Beyond transfusion therapy: new therapies in thalassemia including drugs, alternate donor transplant, and gene therapy

Thalassemias are a heterogeneous range of anemias, caused by globin chain imbalance with ineffective erythropoiesis as its hallmark. This therefore requires a range of measures, from simple observation, to intensive transfusion with chelation and to potentially curative therapies. Unfortunately, geographic regions with the highest prevalence are often those with the lowest level of economic development, often lacking in universal health care provision. With an estimated global birth rate of 50 000 to 60 000 per annum and a global prevalence of 288 000 for severe β-thalassemias, for economic reasons alone, prevention is an important component of overall management.

Transfusion combined with chelation has undoubtedly been an effective treatment modality when it has been efficiently delivered. The reliability with which safe blood, iron chelation, and complex monitoring varies not only between countries but even within countries. Since the introduction of iron chelation in the mid- to late 1970s, life expectancy and comorbidities have markedly improved.¹  Deaths from cardiomyopathy have decreased and this improved further after the introductions of oral chelation and cardiac magnetic resonance monitoring. There is a strong cohort effect on survival (predicated by the date of introduction of chelation to the population in question) but patients with transfusion-dependent thalassemia (TDT) should now expect to live for 50 years or more, provided treatment is not sporadic or interrupted.¹  Quality of life (QOL) has also been improved further by the introduction of orally active chelation. For the best results, however, chelation needs to start before irreversible iron-mediated tissue damage has occurred and for patients now alive in their 50s, regular chelation with desferrioxamine 5 to 7 nights a week by subcutaneous slow infusion only became available after 10 years of age, when irreversible tissue damage had already typically occurred, particularly to the endocrine system. Now, with an aging thalassemia population, in centers where chelation has been established for 4 decades or more, liver disease and bone pain are increasing issues. For example, bone pain has been reported in >80% of patients older than age 40 in our clinics at University College London Hospitals and Whittington. It remains to be seen whether cohorts that have been optimally chelated from a young age will experience less pain, osteoporosis, cirrhosis, and carcinomas than current patients >40 years of age.

This success comes at a price and, unfortunately, the challenge of providing blood, chelation therapy, adequate monitoring, and patient support can threaten to overwhelm health care systems,²  even in areas with relatively well-developed economies such as Cyprus and Sardinia. Consequently, effective treatment is often not delivered: for example, as recently as a decade ago, it was estimated that only 12% of children with TDT globally receive adequate transfusion therapy, and <40% of those transfused receive adequate iron chelation. It has been estimated in the United Kingdom that a lifetime of treatment to 50 years of age is about £483 454 ($646 934).³  Costs have increased because of newer chelator regimes and monitoring techniques such as magnetic resonance imaging of heart and liver iron: for example. increasing 32% in the United Kingdom in the past 16 years.³  Indirect costs also need to be added: a large international study showed that adult TDT patients in paid employment lose a mean of 2.1 working days per month and children 2.8 days of school per month, from treatment and monitoring. A further challenge with chronic transfusion and chelation is adherence to treatment.⁴  To minimize adherence issues, long-term care also requires lifelong specialist, multidisciplinary input, often not organizationally available, even in highly developed economic regions. For all these reasons, curative treatments or for treatments that decrease the requirements for blood transfusion have been eagerly sought.

The effectiveness and risk factors of allogeneic transplantation with HLA-matched stem cells from a sibling were established in the 1980s (Table 1). These remain largely unchanged as: class 1, 2, or 3, based on the presence of 1 to 3 risk factors (out of hepatomegaly, liver fibrosis, or poor previous chelation)⁵  with age older than age 14 is an independent risk factor. Increased risk of graft rejection, probably related to repeated transfusions⁵  as well as massive expansion of the erythropoietic marrow required high-intensity conditioning initially using busulfan and cyclophosphamide. There has been gradual improvement in outcome over the past 20 years, resulting from better risk stratification, targeted dose adjustment of intravenous busulfan, modified conditioning regimens, and improvements in supportive care. Today, if hematopoietic stem cell transplantation (HSCT) is performed from an HLA-matched sibling donor before 14 years of age, procedure-related mortality is <10% (Table 1); if performed before 5 years of age, mortality is reduced further to ∼5%.⁶  Risk in adults (aged 18-45) remains higher than children, so that although overall survival is 88%, only 69% to 81% of adults show thalassemia-free survival.⁶  Remaining complications include about a 10% risk of severe acute graft-versus-host disease (GVHD) (grade 3-4) and 2-year risks for limited or extended chronic GVHD of 15% and 6%, respectively. Graft rejection/failure is another important complication, which was reduced by the use of ATG (Table 1).⁷  GVHD may also be reduced by this maneuver.⁷  Health-related QOL (HRQOL) measures are generally good in children surviving HSCT: chronic GVHD was the most significant factor lowering HRQOL. Twenty years after HSCT, better HRQOL outcomes have been reported than in those having conventional noncurative therapy.⁸  The long-term effects of HSCT on fertility are not well described, however, particularly in males and while fertility might occasionally be preserved after HSCT, this is a rare event. Overall consensus statements now recommend that patients with a matched sibling donor should be offered HSCT at an early age.⁹ 

