Gene transfer studies for the treatment of hemophilia began more than two decades ago. A large body of pre-clinical work evaluated a variety of vectors and target tissues, but by the start of the new millennium it became evident that adeno-associated viral (AAV)-mediated gene transfer to the liver held great promise as a therapeutic tool. The transition to the clinical arena uncovered a number of unforeseen challenges, mainly in the form of a human-specific immune response against the vector that poses a significant limitation in the application of this technology. While the full nature of this response has not been elucidated, long-term expression of therapeutic levels of factor IX is already a reality for a small number of patients. Extending this success to a greater number of hemophilia B patients remains a major goal of the field, as well as translating this strategy to clinical therapy for hemophilia A. This review summarizes the progress of AAV-mediated gene therapy for the hemophilias, along with its upcoming prospects and challenges.
Hemophilia is the X-linked bleeding diathesis caused by mutations in the genes encoding factor VIII (FVIII) or factor IX (FIX), respectively the cofactor and the enzyme responsible for catalyzing the conversion of factor X to activated factor X in the intrinsic pathway of the coagulation cascade. The disease is characterized by recurrent bleeds, primarily into the joints and soft tissues, but bleeding into other closed spaces such as the intracranial space may also occur and may be associated with considerable morbidity or mortality (1). Hemophilia A and hemophilia B are indistinguishable clinically and were first distinguished in the clinical coagulation laboratory in the 1950s (2,3). The incidence of hemophilia is ∼1 in 5000 male births (4), hemophilia A being about four times as common as hemophilia B. Clinically, patients are classified as severe, moderate or mild; severely affected patients constitute the largest group, and have <1% normal circulating levels of FVIII or FIX. Mildly affected patients have ≥5% of normal levels, and are free of the spontaneous bleeding episodes that characterize severe disease; moderately severe patients have factor levels between 1 and 5%, and their clinical presentation is also intermediate between severe and mild. Currently hemophilia is managed by intravenous infusion of clotting factor concentrates, which can be given prophylactically, or ‘on demand’, i.e. in response to a bleeding episode. Most moderate or severe patients administer factor somewhere between 20 and 100+ times/year.
Gene Therapy for Hemophilia: Rationale and Early Trials
Since the isolation of the genes encoding FVIII (5) and FIX (6), hemophilia has been an attractive target for investigation of gene therapy approaches, and the level of activity in terms of clinical trials of gene therapy for hemophilia reflects this (www.clinicaltrials.gov). Characteristics that support the attractiveness of this target include: (i) latitude in the choice of the target tissue. Biologically active clotting factors can be synthesized in a range of cell types, and will be effective so long as the gene product reaches the circulation. (ii) Wide therapeutic window. Most individuals with hemophilia are severely affected, with <1% of normal levels of clotting factor activity, but raising levels even modestly into the moderately severe range (>1, <5%) will markedly improve the clinical phenotype; raising levels into the mild range (≥5%) will prevent spontaneous bleeding episodes and greatly reduce the patient's dependence on exogenously infused clotting factor. On the upper end, raising the level to 100% still leaves the patient within the normal range. Thus, a wide range of transgene expression falls into the therapeutic window. (iii) The existence of small (genetically engineered mice) and large (naturally occurring dog) animal models of hemophilia (reviewed in 7). This has meant that most strategies can be evaluated in animal models prior to clinical trials in humans. (iv) The transgene product is easy to measure (in any hospital coagulation laboratory) from a blood sample and is an accepted endpoint for product registration since it correlates well with the severity of the disease and clinical outcome in terms of the annualized bleeding rate.
