Allogeneic hematopoietic stem cell transplantation (HSCT) is currently the only curative therapy for patients with primary myelofibrosis (PMF), myelofibrosis secondary to polycythemia vera (PV-MF), or essential thrombocythemia (ET-MF). The effect of transplant relies both on the initial activity of the preparative regimen in reducing disease burden in the marrow and extramedullary sites, and on an immunological effect mediated by T-lymphocytes in the graft against clonal hematopoietic cells. This graft-versus-tumor effect was initially demonstrated by a 5-year survival of 54% for patients receiving an unmanipulated human leukocyte antigen (HLA)-matched related graft compared to 26% for those receiving a T-cell-depleted graft or a graft coming from an alternative donor. Complete normalization of marrow fibrosis, restoration of normal hematopoiesis, and reduction of splenomegaly to within normal limits usually occur over 12–24 months following transplantation.

Given the marked heterogeneity in the clinical course of myelofibrosis and possible risks associated with transplantation, it is vital to have a reliable prognostic system to select patients who will benefit most. All the transplant studies reported to date utilized the Lille scoring system based on anemia, leukopenia, or leukocytosis, classifying patients as low, intermediate, or high risk.

An International Prognostic Scoring System (IPSS) was recently proposed and encompassed five independent risk factors at diagnosis: age >65 years, WBC >25 × 10⁹/L, Hb <10 g/dl, peripheral blood blasts >1%, and constitutional symptoms. A stratification of patients at any time point of the disease was obtained after finding that the development of anemia had double the adverse impact on survival and assigning it a score of 2 in the dynamic IPSS (DIPSS). Unfavorable cytogenetic abnormalities were shown to significantly impact posttransplant survival as well as risk of leukemic transformation in myelofibrosis. DIPSS was modified (DIPSS Plus) to include red cell transfusion dependency, platelet count <100 × 10⁹/L, and unfavorable karyotype. Based on the presence of 0 to 4 or more factors, patients at low-, intermediate-1, intermediate-2, and high-risk disease have a median survival of 15.4, 6.5, 2.9, and 1.3 years, respectively. Recently, a mortality rate >80% at 2 years was detected in patients with monosomal karyotype, inv(3) and i(17q) abnormalities, or any two of the following factors: peripheral blast percentage >9%, white blood count 40 × 10⁹/l, or other unfavorable karyotype.

The DIPSS was retrospectively applied to a cohort of transplant patients in two separate analyses with similar results, confirming the prognostic value of this system in a transplant setting.

Leukemic transformation (LT) is often a catastrophic event in myelofibrosis, and chances of success with chemotherapy are dismal, with a median survival of 2.7 months. A scoring system to predict LT, including high-risk karyotype, peripheral blasts >2%, and platelets <50 × 10⁹/L, has been proposed. Transplant at the time of LT has poor outcome; however, in a small series of 13 patients transplanted in the blast phase, 49% patients were alive after a median follow-up of 31 months.

Progress in HSCT over the past 2 decades was mostly related to the development of novel hematopoietic cell transplantation (HCT) preparative regimens at reduced intensity, improved antimicrobial prophylaxis, and improved graft-versus-host disease (GVHD) prophylaxis. A number of nondisease factors such as age, donor type, and comorbidities impact outcome following HSCT in myelofibrosis.