Modifications to conditioning regimes have improved outcomes in high-risk class 3 patients (Table 1). Intensive pretransplant transfusion, intensive chelation, and hydroxyurea combined with improved myeloablative conditioning, improved outcomes (Table 1, Sodani, 2004). Sinusoidal obstruction syndrome (veno-occlusive disease [VOD]) is another important complication, particularly in those with significant preexisting liver dysfunction and conditioning with treosulfan rather than busulfan may also improve outcomes. Another approach is to use a “reduced intensity” regimen in the knowledge that partial mixed chimerism of donor- and recipient-derived stem cells can produce adequate and stable hematopoiesis (Table 1, Andreani, 2008).

Because HLA-matched related donors are available for only about one-quarter of patients (except in countries with larger families or communities with consanguineous marriages), a goal has been to extend the donor pool. In principle, this can be achieved by haploidentical or by matched unrelated donor (MUD) transplantation. MUD potentially extends the donor pool close to 50% of patients of Caucasian origin. High-resolution molecular typing for both HLA class I and II loci has been important. Improved conditioning regimes including ATG and methotrexate for GVHD prophylaxis also improved outcome (Table 1). Because of the greater risks with MUD, expert panel recommendations are more cautious than for sibling allografts,⁹  and suggest that if a well-matched MUD is available, allogeneic HSCT with high-resolution molecular typing for both HLA class I and II loci is a suitable option for a child with life-long control of iron overload and in the absence of iron-related tissue complications.

Cord blood (CB) transplant) has been explored as a way to decrease acute and chronic GVHD. However, for matched sibling transplantations, thalassemia-free survival did not differ between cord blood transplant and blood marrow transplant.¹⁰  Unrelated CB transplantation has also been explored for patients lacking matched family or matched unrelated donors. Unfortunately, high rates of graft failure and delayed hematopoietic recovery have been seen, mainly because of inadequate cell dose (Ruggeri, Table 1). This limitation might be mitigated by using ≥ 1 CB donor unit or by giving CB together with T cell–depleted HLA-haploidentical CD34 with cells. Current recommendations for CB transplant suggest using units containing at least 3.5 × 10⁷ total nucleated cell/kg recipient body weight before cryopreservation, and with less than 2 HLA disparities.

HLA-haploidentical family donors may be another way to extend the donor pool. Initial reports, using T-depleted CD34 donation, showed 27% graft rejection a low thalassemia-free survival (Sodani, 2004, Table 1). There have been recent improvements in graft rejection and overall survival followed using T-cell receptor αβ⁺ (TCRαβ⁺)/CD19⁺–depleted grafts (Gaziev 2018, Table 1). However, there was delayed immune reconstitution with associated morbidity and mortality. Furthermore posttransplant lymphoproliferative disorders such as autoimmune hemolytic anemia or thrombocytopenia have been seen. Procedure-related mortality, as well as morbidities, therefore remain greater than in matched sibling transplantation⁹ ; as a result, HLA-haploidentical family donation should currently be confined to clinical trials.¹¹ 

Although costs vary between countries, the estimated costs of an allogeneic HSCT from a sibling are almost invariably lower than of conventional therapy with lifelong blood transfusion and iron chelation (CT).^3,12  Cost analysis makes assumptions about life expectancy with transfusion plus chelation difficult and vary considerably, depending on health care resources and geographic organizational issues. In the United Kingdom, the approximate cost of an allogeneic blood marrow transplant¹³  is equivalent to about 8.7 years of CT for a patient surviving 50 years.³  The advantages of transplantation over CT will be greater when the latter cannot be provided to the necessary standards for financial or organizational reasons. Comparison of outcomes with either HSCT or CT, in Sardinia, where both are available to a high standard, showed that overall survival of 83% in transplanted patients was similar to CT with overall survival of 85%.¹⁴  Outside matched sibling allograft, risks are still higher and there is less consensus about the optimal risk/benefit ratio. Long-term follow-up at least yearly is recommended for after HSCT.

