Looking to the future of gene therapy for hemophilia A and B

KEYWORDS:

• Factor VIII
• factor IX
• gene therapy
• hemophilia A
• hemophilia B
• AAV
• gene editing

Congenital hemophilia A and B are caused by mutations in the genes encoding coagulation factor (F)VIII and IX, respectively, resulting in lifelong bleeding disorders. After decades of experimentation and clinical fits and starts, the long-awaited first regulatory approvals of hemophilia gene therapies came in 2022, one each for hemophilia A and B [1,2]. Several other adeno-associated viral (AAV) vector-based gene therapies for hemophilia A and B targeting the liver are still in clinical trials with multiyear follow-up data. At the current state of the art, gene therapy can induce long-term therapeutic and sometimes curative FVIII or FIX expression, but it has also shown several important limitations. Wide variability and unpredictability of efficacy, as well as safety uncertainties, have emerged in all trials, with FVIII level decline additionally complicating hemophilia A gene therapies [3–6]. These imperfections necessitate further improvements to the current technology or may require a different gene transfer system to be overcome.

Long-term follow-up from several clinical programs has shown important differences between hemophilia A and B gene therapies [7]. Whereas participants with hemophilia B have maintained stable low levels and stable higher levels of FIX expression at least 7 and 3 years after dosing, respectively (unless they showed signs of cytotoxic T-cell responses toward transduced hepatocytes that persisted despite or reappeared after cessation of immunosuppression [8,9], FVIII levels have been declining in all trials when using the highest vector doses, as much as ~6-fold after 6 years in the longest-running study, which used the highest vector dose in the field (6E13vg/kg), from ~58 to ~10IU/dL. In this Phase 3 study, 13% of participants returned to protein replacement prophylaxis 3 years post-dosing. Lower vector doses produced more stable but lower FVIII levels, in moderate or mild hemophilia ranges [10]. Both hemophilia A and, to a lesser extent, B studies have seen wide variability in expression levels, from no expression to supraphysiological factor levels necessitating thromboprophylaxis [11]. In addition, more hemophilia A trial participants showed signs of mild liver toxicity, with inconsistent evidence for cellular immune response and variable responsiveness to immunosuppression. In Phase 3 trials of the recently approved products, 85.8% and 16.7% of participants with hemophilia A and B, respectively, had ALT elevations [12,13].

These discrepancies point to fundamental differences in gene transfer biology and biosynthesis of the transgene product between AAV-FVIII and AAV-FIX. Unlike FIX, which is naturally made in hepatocytes, FVIII is produced by liver sinusoidal endothelial cells (LSECs) and extrahepatic tissues, and thus transgenic FVIII expression in hepatocytes is ectopic. Also, FVIII is remarkably difficult to express in heterologous cells, and shows a tendency to misfold and activate unfolded protein response, which can trigger cellular stress (which may be interlinked with cellular immune responses) [14]. Additionally, the large FVIII transgene size imposes a more complex and less efficient transduction mechanism compared to AAV-FIX. Poor transduction efficiency necessitates high vector doses, which may contribute to toxicities.

Recombinant AAV has proven itself as a powerful platform for gene transfer, but first-generation vectors have limitations that are being addressed by continued optimization of the coding and regulatory elements of the expression cassette. The inclusion of the hyperactive FIX-Padua variant in AAV-FIX vectors has dramatically improved the efficacy of gene therapy for hemophilia B. Similar hopes drive works on biobetter FVIII variants that would fold and secrete better, permitting lower vector doses and possibly higher circulating FVIII without increasing FVIII biosynthesis. One such variant is FVIII-V3, which contains six N-glycosylation sites from the B domain compressed within a 17-amino-acid peptide and produced 3-fold higher FVIII levels in preclinical studies. It was administered as AAV-HLP-hFVIII-V3 to three participants in the GO-8 study in 2018, with sustained expression in the first year, but no update since. Several other FVIII variants designed to improve folding, stability, and secretion have been described, including ET3, N6, X5, X10, F309S/L303E, IR8, and QQ. ET3 is a porcine-human hybrid FVIII which substitutes 9% of the human FVIII amino acids with porcine sequences, showing up to 100-fold improved secretion in preclinical studies [15]. An AAV vector encoding FVIII-ET3 has entered clinical trials.

It is certainly encouraging that after treatment of a large number of patients no inhibitors (neutralizing antibodies against FVIII or FIX) formed in patients that have received hepatic AAV gene transfer. History of inhibitor formation has been an exclusion criterion, and tolerance to FVIII or FIX was maintained after gene transfer in patients that had not produced inhibitors during prior protein replacement therapy. This outcome is likely favored by the microenvironment of the liver, which often promotes tolerance induction to hepatic expressed protein antigens. It will be interesting if tolerance to FVIII will be observed in patients receiving gene therapy in the future after having mostly been treated with the bispecific antibody emicizumab and may thus be immunologically naïve to FVIII. Encouraged by a series of studies in canine hemophilia A, a recently initiated clinical trial aims to take this concept even further and employ hepatic AAV gene transfer to reverse inhibitors that have formed in FVIII protein therapy [16].

