbazine, thiotepa, triethylene melamine, vinblastine and vincristine. Of these, no adverse drug reactions were found for tositumomab, mercaptopurine, methacycline, thiotepa and triethylene melamine that were indicated for treatment of Hodgkin’s lymphoma. Letter to the Editor 1464 real-world data. The PD-1 and PD-1 ligand pathway provides the factors were not collected systematically, including patient inhibitory signals required for regulation of both central and background information, underlying conditions, and peripheral T-cell tolerance, and previous preclinical studies concomitant drugs. showed an increased risk of GvHD-related lethality in two murine Our results show only a signal for an increased risk that must be models of acute GvHD after immune checkpoint inhibition.7,8 confirmed in further evaluations. However, careful consideration Interestingly, a recent report showed, in allogeneic and should be given to the possibility of an increased risk of GvHD xenogeneic murine GVHD models, that interactions of PD-L1 with when planning to administer an immune checkpoint inhibitor PD-1 on donor CD8+ T cells cause anergy, exhaustion and after allo-HSCT or perform allo-HSCT after an immune checkpoint apoptosis, thereby preventing GvHD in GvHD target tissues, inhibitor treatment in HL patients and appropriate prophylaxis whereas the interactions of PD-L1 with CD80 augment CD8+ T-cell with careful monitoring for GvHD should be considered. expansion without increasing anergy, exhaustion or apoptosis, resulting in strong GvL effects in lymphoid tissues.9 A possible accentuation of GvL effects by PD-1 blockade would be a valuable CONFLICT OF INTEREST theme, but whether GvHD is manageable in a clinical setting YO reports personal fees from Novartis Pharma KK and Sanofi KK, outside the needs further investigation. submitted work; AT reports grants from Bristol-Myers, Pfizer, Chugai Pharmaceutical, In view of the growing concerns about the possible associations Daiichi Sankyo and Sumitomo Dainippon-Pharma, and personal fees from Taiho between GvHD and PD-1 blockade, two small retrospective Pharmaceutical, Rohto Pharmaceutical, Celgene, Novartis, Ohtsuka Pharmaceutical studies have been recently published. One study investigated 20 and Sysmex, outside the submitted work; TT and KY declare no conflict of interest. HL subjects treated with nivolumab as a single agent who relapsed after allo-HSCT; nivolumab-induced GvHD was reported AUTHOR CONTRIBUTIONS in 6 (30%) subjects, and all of them had a prior history of acute GvHD.5 All six patients were managed by standard treatment of YO analyzed data, drafted the paper and contributed analytical tools; TT acute GvHD. Of the six patients, one each died of GvHD, multiple designed research and wrote the paper; KY performed research and analyzed organ failure and progressive disease. The remaining three were data; and AT designed research. alive and still in response at the end of the observation. Another study assessed safety and efficacy of allo-HSCT after PD-1 Y Oshima1, T Tanimoto2, K Yuji1 and A Tojo1 blockade therapy, nivolumab or pembrolizumab, in 39 patients 1The Institute of Medical Science, The University of Tokyo, with relapsed/refractory lymphoma; after a median follow-up of Tokyo, Japan and 12 months, the 1-year cumulative incidences of grade 2–4 and 2Jyoban Hospital of Tokiwa Foundation, Fukushima, Japan grade 3–4 acute GvHD were 44% and 23%, respectively, and the E-mail: [email protected] 1-year incidence of chronic GvHD was 41%.8 In addition, 7 patients developed a noninfectious febrile syndrome shortly after allo- HSCT requiring prolonged courses of steroids. Notably, the REFERENCES authors found that PD-1 blockade may cause persistent depletion 1 Armand P. Immune checkpoint blockade in hematologic malignancies. Blood of PD1+ T cells and alterations in T-cell differentiation that might 2015; 125: 3393–3400. reflect long-lasting effects triggered by prior PD-1 blockade. 2 Angenendt L, Schliemann C, Lutz M, Rebber E, Schulze AB, Weckesser M et al. Such an association between a novel agent and GvHD warrants Nivolumab in a patient with refractory Hodgkin’s lymphoma after allogeneic stem 51 – further studies. For example, an anti-CCR4 antibody, mogamuli- cell transplantation. Bone Marrow Transplant 2016; :443 445. zumab, that is approved for treatment of adult T-cell leukemia/ 3 Haverkos BM, Schowinksy J, Kaplan J, Kamdar M, Kanate AS, Saad A et al. Checkpoint blockade for treatment of relapsed lymphoma following allogeneic lymphoma in Japan depletes regulatory T cells (Tregs) for at least a hematopoietic cell transplant: use may be complicated by onset of severe acute few months. Whereas nivolumab would enhance antitumor graft versus host disease. Blood 2016; 128: 22 (abstract 1163). immunity by suppressing function of T cell by inhibiting PD-1 4 Schoch LK, Borrello I, Fuchs EJ, Bolanos-Meade J, Huo JS, Gojo I et al. Checkpoint pathway, mogamulizumab is supposed to enhance autoimmunity inhibitor therapy and graft versus host disease in allogeneic bone marrow or antitumor immunity by depleting Tregs. A retrospective study transplant recipients of haploidentical and matched products with post transplant indicated that pretransplantation treatment with the mogamuli- cyclophosphamide.