The size difference between the cDNA for FIX (2.8 kb if the long 3′UTR is included) and FVIII (∼4.4 kb even for the B-domain-deleted construct) explains the differences in vector choice in the early trials. The first wave of gene therapy trials for hemophilia A, starting in 1998, utilized retroviral (8), adenoviral (sponsored by GenStar Therapeutics, unpublished) and plasmid vectors (9). Retroviral and adenoviral vectors were delivered intravenously whereas plasmid vectors were ex vivo electroporated into autologous fibroblasts, which were then implanted on the patient's omentum in a laparoscopic procedure. The initial trials for hemophilia B (vide infra), both used adeno-associated viral (AAV) vectors, delivered to either skeletal muscle or to the liver via infusion into the hepatic artery in the interventional radiology suite. All of these trials were first in class, and all appeared generally safe, but none achieved long-term expression at therapeutic levels. However, infusion of an AAV vector into the liver in a subject with severe hemophilia B (10) clearly resulted in therapeutic levels of expression (>10% normal) for a period of several weeks, and laid the groundwork for the current generation of trials, which all involve hepatic transduction by AAV vectors infused intravenously.
AAV Vectors for Hemophilia B
AAV vectors are engineered from a parvovirus (11). The recombinant vector has tropism for a range of target tissues including the liver, cell types in the retina and the central nervous system, skeletal muscle, and cardiac muscle, among others (reviewed in 12). The DNA sequences carried by recombinant AAV vectors are stabilized predominantly in an episomal form so that long-term expression can occur only with delivery into long-lived, post-mitotic cell types; the vector DNA integrates at a very low frequency and is typically lost from replicating cells (13). One of the main limitations of AAV vectors is that they cannot package inserts of more than ∼5 kb (Fig. 1) (14); this explains the initial focus on hemophilia B in the AAV work. Studies in the large animal model of hemophilia B (15) established clear proof of concept, showed a favorable safety profile and accurately predicted dosing requirements in human subjects. Based on these data, eight subjects were enrolled in the first muscle-directed, AAV-based clinical trial for hemophilia B (16, 17). Importantly, no vector-related toxicity was observed, and there was evidence of FIX protein expression in muscle cells up to 10 years after AAV2-FIX administration (18). However, circulating FIX failed to rise to >1% and disease phenotype was not improved, suggesting that the secretion of the synthesized transgene product into the circulation was not efficient.
In the first liver-directed AAV trial for hemophilia B, a single-stranded AAV2 vector expressing human FIX was infused via the hepatic artery into seven subjects (10). Efficacy was observed in the first of the two subjects that received the highest vector dose of 2 × 1012vg/kg, with peak FIX levels reaching ∼10% of normal. Unexpectedly, an asymptomatic, self-limited rise in hepatic transaminases was observed around week 4 after vector infusion that coincided with the onset of a gradual loss of FIX activity. Both of these events were attributed to the destruction of transduced hepatocytes by AAV capsid-specific memory CD8+ T cells (19). This observed immunogenicity against the capsid had not been predicted by any animal model, and several hypotheses were formulated to explain it. Among others, uptake by dendritic cells of the AAV2 virion in a process mediated by binding to heparan sulfate proteoglycans followed by the activation of capsid-specific T cells (20) or the presence of alternative open reading frames in the FIX coding sequence (21) were proposed as the culprits. Notably, after a decade of intense work, the immune response against the capsid remains a poorly understood phenomenon that is not well-modeled in mice (22). The other subject in the high-dose cohort yielded the second valuable lesson learned from that trial, i.e. pre-existing anti-AAV neutralizing antibodies (NAbs), even at modest titers, are able to prevent successful liver transduction after systemic vector administration.
The second liver-targeted AAV trial for the treatment of hemophilia B, conducted by investigators at St Jude Children's Research Hospital and University College London, differed from the first study in two main aspects: (a) it utilized a self-complementary vector genome that was (b) packaged into an AAV8 capsid, administered by peripheral vein infusion. Based on the pre-clinical data available at the time, both modifications were expected to result in significantly higher FIX levels although the extent of any contribution of these two factors is now unclear (23,24). Three vector doses were used (2 × 1011, 6 × 1011 and 2 × 1012vg/kg), with the high-dose mediating peak expression levels at 8–12% of normal (25). More recently, data from 10 patients were reported, with a follow-up period of up to 4 years (26). Several observations of paramount importance were made in this study. First, all patients achieved long term, stable FIX expression with average FIX levels of ∼5% of normal in all six patients in the high-dose cohort. Secondly, in four of these six patients, a transient increase in LFTs was observed between weeks 7 and 10 after AAV administration, likely as a result of a T-cell response against the AAV8 capsid. Notably, the prednisolone treatment was able to control this response and serum alanine aminotransferase (ALT) levels returned to normal within days. Elevated ALT episodes were not recurrent and no late toxicity was reported, establishing a favorable safety profile for this gene transfer protocol. Undoubtedly, these successful results represent a milestone in the gene therapy field, and a goal of much ongoing work is to replicate and extend them.
