Genetic medicine gestating for α-thalassemia

In this issue of Blood, Chappell et al describe a pair of innovations, a novel mouse model and an effective gene therapy approach, that renew prospects to develop definitive therapies for severe α-thalassemia.¹ 

Thalassemias result from imbalanced production of the α- or β-globin chains of hemoglobin. β-Thalassemia has dominated the headlines of late, with 2 gene-modified autologous hematopoietic cell therapies now approved by the Food and Drug Administration: first, betibeglogene autotemcel, relying on lentiviral globin gene addition, and most recently exagamglogene autotemcel, based on fetal hemoglobin induction by gene editing the BCL11A erythroid enhancer.^2,3 Each of these therapies was codeveloped for sickle cell disease, since antisickling globin addition or fetal globin induction may also prevent sickle hemoglobin polymerization.

In contrast, relatively little attention has been paid to developing genetic medicine for α-thalassemia, which can represent an especially severe form of thalassemia. Deletion of all 4 of the duplicated α-globin genes results in the hemoglobin Bart’s hydrops fetalis syndrome (BHFS) and produces fetal demise. Heterozygosity for α⁰ alleles is relatively common, especially in Southeast Asia, where population frequencies up to 5% are observed,⁴ yet fewer than 100 homozygous patients worldwide have been reported to survive BHFS, all receiving rescue treatment with intrauterine or immediate postnatal transfusions followed by indefinite regular transfusions or bone marrow transplant.⁵ 

One challenge in studying this disorder has been its severity; the mouse model of complete α-globin deficiency suffers late gestation lethality, similar to the human condition.⁶ Chappell et al developed an adult-stage mouse model that builds from their prior work in which they showed that delivery of RNA by lipid nanoparticles (LNPs) to hematopoietic stem and progenitor cells (HSPCs) could be augmented by decorating the particles with an anti-CD117 antibody and that Cre messenger RNA is a particularly potent cargo, perhaps because relatively few copies need to be expressed to achieve recombination.⁷ In an impressive display, the authors first generated a floxed α-globin mouse line, then targeted hematopoietic cells ex vivo with LNPs, which led to extremely efficient gene deletion of all copies of α-globin. Finally, upon transfer to myeloablated recipients, α-globin-deleted HSPCs yielded a severe α-thalassemia phenotype in adult mice.

Affected adult mice showed β-globin precipitates on red blood cell membranes, imbalanced α/β globin ratio, hypochromic erythroid indices, and a preponderance of hemoglobin H (β₄ tetramer), all as expected for severe α-thalassemia. There was evidence of ineffective erythropoiesis with erythroid hypercellularity of bone marrow and especially spleen (the latter being the predominant mouse erythropoietic organ). Erythrocytes had short half-life and abnormal morphology, with reticulocyte count correspondingly elevated, indicative of hemolysis. Paradoxically, the mice did not demonstrate anemia but rather erythrocytosis. This seemingly contradictory finding could be accounted for because hemoglobin H (HbH) has markedly elevated oxygen affinity, failing to deliver oxygen to tissues. The animals showed elevated erythropoietin, suppressed hepcidin, and iron overload and died approximately 7 to 8 weeks posttransplant, around the time when erythrocytes produced prior to thalassemia onset would be expected to have fully turned over. Strikingly, these mice had high levels of circulating thrombin-antithrombin complexes and microvascular occlusions in their lungs and brain.

This model provides insights into the pathophysiology of severe α-thalassemia, where tissue hypoxia prevails, associated with a vicious cycle of oxygen delivery failure, triggering accelerated erythropoiesis, which produces red blood cells carrying essentially useless hemoglobin. This fits clinical observations from surviving patients with BHFS, where tissue hypoxia and silent ischemic infarcts have been observed in the face of hemoglobin levels typical for transfusion-dependent β-thalassemia, suggesting functional anemia due to presence of nonfunctioning HbH cells.⁸ Whether the microvascular occlusions observed in the lungs and brains of the mice correlate with authentic human BHFS pathology requires further investigation.

The second welcome contribution from Chappell et al is the development of a novel potent lentiviral vector for α-globin gene addition, closely hewing to successful design features of β-globin gene therapy vectors, with β-globin regulatory elements driving high-level erythroid expression. As demonstrated in erythroid cell lines, primary human HSPCs in which the α-globin gene has been knocked out by CRISPR-Cas9, and in the mouse model, approximately 1 vector copy yields the output of a single endogenous α chain. This implies that 1 copy might convert a patient’s hematology from BHFS to HbH disease and 2 copies might lead to α-thalassemia trait. Lentiviral transduction ex vivo was able to rescue survival of recipient mice transplanted with otherwise lethal α-thalassemia, improving many of the hematological features (see figure). Although vector potency, expression heterogeneity based on insertion position and copy number, insertional toxicity, and scalability and sustainability of lentiviral therapies need to be considered, numerous patients might stand to benefit with intermediate forms of α-thalassemia, namely HbH disease with a single normal copy of α-globin, who present with moderate to severe hemolytic anemia, iron overload, and intermittent or regular transfusion requirement.

But the most severe form of α-thalassemia, with the greatest unmet need, is BHFS, which manifests in utero. By the time the diagnosis is made, irreversible injury, including neurodevelopmental insult and congenital anomaly, may have already occurred. It seems apt that the mouse model described by Chappell et al exploited LNP technology (albeit applied ex vivo), since one of the questions the study begs is whether in vivo delivery application of nanotechnologies, such as LNPs, engineered viruses, or virus-like particles, could enable in utero delivery to genetically modify hematopoietic stem cells (HSCs) in situ. Prenatal gene therapy has theoretical advantages like the low absolute number of target HSCs that may limit gene therapy reagent requirement, the proliferative nature of fetal HSCs that may facilitate efficient gene correction, and the immune privilege of the fetus that may protect against inflammatory responses. Substantial challenges remain, including methods for efficient cellular delivery to and gene modification of HSCs in situ, high-particle doses that are needed to overcome in vivo sinks including hepatocytes and reticuloendothelial cells, and immunogenicity.⁹ Beyond these technical considerations, the real-world and ethical challenges of offering advanced complex therapeutics to patients facing hydropic gestations mostly in global resource-limited settings portend vexing issues ahead before innovative genetic medicine might be equitably delivered to patients in need with severe α-thalassemia.

Conflict-of-interest disclosure: D.E.B. declares no competing financial interests.