Gene therapy thalassemia has already become a clinical reality using lentiviral (LV) vectors to replace key elements of the β globin gene¹⁵  and other approaches are in advanced preclinical stages. The use of autologous modified stem cells for engraftment has the obvious advantage of providing HSCT for all patients. Further theoretical advantages are the lower procedure-related mortality and the lack of need for immune suppression, as well as no risk of acute or chronic GVHD or immune-mediated graft rejection. Some transplant-related issues will remain while myeloablative regimes are used, however; these include VOD and long-term effects on fertility. Overall, the clinical benefit of gene therapy must outweigh the risks of myeloablative conditioning regimens; also, the long-term safety will need to be demonstrated with respect to the oncogenic potential of gene therapy itself (see the following section).

In principle, correction of the abnormal or missing gene can be performed by gene addition, such as using LV vectors or by precisely correcting single-point mutations using designer nucleases. Because there are >200 β-thalassemia mutations, a strategy that can be applied irrespective of the individual β mutation is preferable. Early studies with murine thalassemia models used recombinant γ-retroviral vectors, derived from the Moloney leukemia virus, as carriers to introduce the human β-globin gene in HSC. Limitations included low vector titers, low expression of β-globin, and unstable vectors with a tendency to rearrangements. β-globin expression was then increased by incorporation of the core elements of the hypersensitive sites 2, 3, and 4 of the human β-globin locus control region. However vector rearrangements and position effects were still seen. Self-inactivating LV vectors were then evaluated for which overcame some of these issues; it is these that trials are most advanced

Here the intent is to insert a fully functional gene (or part of it) that will result in sufficient correction of globin chain imbalance that is the hallmark of thalassemia. Self-inactivating, LV vectors are now used by several laboratories for clinical trials for β-thalassemia, which have been reviewed in detail recently elsewhere (Table 2).¹⁶  These vectors have many shared features.¹⁷  For gene expression to be adequate, promoters are required, and the insertion site is critical to efficacy and safety. The role of insulators, ostensibly to decrease the risk of insertional mutagenesis is less clear: the first clinically successful LV vector, HPV569, contained 2 insulator elements. However, because of low engraftment in 3 of 4 patients and because of a transient clone in 1 patient, the insulators (cHS4) were omitted from the subsequent vector BB305, when Bluebird Bio (BBB) became involved (Table 2). Insulators are included in the TNS9.3.55 vector used at Memorial Sloan Kettering Cancer Center¹⁷  but not in the vector developed in the Italian GLOBE study (Table 2).¹⁶  The potential for long-term correction of thalassemia, using LV-transduced HSC was first shown in a murine model of thalassemia intermedia¹⁸  and subsequently clinically in a transfusion-dependent adult Eβ thalassemia patient.¹⁹ 

A second general approach is to promote γ chain synthesis in the developing erythroblast to a point where the α/non-α ratio is sufficiently balanced to prevent intramarrow apoptosis in developing red cells. This has been achieved in human hematopoietic cells by editing genes that suppress γ globin synthesis, such as the BCL11A transcription factor, without affecting other hematopoietic lineages. Editing methodologies include designer nucleases such as zinc fingers or the repurposing of bacterial clustered regularly interspaced short palindromic repeats (CRISPR) to allow RNA-guided cleavage of complementary DNA. By using CRISPR/Cas9 and guide RNA in human HSC, a 40% reduction in BCL11A mRNA resulted in corresponding twofold increase in γ-globin transcript ex vivo.²⁰ 

