Moving Forward Toward a Cure for Hemophilia B

AAV vectors are among the most promising to treat hereditary diseases by gene therapy. Long-term expression of therapeutic genes has been demonstrated in preclinical models and in clinical trials after AAV delivery in various tissues. In particular, AAV gene transfer to the retina resulted in long-term correction of RPE65 deficiency, a rare form of congenital blindness.^1,2,3 Similarly, AAV-based gene therapy to the skeletal muscle resulted in a sustained reduction of postprandial chylomicron levels in patients afflicted by lipoprotein lipase (LPL) deficiency,^4,5 possibly reducing the risk of pancreatitis, one of the hallmarks of this genetic disease. This latest clinical advance constitutes the basis of the first Marketing Authorization Approval—albeit under specific restrictive conditions—of a gene therapy product (i.e., Glybera, developed by uniQure) by the European Medicines Agency.⁶ AAV has also been explored since the 1990s as a vector to treat hemophilia B via clotting FIX gene therapy.⁷ Although FIX expression could be detected after liver-directed gene therapy, the levels were transient.⁸ This transience was possibly due to the immune clearance of transduced hepatocytes by AAV-specific T cells,⁹ consistent with the occurrence of transient liver toxicity as measured by elevated serum transaminases. In the muscle, however, FIX expression was sustained for 10 years and no transient toxicity was apparent, but the FIX production by the myofibers was insufficient to yield detectable FIX levels in the blood.^10,11

In a report published in the New England Journal of Medicine,¹² Nathwani and colleagues (University College London (UCL) and St. Jude Children's Research Hospital) provided an update on an AAV serotype 8 (AAV8) vector-based gene therapy trial for hemophilia B (ClinicalTrials.gov NCT00979238), extending the findings of a previous report.¹³ In this trial, patients suffering from severe hemophilia B (<1% FIX) were injected by peripheral vein administration with an AAV8 vector encoding a codon-optimized FIX. A single intravenous vector infusion in 10 hemophilia B patients resulted in a dose-dependent increase in circulating FIX levels between 1 and 6% of the normal value over a median period of 3.2 years. In the high-dose group (i.e., 2 × 10¹² AAV8-FIX vg/kg), a consistent increase in FIX levels to 5.1 ± 1.7% was observed in all six patients. However, some short-term toxicity was apparent, in that most of the high-dose patients developed transaminitis between week 7 and week 10 after treatment, which could be resolved by short-term immune suppression with prednisolone. It is reassuring that no late toxic effects from the gene therapy were apparent. Most importantly, the levels that were attained resulted in a significant reduction in the bleeding frequency consistent with a diminished requirement for repeated FIX protein injections. This represents a significant improvement compared to conventional prophylactic protein substitution therapy based on plasma-derived or recombinant FIX proteins that required lifelong multiple injections per week.

Obviously, the efficacy and safety of the gene therapy approach described by Nathwani and colleagues would need to be confirmed in larger cohorts in phase III clinical trials, prior to obtaining marketing authorization approval by the European Medicines Agency and the US Food and Drug Administration. Considering that patients received up to 2 × 10¹² vg/kg, it is clear that advances in vector manufacturing are needed to move toward phase III trials and ultimately address the market needs. This could potentially be addressed by using baculovirus-based AAV production systems that are more readily amenable to upscaling than conventional production methods based on triple transfection. Using the same AAV-FIX vector design as the one used in the UCL–St. Jude trial, a clinical trial has been initiated based on the use of an AAV5-based serotype, manufactured using the baculovirus system. It is currently not yet clear whether AAV5 can achieve comparable gene transfer efficiencies compared to AAV8 in humans, although in nonhuman primates their performance was comparable.¹⁴

Clearly, achieving sustained expression of therapeutic FIX levels for 3 years is an important milestone for hemophilia and for the field at large. It is also encouraging that the levels approximated the 5% threshold, which is sufficient to convert a severe to a mild phenotype (5–40%).¹⁵ Nevertheless, the maximum FIX levels attained were insufficient to cure the bleeding diathesis. Indeed, the treated subjects were still at risk of bleeding and occasionally required FIX protein injections to treat their bleeding episodes. Moreover, the majority of patients enrolled in the UCL–St Jude trial receiving the highest vector dose (four out of six) experienced liver toxicity requiring short-term immune suppression with prednisolone. This treatment seemed to have been effective in diminishing inflammation and the ensuing liver toxicity. It was initially hypothesized that transient immune suppression may curtail the AAV capsid–specific T-cell response that accounts for the loss of AAV-transduced hepatocytes as measured by elevated transaminase levels.¹³ This hypothesis is consistent with the results obtained in some of the earlier AAV-based clinical trials for hemophilia and LPL deficiency.^8,16 However, some of the trial subjects in the UCL–St. Jude trial developed transaminitis without showing any detectable AAV-specific T cells, as measured in ELISPOT assays. Conversely, some trial subjects had detectable AAV-specific T cells but failed to develop transaminitis. Hence, other yet unknown factors or immune mechanisms may account for these apparent discrepancies.

