Reduced-Intensity and Non-Myeloablative Allogeneic Stem Cell Transplantation from Alternative HLA-Mismatched Donors for Hodgkin's Lymphoma: A Study By the SFGM-TC (Francophone Society of Bone Marrow Transplantation and Cellular Therapy)

Allogeneic hematopoietic stem cell transplantation (allo-SCT) following a non-myeloablative (NMA) or reduced-intensity conditioning (RIC) is considered a valid approach to treat patients with refractory/relapsed Hodgkin's lymphoma (HL), in particular those having failed a previous autologous SCT (ASCT). When an HLA-matched donor is lacking, a graft from a haploidentical donor (HAPLO), a mismatched unrelated donor (MMUD) or cord blood (CB) might be considered. In this retrospective, registry-based study, we compared the outcome of patients with HL undergoing a RIC or NMA allo-SCT from a HAPLO, MMUD or CB graft.

After detailed review of our database, we found 98 consecutive patients with HL who underwent a NMA/RIC allo-SCT from an alternative HLA-mismatched donor at 24 French and Belgian centers between January 2009 and December 2014. Probabilities of overall survival (OS), event-free survival (EFS) and graft-versus-host disease (GVHD)-free Relapse-free Survival (GRFS) were determined using the Kaplan-Meier method. The cumulative incidences of relapse (CIR), non-relapse mortality (NRM) and cGVHD were studied using a competing risk methodology. We defined GRFS as the probability of being alive without evidence of relapse, grade 3-4 acute GVHD or chronic GVHD (all grades included). Regarding conditioning regimens and intensity, patients in the HAPLO group (n = 34) received a T-cell replete allo-SCT after a NMA (as described by the Baltimore group in Luznik et al, Biology of Blood and Bone Marrow Transplantation, 2008, n = 31, 91%) or a RIC (n = 3, 9%) followed by post-transplant cyclophosphamide (PT CY). Patients in the MMUD group (n = 27) received a variety of NMA (n = 7, 26%) and RIC (n = 20, 74%). All patients in the CB group received a RIC (n = 37, 37%). All but one patient (n = 33) in the HAPLO group received the following GVHD prophylaxis: Cy 50 mg/kg on day +3 and day +4, tacrolimus or CsA and mycophenolate mofetil started on day +5. One patient received only one day of PT CY 50mg/kg at day +3 and also received ATG on day -2 and day -1 for a total dose of 5mg/kg. GVHD prophylaxis for MMUD consisted of CsA or tacrolimus, started on day -3 or day -1 in all patients according to local practice with either mycophenolate mofetil (n = 25, 89%) or methotrexate 15 mg/m² at day +1, 10 mg/m² at day+3 and day +6 (n = 3, 11%). GVHD prophylaxis consisted of CsA and mycophenolate mofetil starting on day -3 for all CB patients.

Median age at allo-SCT was 28 (range: 16-68). The median number of treatments before allo-SCT was 4 (range: 3-6). An ASCT had been performed prior to allo-SCT in 77% (n = 26), 100% (n = 28%) and 100% (n = 37) of cases in the HAPLO, MMUD and CB group, respectively. Disease status at allo-SCT per Cheson 1999 criteria was complete remission in 51% (n = 51), partial response in 32% (n = 31), stable or progressive disease in 11% (n = 11), while no data was available in 5% (n = 5). Positron emission tomography was positive at allo-SCT in 45% (n = 44), negative in 43% (n = 43) while no data was available in 12% (n = 12). Median follow-up was 31 months (range: 3 - 79). In univariate analysis we observed a significantly higher probability of GRFS in patients who received a HAPLO allo-SCT (52% versus 22% and 31% at three years in the HAPLO, CB and MMUD groups, respectively, p = 0.02, Logrank test). Higher GRFS was also observed in patients receiving a NMA conditioning compared to a RIC (50% versus 27% at three years, p = 0.009, Logrank test). We did not observe significant differences in OS, EFS or NRM, according to donor type. Disease status at allo-SCT significantly impacted OS (p<0.001, Logrank test), CIR (p = 0.02, Gray's test), EFS (p < 0.001, Logrank test) and GRFS (p = 0.002, Logrank test). After adjustment for significant covariates, MMUD and CB remained associated with significantly lower GRFS (HR = 2.02, p = 0.03 and HR = 2.43, p = 0.009, respectively, Cox model) compared to HAPLO donors.

In conclusion, significantly higher GRFS was observed in HL patients receiving a haploidentical NMA/RIC T-cell replete allo-SCT with PT CY. How this separation of the GVH from the graft-versus-lymphoma effect occurs requires further investigations. Haploidentical NMA/RIC T-cell replete allo-SCT with PT CY for advanced HL should now be compared with NMA/RIC allo-SCT from related and unrelated HLA-compatible donors, ideally within prospective trials.