Another way to enhance fetal hemoglobin (HbF) is to disrupt γ−δ intergenic HbF silencers. This has been recently demonstrated in human progenitor cells, using CRISPR/Cas9 to edit a 13.6-kb genomic region encompassing the δ- and β-globin genes.²¹  At least 2 such approaches are being initiated in clinical trials for thalassemia (Table 2). The first uses zinc finger nuclease to boost HbF by disruption of an erythroid specific enhancer of the BCL11A gene (Sangamo, Bioverativ). The second is CTX001, a gene-edited cellular product that uses CRISPR Cas9 technology to target an erythroid-specific enhancer of the BCL11A gene (Table 2), which has been are approved for clinical trials in several countries.²²  The latter demonstrates a high ex vivo editing rate and no identified off-target sites in preclinical models. The safety and specificity of the CRISPR approach to gene editing is currently debated. A key issue is the so called “off target” effects (OTEs). These occur under conditions in which, although cleavage at sites with perfect complementarity to the guide RNA (gRNA) is highly efficient, this can also occur at sites where 1 or more bases are mismatched with the gRNA. In principle, OTE could lead to loss of a key stem cell function or could have oncogenic potential by activating oncogenes or inactivating a tumor suppressor genes. Methods that result in higher sustained levels of cellular Cas9 and gRNA tend to have higher OTE. At high editing efficiencies, double-strand DNA breaks are induced by Cas9 and, although these usually lead to cell death through a P53/TP53-dependent mechanism, cells with mutated P53 may be resistant, risking the emergence of mutated cells.²³  It remains to be seen whether these concerns are real in clinical use and/or whether strategies to minimize OTE such as with high-fidelity Cas9 variants are effective.

Irrespective of the gene modification, transplantation typically involves 4 steps: mobilization of autologous CD34⁺ stem cells, ex vivo correction of the abnormal gene, myeloablative conditioning, and the reinfusion of sufficient corrected stem cells. Autologous CD34⁺ cells can be obtained by direct collection from bone marrow aspiration¹⁶  or from stem cell mobilization.^15,19  Mobilization of CD34⁺ cells has been generally achieved using granulocyte colony stimulating factor with or without plerixafor (Table 2). Myeloablation has been achieved using busulfan monotherapy, for example, with total doses of 12.8 mg/kg used over 4 days in the BBB studies. The procedure has been generally well tolerated with adverse events and frequencies typical of an autologous transplant.¹⁵  A less intensive myeloablative regime would presumably lead to lower VOD risk, which although less frequent than in allogeneic transplantation, is not 0.¹⁵  Reduced intensity busulfan conditioning has been used in the Memorial Sloane Kettering Cancer Center study (Table 2). By contrast, in the Italian GLOBE study, treosulfan and thiotepa have been used for conditioning, which may reduce toxicity when compared with busulfan.²⁴  Experience thus far suggests that suppression of the bone marrow before HSC by hypertransfusion may improve engraftment. The duration of hypertransfusion and the optimal levels are yet to be determined. Good chelation control before transplant is advisable and is a requirement of some protocols to reduce the risk of liver complications in the peritransplant period. Again, the optimal liver iron concentration values for minimizing hepatic risk have not been formally evaluated. Infertility is a likely consequence of this regime and VOD is also occasionally seen.

After myeloablation, the genetically modified autologous HSPC are returned, where they repopulate the hematopoietic compartment. In 1 trial, infusion was into the posterior superior iliac crest (Table 2) with the aim of enhancing engraftment and minimizing a first-pass intravenous filter.¹⁶  The vector copy number (VCN) in the infused CD34⁺ cells is important to levels of HbA or HbF obtained postgraft. For example, improvement in manufacturing process of LentiGlobin has led to both increased peripheral blood VCN and also increased levels of posttransplant Hb (HbA^T87Q). This has allowed current trials to be extended to more severe forms of thalassemia (β0/β0) (BBB study 212).

