Multiple myeloma is a largely incurable disease that has benefited from the advent of novel agents over the last decade. Allogeneic hematopoietic cell transplantation (allo-HCT) remains the only potentially curative treatment; however, its use is controversial given its high morbidity and mortality. Alloreactive immune effector cells originating from a myeloma-free graft exert graft-versus-myeloma (GVM) effects leading to long-term control of disease (1, 2). Evidence for this GVM effect is provided by reports of sustained molecular remissions achieved by donor lymphocyte infusion (DLI) after allo-HCT (3). Currently, high-dose melphalan followed by autologous hematopoietic cell transplantation (auto-HCT) remains the standard-of-care for suitable patients with MM, as demonstrated by the recent report of the IFM 2009 trial (4). Historically, there have been significant efforts to modulate different elements of allo-HSCT to improve outcomes while minimizing regimen-related toxicity and graft-versus-host disease (GVHD). In the following sections, we will discuss different components of allo-HCT in MM and how they may guide our optimal choice of therapy for MM. There include: patient selection, graft source, conditioning regimen, posttransplant maintenance therapy, and the role of DLI. This review examines these crucial steps of allo-HSCT separately in the context of emerging new data.

The risk–benefit ratio of allo-HCT in standard-risk MM is questionable. The BMT CTN 0102 trial randomized standard risk MM patients (defined by Beta-2 microglobluin <4 mg/L, absence of chromosome 13 deletion, and no progression after initial chemotherapy) to tandem auto-HCT (n = 366) or auto-HCT followed by matched related donor (MRD) allo-HCT (n = 156) (5). There was no difference in relapse rate or overall survival (OS) between the two arms with more treatment-related mortality (TRM) in patients who received auto-HCT followed by allo-HCT (4 versus 11%, p-value <0.0001). Similarly, another large study, HOVON-50 trial, did not support the broad application of allo-HCT in MM (6). In this trial, the outcomes of 122 patients who had an HLA-identical sibling donor (donor group) and underwent allo-HCT after a planned auto-HCT were compared prospectively to 138 patients without a donor who underwent maintenance therapy or second auto-HCT after the first auto-HCT. PFS was significantly prolonged in patients who proceeded to allo-HCT compared to maintenance or second auto-HCT. However, after a median follow-up of 77 months complete remission (CR) rate as well as OS was similar between the two groups with higher 6-year TRM in the allo-HCT group (16 versus 3%, p-value <0.001). In the European Group for Blood and Marrow Transplantation Non–Myeloablative Allogeneic Stem Cell Transplantation in multiple myeloma (EBMT-NMAM2000 study), patients with an HLA-identical sibling (n = 108) receiving auto-HCT/RIC allo-HCT were compared to 249 patients that received tandem auto-HCT. This study demonstrated improved PFS and OS in the auto-HCT/RICallo-HCT arm (OS 49 versus 36% at 96 months, respectively, p = 0.030), at the expense of higher TRM (13 and 3%, respectively). Interestingly, the RIC/allo-HCT approach seemed to overcome the poor prognostic implication of del(13) (7). However, none of the patients received any novel agents during induction or maintenance and FISH data was not available for all patients for risk stratification.

Crucial to interpreting these studies is the fact that novel agents and auto-HCT have enhanced OS of MM patients from a median of 3–5 years in the 1990s to over 10 years currently (8). Therefore, it is harder to justify allo-HCT for standard risk disease in the era of novel agents (9, 10).

Myeloablative conditioning regimens are generally associated with higher TRM rates mostly due to infection, GVHD, and regimen-related toxicities (35). The most common conditioning regimens incorporate Cyclophosphamide (Cy) and total body irradiation (TBI); Busulfan (Bu) and Cy; Bu/Cy/TBI and Melphalan (Mel) with TBI (Table 1). In the S9321 clinical trial, patients in the allo-HCT arm received the myeloablative conditioning regimen of Melphalan plus TBI. TRM was 53% leading to early closure of the allo-HCT arm (36). In another trial, the Haemato-Oncology Foundation for Adults in the Netherlands (HOVON-24) study, TRM was 30% with dismal PFS and OS for patients who underwent partially T-cell-depleted myeloablative allo-HCT with Cy/TBI conditioning regimen (37). Bensinger et al. compared the outcome of 144 patients who underwent allo-HCT utilizing high-dose myeloablative regimens with 134 patient who received RIC and non-myeloablative (NMA) regimens. Intensity of the conditioning regimen was predictive of higher acute GVHD and TRM rates, earlier relapse, and inferior OS (38). Prior auto-HCT was the main predictor of high TRM after allo-HCT across a variety of conditioning regimen intensities. Other groups reported a similar correlation between prior auto-HCT as well as time between auto-HCT and allo-HCT as another poor prognostic factor and predictor of higher TRM (39). These results suggest that prior exposure to melphalan adds to the toxicity of allo-HCT.

Higher TRM rates associated with myeloablative regimens lead to alternatively exploring NMA and RIC regimens. The CIBMTR database shows a significant shift from intensive regimens from the 1990s to the early 2000s (42). Most of these regimens contain Mel 100–140 mg/m² plus other agents (43). This approach leads to expansion of allo-HCT eligibility criteria by including older patients. Interestingly, higher relapse rates often negated the lower TRM benefit of RIC leading to similar OS.