Ideally, FIX levels should be increased to normal levels (>50%) and the risk of transaminitis should be further reduced, obviating the need for immune suppression. Consequently, there is still a need to further improve the efficacy and safety of gene therapy for hemophilia. This justifies the development of improved “second-generation” AAV vectors that allow for higher FIX expression levels at lower and thus potentially safer vector doses. This could possibly be achieved by using a hyperactive FIX-R338L (Padua) transgene. We had demonstrated that incorporation of a gain-of-function R338L mutation in the FIX protein resulted in a 5- to 10-fold increase in clotting activity using both lentiviral and AAV vectors.^17,18 This hyperactivating mutation was previously identified in thrombophilic patients who express FIX protein with more than eight times the normal specific activity, presumably with more efficient generation of thrombin.¹⁹ When incorporated into a gene therapy vector, this gain-of-function R338L mutation translated into a significant dose advantage, as substantially lower vector doses could be used to achieve a comparable therapeutic effect.

These encouraging studies prompted preclinical studies in canine hemophilia B models^20,21,22 that confirmed our initial results in hemophilic mice regarding the enhanced functionality of FIX-R338L Padua.^17,18 These preclinical studies justified the use of the hyperactive FIX-R338L for clinical applications in patients suffering from severe hemophilia B. It is noteworthy that a clinical trial is ongoing in patients suffering from severe hemophilia B, based on AAV8-mediated liver-directed gene therapy of the hyperactive FIX-R338L Padua (ClinicalTrials.gov NCT 01687608). It is particularly encouraging that FIX activity levels at 10% levels or higher had been reported.²³ Another strategy that could be used to boost FIX expression levels is based on the design of more robust liver-specific promoters. We have recently identified hepatocyte-specific transcriptional cis-regulatory modules (CRMs) using a computational strategy that increased FIX levels more than 10-fold.^18,24,25 Vector efficacy could be enhanced further, by combining these novel hepatocyte-specific cis-regulatory modules with a codon-optimized hyperfunctional FIX-R338L Padua transgene, yielding one of the most robust vector designs for hemophilia B gene therapy to date. Finally, the use of novel hepatotropic AAV vectors that selectively increase gene transfer efficiency into human hepatocytes compared to AAV8 could potentially result in higher FIX expression levels.²⁶

Because the clinical trial data revealed a vector dose–response, administering higher vector doses could, theoretically at least, result in increased circulating FIX levels. However, it is likely that increasing the AAV vector dose would also inadvertently increase the risk of liver inflammation and toxicity. Though this undesirable side effect could potentially be mitigated by transient immune suppression, it is not clear whether there would be a threshold vector dose, above which transient immune suppression would become less effective. In a worst-case scenario, the high vector doses could provoke an uncontrolled inflammatory reaction that could potentially be dangerous. This could be compounded by a more potent innate immune reaction at higher AAV vector doses.²⁷ From a safety perspective, this would again justify the need to generate more robust gene therapy vectors that express higher levels of functional FIX proteins. At first glance, it seems reassuring that vector doses exceeding the highest dose used in the trial (i.e., 2 × 10¹² vg/kg) had not been associated with any dose-limiting toxicity in preclinical animal models. However, it has been particularly challenging to model this AAV-specific immune response and ensuing liver toxicity in mice, dogs, or even nonhuman primates. Recently, Herzog and colleagues established an alternative murine model that overcomes this limitation and that is well suited to assess this potential risk.²⁸ In this model, ex vivo–expanded AAV capsid-specific CD8⁺ T cells are adoptively transferred in mice that were subjected to AAV2- or AAV8-based gene therapy. Consequently, this provoked the loss of AAV2- or AAV8-transduced hepatocytes consistent with transaminitis.

Would it be possible to diminish the risk of liver toxicity by modifying the vector rather than by subjecting the patient to immune suppression, which is in itself not a risk-free intervention? It is unlikely that changing AAV serotypes would suffice, because they share common T-cell epitopes that are recognized by the peripheral blood mononuclear cells obtained from the treated patients.⁹ Instead, it would make more sense to modify the intracellular processing of the AAV capsids to the extent that the presentation of antigenic peptides on major histocompatibility complex class I is reduced. For instance, this could be accomplished by using capsid mutants that are less susceptible to proteasomal degradation and concomitantly resulted in reduced hepatotoxicity.²⁸ We have recently generated an alternative “immune stealth” AAV capsid that inhibits the proteasome directly and efficiently prevents major histocompatibility complex class I presentation of AAV capsid–derived peptides, without interfering with the transduction efficiency per se (unpublished observations).

The recent clinical findings by Nathwani and colleagues¹² constitute an important step in the right direction. Nevertheless, gene therapy for hemophilia B is not quite ready for ‘prime-time'. A bona fide cure of hemophilia B has not yet been established and some short-term safety issues remain to be addressed. Consequently, these clinical data justify the development of improved ‘next-generation' AAV vectors that allow for higher FIX expression levels at lower and thus potentially safer vector doses. This could possibly be achieved by using hyperactive FIX (Padua), more robust cis-regulatory modules to drive the therapeutic transgene and/or improved hepatotropic AAV capsid variants. It is therefore important to continue to move forward cautiously towards the development of a safe and effective cure for hemophilia in order to ultimately maximize the benefits for those patients and their families that are afflicted by this severe bleeding disorder.