Interim combined data have been reported from 2 phase 1-2 studies of autologous HSC transduced ex vivo with the LV BB305 (BBB) vector to 22 patients (aged 12-35) (TDT trials numbers NCT01745120 and NCT02151526).¹⁵  At 26 months (range, 15-42) after infusion, 12 of 13 patients with a non-β(0)/β(0) genotype were transfusion free with HbA(T87Q) levels from 3.4 to 10.0 g/dL, and total hemoglobin ranged from 8.2 to 13.7 g/dL. Furthermore, in 9 patients with the more severe IVS1-110 mutation, the annual transfusion volume was decreased by 73%. No safety issues were attributed to the BB305 vector in either studies 204 or 205. Nine serious adverse events were reported, including 2 episodes of veno-occlusive liver disease attributed to busulfan conditioning. No replication-competent lentivirus was detected in the patients in either study, and serial monitoring of vector integration sites in blood samples has consistently shown polyclonal profiles of unique integration sites without dominant. However, the overall long-term risks remain unclear: in particular, it is unknown whether there is increased oncogenic potential in humans over several decades. However, no clonal dominance related to vector integration has been thus far observed in the LentiGlobin studies. This is a rapidly evolving field and other clinical studies also using self-inactivating LV vectors or CRISPR both in thalassemia and sickle cell disease are less advanced but also promising (Table 2). Ultimately, a conditioning regime that did not have the risk of infertility would also be welcome.

Enhancers of erythropoiesis for thalassemia patients have been known for >2 decades, working either by direct stimulation: erythropoiesis-stimulating agents or by enhancing HbF synthesis such as with hydroxyurea (Hu) and butyrates. Unfortunately, responses have generally been highly variable and unpredictable so that, unlike sickle cell disease, their place in the management of TDT and non-TDT is still unclear. For reasons of space, this section will focus only on agents that have reached the stage of clinical use or clinical trials.

In non-TDT, erythropoietin (EPO) levels decrease with age, as does erythropoietic response to EPO, so that Hb values also tend to fall. EPO acts by binding to the EPO receptor on the earliest erythroid progenitors, burst-forming unit-erythroid, with subsequent activation of Jak2 kinase and downstream Stat5. EPO also mediates erythroblast expansion though suppression of Fas-FasL coexpression, a potentially undesirable effect in non-TDT, risking increased and unwanted expansion of extramedullary erythropoiesis. EPO doses have typically been 1000 IU/kg weekly or twice weekly and mainly short term in uncontrolled trials so that the long-term risks of extramedullary expansion are not well defined. Highly variable responses have been reported: from no response up to 3 g/dL in some non-TDT patients, and with transfusion independence in some hitherto transfused patients. Pegylated EPO is now in a phase 1 clinical trial for β thalassemias (NCT02950857), but this may risk increased extramedullary hematopoiesis if used alone. Surprisingly, in the light of recent reports of iron restriction enhancing erythropoiesis (see the following section), some trials used iron supplementation to further increase the Hb response.

A secondary effect of ineffective erythropoiesis (IE) and hypoxia in thalassemias is the EPO and Jak2 phosphorylation-dependent expansion of immature erythroblasts that are unable to bypass apoptosis at the polychromatophilic stage, to 5 to 6 times that of healthy controls. Chronic anemia in thalassemia skews the erythropoietic response, so that proliferation outpaces differentiation. In principle, inhibitors of Jak2 kinase may decrease IE by improving the balance between proliferation and differentiation. Murine studies in a thalassemia model showed improvement in IE and decrease in spleen size, possibly at the cost of decreased overall erythropoietic activity. Furthermore, in humans with myelofibrosis, a Jak2 inhibitor (INCB018424) reduced spleen size. A phase 2A open label single arm study was therefore undertaken of ruxolitinib (INC424, Jakavi) (NCT02049450) in 30 TDT patients with splenic enlargement²⁵  to determine whether transfusion requirement and/or spleen size could be decreased. Treatment was given for 30 weeks at a starting dose of 10 mg twice daily. Overall, transfusion requirement did not decrease (12 decreases, 7 increases, 8 did not change). However there was a significant 26% reduction in splenic volume at 30 weeks. Thus, although the lack of decrease in transfusion requirements are consistent with murine studies, ruxolitinib may have a role in reducing spleen size in TDT patients with advancing splenomegaly. In principle, after reduction of spleen size and cessation of ruxolitinib, transfusion requirement may then decrease, so that intermittent therapy may be applicable in selected TDT patients with splenomegaly.

