Gene therapy, an ongoing revolution

In this issue of Blood, Buchlis and colleagues describe the long-term persistence (up to 10 years) of factor IX (FIX) expression in adeno-associated virus serotype 2 (AAV-2)–injected muscles of a patient with hemophilia B.¹ 

This AAV-2 contained a human FIX minigene under the dependence of a cytomegalovirus promoter. The patient had received an intermediate dose (6.0 × 10¹¹ viral genomes [vg] per kg at several sites of both vastus lateralis muscles) during a dose-escalation trial,²  10 years before his death, which was unrelated to the procedure. Despite evidence of gene transfer and expression 10 months after AAV injection and for the rest of his life, his circulating FIX levels remained subtherapeutic (< 1% of normal), presumably because the injected doses were too low. Nevertheless, this study underlines the fact that, despite potential immune responses to any transgene product, a transferred gene can remain transcriptionally and translationally active for many years with no observable inflammatory infiltration at the injection site.¹ 

Gene therapy may eventually become a realistic option for many monogenic diseases, and 2779 gene therapy studies are currently listed in the ClinicalTrials.gov registry. Several strategies to deliver therapeutic genes have been tried, including direct injection of a plasmid encoding the gene of interest into the target tissue (eg, muscle), which has so far achieved only low transduction efficiency³ ; and the use of modified viruses to carry the gene to target cells. The first noteworthy success was obtained by ex vivo transduction of CD34⁺ bone marrow cells with a defective retroviral vector and reinjection into 10 boys with X-linked severe combined immunodeficiency. However, because of insertional mutagenesis, 4 boys developed T-cell acute lymphoblastic leukemia.⁴  AAV vectors have evolved over the past decade and now represent particularly interesting in vivo gene delivery vehicles. Contrary to retroviral vectors, AAV vectors remain essentially episomal after gene transfer, thus minimizing the issue of insertional mutagenesis. The nonpathogenic nature of AAV vectors is a further advantage. Initially however, AAV vectors could only be produced in small amounts for research purposes and for trials involving few injections and low doses.^2,5-9  Efficient AAV gene transfer has nonetheless been achieved, with some clinical benefit, in Leber congenital amaurosis, for example.⁹  This bottleneck has now been overcome by the development of scalable production methods such as the insect cell/baculovirus system, enabling preclinical toxicity studies in large animals and the first human dose-escalation trial of whole-body treatment by peripheral vein administration in hemophilia B patients.¹⁰  The patients in this latter study received a self-complementary AAV-8 vector expressing a codon-optimized human FIX transgene and experienced a real clinical benefit. For example, blood FIX levels reached 8% to 12% of normal in the 2 patients who received the highest dose (2 × 10¹² vg/kg), allowing them to stop prophylaxis with FIX concentrate.¹⁰ 

Following this proof of concept of an AAV-based gene transfer strategy, attention is turning to the immune response that could potentially compromise the immediate or long-term efficacy of this approach. Indeed, transgene expression can be compromised by immune responses to AAV^5,7,8  and/or to the exogenous wild-type protein.¹¹  Natural AAV seroprevalence rates vary widely in the general population, depending on the AAV subtype (from 35% to 75% for AAV-8 and AAV-2, respectively).⁸  All seronegative patients who received an injection of AAV, by the intramuscular or intravenous route, became seropositive in the following weeks.^7,8,10  High titers of anti-AAV neutralizing antibodies may block transfection, whether AAV is injected intramuscularly,^7,8  via the hepatic artery,⁵  or intravenously, at least in animal models.⁸  Possible ways of overcoming this barrier in patients with high titers of neutralizing antibodies (naturally or after a first AAV injection) include the use of an AAV serotype that does not cross-react with the seroconversion serotype, plasma exchange to lower the antibody titer, and/or immunosuppressive therapy. A further concern is the cytotoxic T-cell response to the exogenous wild-type protein, a reaction observed after intramuscular injection for limb girdle muscular dystrophy^7,11  and intrahepatic artery or intravenous injections for hemophilia B (targeting hepatocytes),^5,10  sometimes compromising gene expression a few weeks after AAV injection. Muscle inflammation was observed in the former patients,^7,11  and transient hepatitis in the latter.^5,10  Short-course corticosteroid therapy has been proposed to control this reaction,¹⁰  and also immunosuppressant therapy of the type used to prevent graft-versus-host disease.

We are thus at the beginning of a gene therapy revolution for patients with monogenic diseases. Differences in the reported frequency of detectable immune responses across clinical studies may be due to differences in the AAV doses, the quality (purity) of the vector preparation, the route of administration, the AAV serotype, or the promoter. Immunomodulation, as currently used in organ transplantation, will no doubt be necessary, but the optimal choice of drugs and schedules, and the patients concerned, remain to be determined. In view of these concerns, the results from Buchlis and colleagues are particularly encouraging, as they show that therapeutic genes delivered by an AAV vector can be expressed for at least a decade.