In initial studies with myeloablative conditioning (MAC) regimens, patient age at the time of transplantation represented an important prognostic factor, with a 5-year survival of only 14% in PMF patients older than 45 years compared with 62% in younger patients. Retrospective studies both from the Nordic and the Australia–New Zealand cooperative groups confirmed the adverse prognostic impact of age on the outcome of allogeneic stem cell transplantation (allo-SCT). Transplant-related mortality significantly decreased upon introduction of reduced-intensity conditioning (RIC). However, multiple factors have been suggested as independent prognostic factors for toxicity in RIC transplants. In contrast with previous reports, Alchalby et al. (2010) initially showed that patients with wild-type JAK2V617F had an inferior survival after allogeneic HCT. In a following study from the same group, JAK2 status, constitutional symptoms, and age >57 years independently predicted poor survival. The recently discovered calreticulin (CALR) mutation seen in 30% of MF cases has demonstrated predictive potential in transplant outcomes. Patients with mutated CALR had a better OS after ASCT than with wild-type CALR (4-year OS 82 vs 56%, respectively, P = 0.043). More specifically, patients with mutated CALR had the best prognosis, patients with JAK2 or MPL mutations had an intermediate prognosis and triple negative patients had the worst prognosis. Another single-center study demonstrated that spleen size >22 cm, transfusion history >20 units, and any donor other than a matched sibling donor (MSD) (e.g., a matched or mismatched unrelated donor, or a mismatched related donor) predicted an adverse outcome. Analysis of the French stem cell transplantation registry showed that factors favorably affecting engraftment were splenectomy before HCT, a human leukocyte antigen (HLA)-matched sibling donor, peripheral stem cell use as a source of stem cells, and the absence of pretransplant thrombocytopenia. In the GITMO experience, a longer interval between diagnosis and transplant negatively impacted survival after HSCT.

Several studies, including a large prospective trial, have reported similar outcomes in patients undergoing matched related (MRD) and matched unrelated donor (MUD) transplants, while outcomes of mismatched donors were significantly inferior. Contrary to this, using myelofibrosis retrospective data from the Center for International Blood and Marrow Transplant Research, Ballen et al. (2010) showed a 1-year nonrelapse mortality (NRM) rate of 27% for an HLA-identical HCT and 43% for an HLA-matched unrelated HCT. GITMO data also revealed a reduction in NRM associated with the choice of a matched sibling donor. The SFGM-TC registry, including patients with myelofibrosis transplanted between 1997 and 2008, showed engraftment probability and overall survival (OS) to be significantly inferior in nonsibling donors compared to matched sibling donors. An Italian study also indicated that having a donor other than a matched sibling was an independent poor prognostic factor. The ambivalent results from these different studies could be attributable to small sample sizes, retrospective design, patient heterogeneity, and the different chemotherapy regimens utilized. In a prospective study of RIC-HSCT in 66 patients, with a median follow-up of 24 months, 78% in the related group are alive compared to 44% in the unrelated group at 12 month follow-up. Possible interpretation of these different results could be the conditioning regimen utilized, the combination of older age and more advanced disease in the unrelated group, or the degree of HLA compatibility. At this time, for patients <65 years of age, in good overall condition, and with intermediate-2 or high-risk disease, a transplant from an MRD or MUD is indicated due to their poor prognosis. In a small study of 14 patients with marrow fibrosis secondary to different disorders, including PMF, cord blood was shown to be a feasible alternative graft source but only in the setting of a clinical trial at this time.

Significant splenomegaly at the time of transplant may adversely impact time to engraftment, so the question of whether splenectomy should be offered to these patients has been addressed. A previous analysis at Mayo Clinic showed a 27.7% risk of perioperative complications and a 6.7% mortality rate for splenectomy in myelofibrosis patients. Conflicting results have been reported on the survival benefit of pretransplant splenectomy. The European Group for Blood and Marrow Transplantation (EBMT) study reported a threefold increased risk of relapse after splenectomy, although it is argued that this may have been a reflection of more advanced disease. In contrast, an Italian study reported a reduced risk of relapse in patients who were splenectomized before HSCT. Review of the data from the Fred Hutchinson Cancer Center showed that after adjustment for DIPSS score and the Hematopoietic Cell Transplant Comorbidity Index (HCT-CI), patients who had a splenectomy were at lower risk for mortality compared with patients who had not (HR = 0.44). Retrospective data from the Center for International Blood and Bone Marrow Transplant Research (CIBMTR) and two other studies failed to show an effect of splenectomy on disease-free survival (DFS) or OS. Postulated setbacks of splenectomy in the setting of HSCT include a worsened severity of GVHD due to altered immunomodulation and an excess risk of acute myeloid leukemia transformation.

In the absence of any intervention for splenomegaly, successful engraftment can still occur. In a study of 10 patients with extensive splenomegaly, a progressive reduction of splenomegaly within 12 months post RIC transplant was demonstrated, and it paralleled the reduction of marrow fibrosis. Given the absence of strong favorable evidence and inherent operating risks, at this time routine splenectomy is not advocated prior to transplant.