HbF-inducing agents increase γ-globin, which binds to unpaired α-chains in β thalassemias, thereby decreasing globin imbalance, ineffective erythropoiesis, and hemolysis. Many drugs increase HbF but few have undergone systematic clinical evaluation. γ-Globin can be increased by cytostatic agents that disturb the balance between proliferation and differentiation: such as Hu. In non-TDT, responses have been highly variable, from very little to >2 g/dL. QOL improvement has been reported when Hb increased with EPO plus hydroxyurea by 1.6 g/dL or by 0.7 g/dL using Hu alone.²⁶  Decreased extramedullary hematopoiesis, pulmonary hypertension, leg ulcers, hypothyroidism, and osteoporosis were reported in a retrospective analysis.²⁷  In TDT, response has been particularly variable. Some reports have been remarkable: 78% of “transfusion-dependent” patients becoming transfusion independent in an Iranian study.²⁸  A Cochrane review found no convincing evidence that Hu reduces transfusion requirement, however. Some of this variability may reflect the local clinical definitions TDT but underlying genetic determinants cannot be excluded.

Short-chain fatty acids such a butyrates activate the Aγ gene promoter and possibly also inhibit histone deacetylase. Despite the occasional dramatic individual responses, initially with intravenous arginine butyrate, Hb responses >1 g/dL have been rare. Responses with orally bioavailable butyrates have been inconsistent and insufficient. Other modulators of HbF have included DNA demethylating agents such as 5-azacytidine and decitabine to hypomethylate the γ-globin genes. Hb increments of about 1 g/dL were reported, but long-term safety and the role in management of non-TDT is not established. Pharmacological enhancers of HbF are under preclinical investigation, such as LSD-1 inhibitors, and may have application for non-TDT. Other targets for HbF manipulation, such as SOX3, KLF1, HBSL1-MYB intergenic region, DRED complex (TR2 and TR4), and Lim domain biding 1 have yet to be evaluated clinically.²⁹ 

These agents were originally conceived to improve bone mineral density in postmenopausal women, preventing osteoclast-dependent bone resorption through inhibition of activin-dependent signaling, but sotatercept unexpectedly increased Hb values. This was later shown with luspatercept (ACE-356), but unlike in sotatercept, increases in bone mineral density were not seen, possibly because ACE-356 does not bind with high affinity to activin A. In murine studies using murine equivalents of these drugs (RAP-536 or RAP-011), Hb increases were seen in both healthy animals and models of anemia that included: thalassemia, chronic renal disease, myelodysplastic syndromes, Diamond Blackfan anemia, and iron-restricted anemia.

Sotatercept (ACE-011), is a recombinant human homodimeric activin type IIa receptor fusion protein comprising the extracellular domain of the human activin type IIA receptor, fused to the Fc domain of human immunoglobulin G1 with 2 disulfide-linked chains, dimerized through the Fc region. Luspatercept (ACE-536) is a similar fusion protein, but for activin receptor type IIB. These act as traps for a wide range of ligands of the transforming growth factor-β superfamily and inhibit signaling by binding ligands extracellularly, sequestering them away from their receptors. Linking clinical effects to inhibition of a single ligand can be challenging, but studies have also provided insights into the regulation of hematopoiesis itself. Effects appear EPO-independent: the kinetics of response are too rapid for an EPO-like effect,³⁰  occurring at later stages of erythropoiesis³¹  and, unlike EPO responses, appear to require accessory cells. Modulation of GDF-11 has been implicated as the mechanism of action in thalassemia³¹ : GDF11 is upregulated in spleen and erythroid cells of thalassemic animals and inhibits murine erythroid maturation. GDF11 expression and recombinant human homodimeric activin type IIB receptor in erythroid precursors decreases progressively with maturation.³¹  Inactivation of GDF11, as with these drugs, decreases oxidative stress and α-globin membrane precipitates in red blood cells. GDF11 also corrected the abnormal ratio of immature/mature erythroblasts in thalassemic mice by inducing apoptosis of immature erythroblasts through the Fas–Fas ligand pathway.³² 

