Allogeneic stem cell transplantation (allo-SCT) remains the only curative therapy for congenital, severe haemoglobinopathies. Nevertheless, post-transplant morbidity and mortality are of concern. Toxic mortality is more evident with increasing patient age, with the use of peripheral stem cells, with non-HLA-identical donors, and above all with iron overload, which in turn depends on adequate iron chelation. The subdivision of patients according to Pesaro risk classes remains significant in identifying patients with an increased risk of transplant mortality ^[1][2][1,2].

2. General Results

An extensive analysis of transfusion-dependent thalassemia (TDT) patients receiving allo-SCT in Europe between 2000 and 2010 was reported by the EBMT. In this retrospective study, including 1493 patients, most were paediatric (91%, median age 6.6 years), and only 8% were aged more than 18 years (median age 22.9 years). Donors were HLA identical siblings (HLAid sib) in 71%, while only 14.1% were matched unrelated (MUD). The stem cell source was bone marrow in most cases (BM) (67%), peripheral blood stem cells (PBSCs) in 20%, and cord blood (CB) in 3.9%. The type of conditioning regimens was not reported. Overall, the 3-year OS and EFS were 88% and 81%, respectively. However, survival was improved in patients with defined characteristics such as age less than 14 years, donor HLAid sib, and BM. The incidence of severe (grade 3–4) acute graft vs. host disease (aGVHD) was 9%, and the incidence of extensive chronic GVHD (cGVHD) was 6%. As expected, the incidences of aGVHD and cGVHD were lower when BM was used as the stem cell source and HLAid sib was used as the donor [3]. As reported in the recent Cochrane Review [4], randomized or near-randomized studies are lacking, and thus far, most clinical data and suggestions come from retrospective, frequently registry-based studies. Thus, selection bias should be considered when we are discussing allo-SCT in TDT. Factors such as HLA-identical family donors, myeloablative conditioning regimens, bone marrow as a stem cell source, and early recipient age (see below) are considered of the utmost importance because of their impact on clinical outcomes. The first point to address is recipient age. Indeed, it is clear that the main outcome endpoints are jeopardized when the age of the recipient is increased. However, it is less clear what the age cut-off is that can help to separate patients with high vs. low risk. In the Cochrane Review [4], the cut-off age retrieved by the studies analysed was 14 years. The EBMT registry reported on TDT patients allografted between 2000 and 2010 (n = 1493) (3). Most patients were aged less than 18 years with a median age of 6.6 years, while only 9% were adults with a median age of 22.9 years. Clearly, overall survival (OS) and event-free survival (EFS) were better in patients aged less than 14 years. Indeed, in the group that was greater than 14-years-old, the OS was 82%, and the EFS was 74% (compared to more than 90% and 83%, respectively, in the younger groups). In another registry-based study from the CIBMTR, including 1110 patients, the impact of age on the clinical outcome was analysed. The median age of the whole population was 6 years (<1–25), but only 3% of them were aged between 16 and 25 years. EFS and OS were higher in the cohort aged less than 6 years, the intermediate group between 7–15 years, and the lower aged cohort aged 16–25 years (EFS 86% vs. 80% vs. 63%, and OS 90% vs. 84% vs. 63%, respectively) [5]. Many hypotheses can be proposed to explain the strong impact of age on outcomes. The most immediate could be that younger patients have a shorter duration of disease, with a lower exposure to transfusions and iron overload. This means that organ integrity can be preserved or is less affected, leading to better tolerance of intensive conditioning regimens. A second point related to fewer transfusions could be the low risk of allogeneic immunization. However, other factors, such as conditioning regimens, GVHD prophylaxis, ethnicity, and donor age, can contribute to or interact with recipient age. The second point to discuss is the stem cell source. Bone marrow is considered the ideal stem cell source in non-malignant disease due to the reduced risk of both acute and chronic graft versus host disease, which are considered “not suitable” because of the nonneoplastic nature of the disease. In the EBMT study [3], BM was the most frequently used stem cell source (67.8%) compared to peripheral blood (20.3%) and cord blood (3.9%). The authors observed a significant negative impact of PBSCs and CB on EFS and OS compared to BM (76% vs. 85% vs. 82% and 81% vs. 93% vs. 90%, respectively). The lower survival with PBSCs was attributed to a higher incidence of severe aGVHD (8% vs. 3%). On the other hand, in the CIBMTR study [4], 61% of patients received PBSCs and 29% received BM. In their multivariate analysis on factors influencing survival, aGVHD and cGVHD, PBSCs did not enhance the risk of developing GVHD or reduce survival. When BM cannot be collected, regardless of the reason, and PBSCs are the stem cell source available for a definite patient, some parameters of transplantation should probably be modified. For example, GVHD prophylaxis should be reinforced with in vivo T-cell depletion or with ex vivo selective T-cell depletion, such as an αβ-depletion platform. This could reduce the risk of severe GVHD. On the other hand, the use of PBSCs could lower the risk of graft failure because of higher CD34+ cell count. Another relevant variable on the outcome of TDT is the conditioning regimen. The condition regimen plays a different role in the outcome: it is useful because TDT patients have hyperexpanded bone marrow, often with islets of extramedullary haematopoiesis, and ablation of this hyperplastic tissue is necessary to favour engraftment. Furthermore, myeloablative doses can help to eliminate immune-activated B cells/plasma cells and T cells induced by frequent transfusions. The net effect of MAC is an increased probability of avoiding graft failure, one of the most important causes of death. In the first reports from Pesaro, a myeloablative conditioning regimen based on oral busulfan and cyclophosphamide (osBUCY) was simply taken from what was used at that time as a total body irradiation-free conditioning regimen for malignant diseases ^[6][9]. This regimen allowed a low rate of graft failure and good survival ^[1][2][1,2].  