The strategy of cytoreduction with auto-HCT before RIC allo-HCT has been investigated extensively. Several large prospective randomized trials compared auto-HCT followed by RICallo-HCT to tandem auto-HCT as consolidation therapy (Table 2). However, there is a high degree of heterogeneity among these trials, including use of different conditioning regimens and use of novel agents. Overall, it seems that patients who underwent allo-HCT after auto-HCT are at a higher risk of TRM. A meta-analysis of six studies compared 1,192 patients who underwent tandem auto-HCT to 630 patients who received auto-HCT followed by RIC allo-HCT (44). Recipients of auto-HCT and RIC allo-HCT had higher TRM and CR rates, without PFS or OS advantage.

Multiple studies showed the benefit of maintenance therapy with lenalidomide or bortezomib after auto-HCT for MM patients (55, 56). In contrast, there are scant data available on maintenance therapy post allo-HCT. Lenalidomide enhances anti-myeloma NK and NK/T-cell response and prolongs PFS and OS when used as maintenance therapy after auto-HCT (55, 57). Lenalidomide augments NK cell-derived anti-myeloma activity while delaying recovery of regulatory T cells posttransplant (58). However, the possible immunostimulatory effect of this drug may lead to rapid onset and higher rates of acute GVHD (59), therefore, complicating its clinical use post allo-HCT. Dosing and timing seems to be crucial for lemalidomide maintenance post allo-HCT affecting NK and T cell activation (58). There is preclinical and clinical studies evidence suggesting that proteasome inhibitors can be safely used post allo-HCT. Moreover, these agents may suppress GVHD while preserving the GVM effect, and this hypothesis is under investigation in the U.S. BMT CTN 1302 trial utilizing the third generation proteasome inhibitor ixazomib as maintenance.

Bortezomib use after allo-HCT may add an anti-myeloma effect (60) without exacerbating GVHD. Kroger et al. administered maintenance bortezomib (Median: 7 months) for 18 MM patients with stable disease after RICallo-HCT (54). The drug was well tolerated with Grade III/IV neuropathy in three patients receiving cyclosporine, while three patients experienced mild aggravation of existing skin GVHD. In another study, 37 MM patients with residual or progressive disease after allo-HCT received bortezomib. Response rate was 73% with two patients experiencing worsening GVHD.

As discussed previously, most allo-HCT for MM now use RIC regimens at the expense of higher relapse rates. RIC use is especially important in heavily pretreated and/or elderly patients who are not candidates for more intensive conditioning regimens. However, MM is considered to be of only intermediate susceptibility to the graft versus tumor effect. Several strategies have been explored to improve remission rates and decrease relapse risk.

Donor lymphocyte infusion given post allo-HCT is believed to augment the GVM effect. In 1996, Tricot et al. (61) reported CR with a single dose of CD3+ cells in a MM patient who had progressed after allo-HCT. This report was the first “proof of principle” for utilizing DLIs to induce a GVM effect. Although DLIs are mostly given in the context of refractory or progressive disease posttransplant (salvage DLI), this modality of adoptive immunotherapy has also been utilized preemptively posttransplantation (prophylactic DLI) (30, 62–70). Prophylactic DLI given at specific time points or in escalated incremental doses during T-cell reconstitution are believed to enhance donor-derived T-cell reconstitution and to maximize GVM (70) (Tables 3 and 4).

Response rates as high as 52% have been reported (72, 73). However, in many studies, patients received additional therapies including chemotherapy before DLI to decrease “tumor burden” and improve chances of response. In other reports, DLIs have been combined with either immunomodulatory drugs or proteasome inhibitors.

Graft-versus-host disease is a significant risk associated with DLI that should be taken into consideration before selecting candidates for this treatment modality. The risk of both acute and chronic GVHD is reported to be as high as 60% in some studies. Development of GVHD has been associated with higher response rates but, this has not been consistent across different reports (73). In addition, neither the T-cell dose nor converting patients from partial to full donor chimerism has been consistently associated with higher anti-myeloma response.

A study of frequent chimerism monitoring in 155 patients who underwent allo-HCT for MM showed that two-thirds of relapsing patients had full donor chimerism. Interestingly, one-third of patients had extramedullary disease despite donor hematopoietic reconstitution (75).

Donor-derived cytotoxic T-lymphocytes directed against myeloma-associated antigens such as Wilms tumor 1 (WT1) or other cancer testis antigens (76, 77) may enable a GVM effect without GVHD. Tyler et al. (77) reported a correlation between disease outcomes and WT1 expression by plasma cells in the bone marrow. Development of WT1 cytotoxic T-lymphocytes post allogeneic T-cell depleted transplant and DLI was associated with a GVM effect in the absence of GVHD.

Allo-HCT highlights the potential of immune-based cell therapies for MM treatment. Improving risk models incorporating more sensitive minimal residual disease may lead to better patient selection and optimal structuring of different elements of the allo-HCT process in the future. In the U.S., the Centers for Medicare and Medicaid Services is considering MM as an indication for allo-HCT in the setting of an appropriate clinical trial applicable to the Medicare age population. The transplant community will be charged with better defining indications and limitations of this approach. Allo-HCT for MM should continue to be offered within a clinical trial designed to improve and understand the role of this therapeutic modality.

Conception and design: EM and ML. Drafting the article: EM and NEJ. Critical revision of the article: ML and NK. Final approval of manuscript and accountable for all aspects of the work: all authors.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.