While the clinical improvement in patients who achieved stable FIX levels of ∼5% of normal is indisputable, risk for excessive hemorrhage after trauma or surgery would be significantly reduced if stable levels were close to 50%. A more recent Phase 1/2 trial sponsored by Baxalta (clinical trials identifier no.: NCT01687608) also utilized an AAV8 capsid packaging a self-complementary cassette, but expressing FIX Padua. This naturally occurring FIX variant has an activity-to-antigen ratio of around 8–9 (27). A total of seven patients have been treated: two at 2 × 1011 vg/kg, three at 1 × 1012 vg/kg and two at 3 × 1012 vg/kg [World Congress of the International Society of Thrombosis and Haemostasis (Toronto, Canada, June 2015)]. Peak FIX activity values in the third cohort subjects reached 30–60% of normal, highlighting the potential of the Padua variant. However, expression declined sharply around week 6, coinciding with an elevation in ALT levels. One subject in the medium-dose cohort has had sustained FIX activity levels of 20–25% for a year, whereas FIX antigen levels in the two other subjects declined over time. Finally, the first treated subject in the low-dose cohort showed no detectable FIX activity whereas FIX activity remains stable at 3% in Subject 2 (thus, antigen levels must be <1% of normal). The reasons behind the differences in the outcome of these two somewhat similar trials (both utilized the same AAV8 capsid and a self-complementary genome configuration) remain to be determined. A potential explanation could lie in differences (however minor) in vector design or manufacturing, which may significantly alter the kind and/or the magnitude of the immune response. This underscores the need to develop relevant pre-clinical models to build a more complete understanding of vector immunogenicity associated thus far with AAV gene delivery in humans.
The T-cell response against the capsid is not the only limitation that the immune system imposes on AAV-based treatments for hemophilia. As mentioned above, even low levels of pre-existing circulating NAbs against the vector can completely inhibit liver transduction after systemic administration (28). This means that as many as 40% of adult hemophilia B patients may be ineligible to participate in liver-directed AAV trials. Recent data with phylogenetically distant capsids suggest that switching serotypes may not offer a significant improvement (29). Several strategies have been devised to overcome the presence of NAbs. We have shown that empty capsids may be used as decoys (30) and that B-cell depletion using rituximab can decrease anti-AAV antibodies titers in rheumatoid arthritis patients (31). Others have suggested plasmapheresis (32), chemical or genetic modification of the AAV capsid (33,34), saline flushing prior to portal administration (35) and even using naturally enveloped AAV vectors with extracellular vesicles (36).
In addition to the short-term risks posed by the immune response, which primarily relate to efficacy rather than safety, there is the potential risk of late adverse events related to insertional mutagenesis. Although AAV vectors are stabilized mainly in an episomal form, it is clear that some low-level of integration into the host cell genome occurs (37). There have been reports of an increased incidence of hepatocellular carcinoma (HCC) in mice injected with high doses of AAV vectors in the neonatal period (38,39), and integration of sequences derived from wild-type AAV2 has been found in human HCC tissue samples (40). In mice, the incidence of HCC is highly dependent on the AAV vector dose, the ability of the regulatory elements to promote increased transcription of proximal genes and the timing of vector delivery (neonatal versus adult administration) (41). Thus, using the lowest effective dose of a vector devoid of enhancers would seem the safest approach to minimize risk. Long-term surveillance of large animals injected with vector, as well as pharmacovigilance in human subjects, will be required to address the likelihood of this risk in subjects injected as adults. Importantly, tumors have not been observed in human subjects injected as long ago as 2001 (42).