Phase 2 clinical trials in non-TDT with sotatercept showed dose-dependent increments in Hb typically >1 g/dL with improvements in red cell morphology (Table 3).³³  Because of the long plasma half-life of ∼23 days, subcutaneous dosing was given every 3 weeks, with Hb improvement seen within the first 3 cycles in non-TDT. In TDT, a reduction in transfusion burden was also seen. Tolerability was generally good and reported adverse events included grade 3 bone pain in 1 TDT patient and grade 2 phlebitis in a non-TDT patient. There were similar findings in phase 2a trial of luspatercept (ACE-356; NCT01749540) in 31 non-TDT and 32 TDT patients at dose levels titrated up to 1.25 mg/kg every 3 weeks were presented at the European Hematology Association Congress in 2018.³⁴  In NTDT an increase of >1 g/dL over baseline was seen in 71% of patients and of >1.5G in 55% at 3 months. In a patient-reported outcome QOL questionnaire, a significant increase in score was seen in patients showing >1 g/dL Hb increase.³⁴  The 6-minute walk test showed a significant improvement at 48 weeks of treatment and improvement correlated with the Hb increase at this time. In TDT patients, a ≥20% reduction in transfusion requirement was seen in 78% of patients and a ≥30% reduction in 69% of patients after a median treatment duration of 14 months. In patients as a whole, tolerability was generally good with the majority of adverse effects being grade 1-2: bone pain was seen in 38%, headache in 28%, myalgia in 22%, and arthralgia in 19%. Other adverse effects (≥10%) were asthenia, injection site pain, and back pain. This study is now in a 5-year extension phase (NCT02268409). The detailed phase 2 findings with sotatercept and luspatercept are in press.

Luspatercept was selected for phase 3 development, mainly because, unlike sotatercept, it binds only minimally to activin A and thus may have lower off-target effects. Randomized double-blind phase 3 studies are now under way for TDT (NCT02604433: Efficacy and Safety Study of Luspatercept (ACE-536) Versus Placebo in Adults Who Require Regular Red Blood Cell Transfusions Due to Beta [β] Thalassemia) and non-TDT (NCT03342404; A Study to Determine the Efficacy and Safety of Luspatercept in Adults With Non Transfusion Dependent Beta [β]-Thalassemia). The predetermined criterion for efficacy in TDT is a 33% decrease in the number of transfused red blood cell units from baseline. In summary, both sotatercept and luspatercept regularly increase Hb in non-TDT to levels associated with improved QOL. In TDT, the reduction in transfusion burden is likely to decrease the use of chelation and the frequency of hospital visits. It is reassuring that RAP-536 corrected the complications of ineffective erythropoiesis, splenomegaly, and bone pathology in a mouse model of non-TDT.³⁵  However, as transfusion decreases in TDT and autologous erythropoiesis proportionately increases, it will be important to establish that the known complications of non-TDT, such as thrombosis and extramedullary erythropoietic expansion, do not become problematic.

Animal models suggest that by restricting iron delivery to the thalassemic erythron, the efficiency of erythropoiesis can be improved. Iron restriction, using intraperitoneal iron-free human transferrin or by increasing plasma hepcidin levels, improved Hb in thalassemic mice. Both manipulations decrease iron delivery to normoblasts, thereby decreasing heme synthesis, decreasing hemichrome formation, and hence reactive oxygen species–mediated oxidative stress and apoptosis. In principle, plasma hepcidin can be increased by giving exogenous hepcidin(s) or by upregulating hepcidin synthesis. Administration of minihepcidins, short peptides that mimic the activity of endogenous hepcidin, improved ineffective erythropoiesis, anemia, and iron overload in thalassemic mice.³⁶  Hepcidin synthesis can be increased by suppressing Tmprss6, a transmembrane serine protease (matriptase-2) that normally suppresses hepcidin synthesis by deactivating hemojuvelin to form soluble hemojuvelin. An antisense oligonucleotide approach has been shown to decrease Tmprss6 protein, thereby increasing hepcidin and leading to improved anemia in murine thalassemia models.³⁷  Injection of siRNA formulated in lipid nanoparticles directed to the liver improved anemia by ameliorating of IE and improving red cell survival³⁸  in thalassemic mice. Iron chelation has a less clear effect, however, at least when given alone, suggesting that restriction of iron delivery may need to be through the transferrin uptake. This may have a role when combined with hepcidin mimetics, however. In humans, large doses of purified human apotransferrin showed decrements in transferrin saturation or NTBI.³⁹  Several clinical trials are under way or imminently planned to investigate the effect of hepcidin, mini-hepcidin or manipulation through TMPRSS6 on transferrin saturation, NTBI and hence erythropoiesis in non-TDT (Table 3). Secondary effects on iron uptake by myocardium are also under investigation.

John Porter, Department of Haematology, University College London, Paul O'Gorman Building, 72 Huntley St, London WC1E6HX, United Kingdom; e-mail: j.porter@ucl.ac.uk.