In recent years, the definition of MAC based on the degree of haematopoietic toxicity ^[7][14] has been challenged by the introduction of an intravenous formulation of BU (ivBU), the adoption of a PK-based dosage of BU, and the development of more tolerated alkylating agents such as treosulfan (TREO). Overall, all of these changes define MAC with reduced toxicity (RTC). The use of ivBU has significantly improved its therapeutic index, reducing, but not eliminating, pharmacokinetic variability ^[8][15]. The use of PK-guided BU dosage further improved the clinical outcome in both malignant and non-malignant diseases in adult and paediatric settings. The data relative to the impact of PK-guided BU have been reviewed by van der Stoep et al., confirming that, at least in children, a correct BU dosage correlates with clinical outcomes ^[9][16]. The replacement of BU with another alkylating agent, such as TREO, has gained increasing interest. TREO is characterized by renal clearance as an unmodified product, but in general, clearance is variable in the paediatric setting. The results obtained with conditioning regimens including TREO were interesting because, at myeloablative doses, the tolerance was good with a low rate of graft failure and toxic deaths. Two seminal papers reported on these results. Bernardo et al. treated 60 TDT patients with a TREO-based conditioning regimen (associated with fludarabine and thiotepa). The median age was 7 years (1–37), and 70% of donors were unrelated. In vivo T-cell depletion by ATG was used only with UD. Few patients were classified as class 3 Pesaro (7%). The cumulative incidence (CI) of graft failure was 9%, the CI of grade 2–4 acute GVHD was 14%, and the CI of TRM was 7%. No cases of VOD were observed. The 5 y OS and TFS were 93% and 84%, respectively ^[10][17]. Every time the conditioning regimen is modified, mainly to reduce the toxicity and mortality, the engraftment rate can be affected, leading to a mixed donor chimerism (MC). The association between MC levels at day +28 (high when residual host cells were more than 25%) and the risk to reject donor marrow after MAC, BM, and HLAid sibling is well known ^[11][12][24,25]. Using less-intensive conditioning regimens, the MC rate could be different and theoretically higher. When treosulfan-based regimens were utilized before transplantation, the MC rate ranged from 8% to 50% ^[13][14][15][18,19,21]. On the other hand, in the paper from Shenoy et al. ^[16][26], MC was absent, while in another paper ^[17][27], the rate was 28%, but all MC patients reversed their pathological phenotype.  Although a HLAid sib is considered the ideal donor for these patients, the probability of finding such a donor is at best 25%. Thus, alternative donors such as unrelated or nonidentical familial donors have been sought. Of course, with alternative donors, the risk of graft failure and TRM, with a reduction in survival, can be high. In a prospective trial from China, 82 patients with TDT (median age 6 years, range 0.6–12) received an adapted MAC consisting of PK-guided BU, Cy, fludarabine and thiotepa before MSD or MUD stem cell infusion. In the MUD group (n = 52), all patients were infused with PBSCs. The overall survival was similar based on donor type (3-year OS and EFS 90% vs. 92% and 83% vs. 90%, respectively). Graft rejection was observed in three patients (CI 3.7%), and TRM was 8.5%. VOD was observed in 6%. Of note in this study, the authors classified patients using two parameters: age (<4 vs. >8 years, ferritin levels <3000 vs. >5000) and hepatomegaly (<2.5 under costal margin vs. >4 cm) due to the impossibility of performing liver biopsy to define Pesaro class risk ^[18][23]. GVHD prophylaxis with post-transplantation cyclophosphamide (PTCY) with T-cell-replete stem cell infusion from haploidentical donors quickly gained popularity in the field compared to previous systems based on ex vivo T-cell depletion. Mainly used in patients with haematological malignancies, the Baltimore group tested this modality in patients with haemoglobinopathies. Bolanos-Maede et al. first reported a prospective trial consisting of a non-myeloablative conditioning regimen and PTCY in 17 patients (5 TDT). Based on the data from a previous trial on sickle cell disease ^[19][31], the protocol was modified with a higher dose of total body irradiation (from 200 to 400 rads) to try to lower the incidence of graft failure (reported to be 50%). The median age was 16 years (6–31). The treatment was extremely well-tolerated, and only one patient developed graft failure, five had grade 2–3 aGVHD (incidence 29%), and only one had moderate cGVHD. No treatment-related mortality was reported, and all TDT patients became transfusion independent ^[20][32]. 

3. Conclusions

Allogeneic stem cell transplantation, while waiting for gene therapy, remains the only curative option for TDT. Clearly, the scenarios are changing because few patients can receive what is considered the gold standard transplantation consisting of a young patient, HLAid sibling, myeloablative condition regimen, and BM as the stem cell source. On the other hand, changing one or more of these variables must be linked to adapting the others. For example, the same conditioning regimen cannot be used if the recipient’s age is higher than 16 years or more, if the donor is unrelated or haploidentical, or if iron chelation is not optimal due to organ damage. Furthermore, specific prognostic scores have been recently developed for both transfusion-dependent and transfusion-independent thalassemia (31). In this analysis, seven variables were included; namely, heart and liver disease, ALT levels > 42, diabetes, sepsis, haemoglobin levels < 9 g/L, and serum ferritin ≥ 1850 ng/mL; and three categories were identified: low, intermediate, and high risk. The survival rate was significantly different among the three categories (59.8% vs. 26.4% vs. 13.8%). The development of risk scores, such as the Thalassemia International Prognostic Scoring System (TIPSS) ^[21][35], could help to counsel patients and families and to plan treatment, such as allo-SCT, based on objective information, even in patients with no TDT.