AAV Vectors for Hemophilia A
The development of AAV-based therapies for the treatment of hemophilia A is still at an early stage, with one trial sponsored by BioMarin (clinical trials identifier no.: NCT02576795) recently begun. One of the main challenges is fitting the FVIII expression cassette within the restricted packaging limits of AAV. At ∼7 kb, the cDNA for FVIII exceeds the packaging capacity of adeno-associated vectors, which is ∼5 kb (14). The use of a B-domain deleted form of FVIII (4.37 kb in size) circumvents this limitation, but imposes a stringent constraint on the size of the regulatory elements that control FVIII expression. In addition, FVIII expression is significantly lower than that of similarly sized clotting factors such as factor V (43). Codon-optimization of the coding region has been reported to increase FVIII expression over 40-fold compared with the wild-type sequence (44). Moreover, circulating FVIII levels may be further increased by using a variant that contains a 17 amino acid synthetic sequence flanked by 14-aa SQ residues from the N- and C-terminal ends of the B domain (45). While the presence of the synthetic spacer allows for an increase in circulating hFVIII levels, the use of a non-wild-type FVIII sequence in hemophilia A patients raises concerns about increasing the risk of inhibitor development due to its potential neo-antigenicity. Generating an AAV-FVIII vector capable of expressing therapeutic levels of FVIII at a clinically relevant dose without adding any neo-antigens to the protein remains an unmet goal.
The most difficult problems in hemophilia care currently revolve around patients who develop inhibitors, to infused factor. These occur in as many as 20–30% of all hemophilia A patients (46); they are much less frequent among those with hemophilia B. Currently, the standard of care for these individuals is a so-called immune tolerance induction regimen (ITI) in which patients are infused with large doses of clotting factor daily, a strategy that results in eradication of the inhibitor and restoration of normal FVIII pharmacokinetics in ∼74% of cases (47). Although the mechanism by with ITI eradicates the inhibitor is not worked out, this would seem to be an excellent setting for a gene therapy strategy, which exposes the patient to continuous levels of FVIII without the need for daily infusions. Indeed studies in hemophilia A dogs that had developed inhibitors demonstrated eradication of inhibitors following AAV-mediated, liver-directed gene therapy (48). These strategies have not yet been tested clinically.
These recent clinical trials using AAV-mediated gene transfer have underscored the tremendous potential of gene therapy for the treatment of hemophilia. However, an unanswered question is whether episome-derived liver expression will be sustained in a setting of substantial liver proliferation, as in pediatric patients [the liver quadruples in size during the first 4–5 years of development (49)] or those with liver disease (e.g. hepatitis and/or cirrhosis). For these groups of patients, the integration of the transgene to avoid AAV dilution and loss of expression could be especially beneficial. Long-term expression of FIX in hemophilic dogs has been recently described (50). It has also been shown that in vivo site-specific genome editing can be applied successfully with therapeutic benefit, achieving long-term expression of human coagulation factors in mice (51,52). These studies used AAV to deliver zinc-finger nucleases to the target hepatocytes; the extension of the strategy to large animals has been reported in abstract form (53). Plans are underway to investigate this strategy clinically (September 9, 2015—RAC meeting: http://videocast.nih.gov/launch.asp?19141).
A proof-of-concept for the long-standing goal of achieving sustained and therapeutic levels of a clotting factor after a single vector administration has been fulfilled in the case of hemophilia B (26), and a number of trials have been initiated over the past year (clinicaltrials.gov) with the objective of confirming and consolidating this landmark result. Replication of this success in larger patient cohorts is an immediate goal, followed closely by efforts to extend the approach to hemophilia A (45), an objective that has not yet been realized. Extension to those with liver disease, anti-AAV NAbs, FVIII and FIX inhibitors, as well as the pediatric population, are longer term goals for investigation.
Conflict of Interest statement. K.A.H. and X.M.A. are employees of Spark Therapeutics, Inc. The company's hemophilia B program is partnered with Pfizer Inc.
Previous work was supported by the Center for Cellular and Molecular Therapeutics at The Children's Hospital of Philadelphia, the Howard Hughes Medical Institute and the US National Institutes of Health (grants HL64190, HL078810 and HV78203). Current work is supported by Spark Therapeutics, Inc.