Are Alternative Donors Really Still “Alternative?”

HLA-matched sibling donors (MSDs) have long been the gold standard for allogeneic blood or marrow transplantation (alloBMT), but only about 35% of patients have an HLA-matched sibling [1]. The use of HLA-matched unrelated donors (MUDs; and their accompanying donor registries) widened the donor pool, but alloBMT was still unavailable to many individuals, particularly those in need of rapid transplantation for advanced diseases or those belonging to many minority groups. Other alternative donor transplantation approaches, including umbilical cord blood transplantation (UCBT) and modern approaches to HLA-haploidentical (haplo) alloBMT, have made donors available to nearly all patients who might require alloBMT [2–5]. However, it has been unclear whether any of these approaches can produce outcomes truly similar to those seen with MSD alloBMT. The report by Raiola et al. [6] in this current issue of Biology of Blood and Marrow Transplantation, detailing the San Martino Hospital’s collective institutional experience from 2006 to 2012 of MSD and four alternative donor approaches to alloBMT, suggests there now may be clinical equipoise as to the optimal donor.

Several large registry studies have assessed comparative outcomes between MUD and MSD alloBMT. Studies of patients transplanted from the late 1980s to early 2000s suggested that transplantation using MUDs was associated with inferior outcomes for graft-versus-host disease (GVHD), nonrelapse mortality (NRM), disease-free survival (DFS), and overall survival (OS) compared with those seen with MSD alloBMT [7–9]. Furthermore, despite the theoretical potential for augmented antitumor activity that might accompany minor antigen HLA-mismatching, relapse was not lower after MUD allografting [7–9]. Fortunately, outcomes for MUD alloBMT have improved over time likely as a result of, among other factors, improved HLA-typing [10], better supportive care, and the use of antithymocyte globulin [11]. Outcomes for MUD allografting now appear comparable with those seen with MSD alloBMT, with the exception of higher incidences of GVHD with MUDs [12–14].

Similarly, several retrospective studies have compared UCBT with MSD or MUD alloBMT. HLA-mismatched UCBT has been consistently associated with delayed engraftment but similar DFS and OS when compared with MUD and/or HLA-mismatched unrelated donor alloBMT [15–20]. Effects on GVHD, NRM, and relapse have been inconsistent between studies, but most suggested that UCBT was associated with higher NRM but lower GVHD and relapse. In comparisons of UCBT with MSD alloBMT, one study showed similar survival outcomes for individuals ≥ 10 years old treated with high-dose total body irradiation [16], whereas another study in acute myelogenous leukemia patients ≥50 years old showed inferior DFS with UCBT that became similar to MSD and MUD alloBMT after adjustment in multivariate analysis [20].

Early studies of HLA-mismatched related donors were met with unacceptably high rates of severe acute GVHD and graft failure, thus resulting in comparatively poor survival [21]. However, modern approaches to haplo alloBMT, including T cell–depleted “mega-dose” CD34⁺ allografts [3], granulocyte colony-stimulating factor–modulated allografts followed by aggressive pharmacologic prophylaxis [4], and post-transplantation cyclophosphamide (PTCy) [5,22,23], have successfully overcome the challenges encountered in these initial studies. When compared with UCBT in parallel phase II studies using reduced-intensity conditioning, haplo alloBMT with PTCy was associated with improved engraftment, lower severe acute and chronic GVHD, and lower NRM [2]. However, relapse was higher after haplo alloBMT with PTCy, leading to similar progression-free survival and OS. In retrospective comparisons with MSD or MUD allografting, haplo alloBMT with PTCy was associated with similar rates of acute GVHD, NRM, relapse, DFS, and OS as well as lower chronic GVHD in 1 study of a heterogeneous group of hematologic malignancies [24], whereas a study of Hodgkin lymphoma patients showed lower rates of chronic GVHD, NRM, and relapse after haplo alloBMT [25].

The report by Raiola et al. [6] is unique in providing a comprehensive evaluation of 5 different donor types used at the same institution over the same time period (2006 to 2012). All 5 groups received cyclosporine-based GVHD prophylaxis, although antithymocyte globulin also was used for MUD, HLA-mismatched unrelated donor, and intraosseous UCBT, and PTCy was used for haplo alloBMT. The algorithm used by the San Martino Hospital actually provided haplo alloBMT as a safety net available only to those patients for whom other donor types were not available within a time period appropriate for their disease state. Therefore, patients receiving haplo alloBMT were the most likely to have advanced phase disease at alloBMT. Even so, DFS and OS were similar for all 5 groups, with haplo alloBMT patients actually having the highest DFS and OS rates. These favorable outcomes in haplo alloBMT were due to this group having the lowest rates of NRM, grades II to IV acute GVHD, and chronic GVHD as well as relapse rates comparable with other donor types. Notably, immune reconstitution in haplo alloBMT patients was second only to MSD alloBMT, and haplo alloBMT patients had the highest rate of cyclosporine discontinuation by 1 year. Although survival outcomes were statistically similar between UCBT and other groups, the use of UCB grafts compared with MSDs was a negative predictor of OS in multivariate analysis. Overall, the authors concluded that alloBMT using haplo or MUD donors yields comparable survival with MSDs, consistent with previously published results [23,24].

Although these analyses may be confounded, as all retrospective analyses are, with differences in the transplantation approaches and patient characteristics within each treatment cohort, these results question whether MSD alloBMT is really still superior to other donor alloBMT approaches. In fact, the San Martino Hospital group uses haplo donors as its preferred approach to treating active/relapsed acute leukemia [26]. Nevertheless, a major unresolved question is how much of the similarity in outcomes seen in this study are related to other elements of the transplantation platform. Relapse seen in PTCy-treated patients in this study is lower than that reported with reduced-intensity conditioning haplo alloBMT with PTCy [2,22,23], questioning whether relapse is really higher after PTCy and whether more intensive conditioning may overcome any potential for higher relapse. Furthermore, isolating the effect of haplo donors in the transplantation platform is essential, and thus a direct comparison of haplo bone marrow transplantation using PTCy with MSD and MUD bone marrow transplantation using PTCy [27,28] would be illuminating. Even so, the results by Raiola et al. suggest that alternative donor strategies currently may provide outcomes similar to those seen using MSDs and the term “alternative” may no longer be applicable.