Besides AAV (whose genome is mostly maintained in episomal form in transduced hepatocytes), several other gene transfer systems are being tested in the clinic. Some combine genome integration with AAV transgene delivery, some use a different viral vector, and some are fully non-viral. Three clinical programs with new iterations of integrating lentiviral vector gene therapy have recently started. They transduce autologous hematopoietic stem cells ex vivo and infuse them back into participants. One study targets a FVIII-BDD transgene under a platelet-specific promoter (ITGA2B) to megakaryocytes, resulting in the expression and storage of FVIII in platelets (NCT03818763). This trial is recruiting individuals with FVIII inhibitors since platelet-derived FVIII corrected hemostasis in animal models with preexisting anti-FVIII. The other two studies (NCT04418414, NCT05265767) placed the FVIII-ET3 variant under the CD68 promoter, thereby targeting transgene expression to mononuclear phagocytes. The first participants have already been dosed in these lentiviral trials. The main disadvantage of stem cell infusion-based gene therapies is toxic pre-conditioning, which is required to make room in the bone marrow for transgenic stem cells. However, a new generation of non-genotoxic conditioning agents is under development.

Notably, gene editing has been previously tested in a single clinical study participant, who received a combination of three AAV vectors encoding the left- and right-side zinc-finger nucleases and FIX with albumin homology arms but never launched FIX expression. Several other gene editing approaches are being developed in preclinical studies [17]. One uses a combination of AAV8 vector encoding wild-type FIX and lipid nanoparticles (LNPs) delivering Cas9 mRNA and gRNA that targets the cargo to intron 1 at the albumin locus [18]. This approach achieved curative FIX levels in both murine and NHP models. Importantly, mice dosed as neonates showed stable FIX levels, in contrast to animals treated with episomal AAV-FIX, which partially lost expression with growing body size. This result is promising because pediatric patients, who have the most to gain from early hemophilia correction, are excluded from episomal AAV gene therapies. A similar approach delivering both the transgene and the CRISPR/Cas9 machinery in a dual AAV system for hemophilia A is currently being tested in mice. In episomal AAV gene therapy, high transgene expression from a small number of target cells was shown to induce ER stress and limit durability, so hemophilia A may not be well-suited for transgene expression from under the robust albumin promoter. An alternative gene editing approach for hemophilia A uses a dual LNP transgene insertion system, whereby one LNP ferries a FVIII-encoding transposon (DNA) and the other delivers mRNA encoding Super piggyBac (SPB) transposase, which undergoes translation in the cytoplasm, cuts the FVIII transgene out of the transposon, moves it to the nucleus, and integrates it with the genome. In addition to potential pediatric use, this approach could also permit dose titration to achieve desired expression levels because LNPs do not trigger the neutralizing immune responses that prevent repeat dosing of viral vectors [19]. One drawback of SPB is nonspecific genome integration despite a preference for intergenic or intronic regions.

Each of these new approaches promises to overcome one or more but not all of the obstacles that have hindered the AAV gene transfer in the clinic. Ex vivo lentiviral therapies might be more durable, possibly in the presence of inhibitors, but the aggressive and resource-intensive conditioning makes administration challenging compared to a peripheral vein injection of AAV vectors. All AAV orcombined AAV-LNP gene editing approaches may offer durability and permit pediatric use but will still be limited by inter-individual variability in response, concerns over off-target integration events, and anti-AAV immune responses precluding repeat dosing. On the other hand, non-viral LNP-based delivery of DNA may permit repeat dosing and thus reduce variability, but its efficacy has been underwhelming when moving from small to large animal models. Also, LNPs unload their cargo in the cytoplasm, from where it needs to be moved to the nucleus, introducing an additional hurdle to vector efficiency because AAVs release their payload inside the nucleus. Altogether, while alternative gene transfer systems might bring us closer to a long-awaited cure for hemophilia, they have yet to prove their clinical viability and superiority over the current state of the art.

In the US, currently approved gene therapies for hemophilia have a price tag of approximately $3 M per patient, representing the most expensive drugs in the world. Assuming that the average cost for factor replacement therapy for a patient with severe hemophilia is ~$300,000 per year, a gene therapy that lasts for 10 years will not be more expensive and cheaper if it lasts longer. Taking into account inflation, the cost of gene therapy would likely be lower than traditional protein therapy in less than 10 years. However, this simplistic view does not take into account variability in the efficacy and durability of the gene therapy. It has also been suggested that, given various treatment options, demand in the patient community may not be high enough to maintain these prices. Regardless, publications on multi-year follow-up of subjects that participated in early-phase or phase 3 trials are still sorely lacking. Similarly, some of the less successful trials have not yet been published. We therefore appeal to the investigators and sponsors to make this wealth of data available to the patient, treatment, and research communities in the form of peer-reviewed publications.

Declaration of interest

R Kaczmarek is serving on a scientific advisory board of BioMarin and received research funding from Bayer. RW Herzog is serving on the scientific advisory boards and committees of Regeneron Pharmaceuticals, Pfizer, BioMarin, Spark Therapeutics, Hoffman-La Roche, and Prevail Therapeutics.

The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures

A peer review on this manuscript has received prior research funding from Biomarin. prior advisory board for CSL Behring, Genentech, Sanofi Genzyme, and is currently an advisory board member for Regeneron.

A peer reviewer on this manuscript has received financial support for research from Anthos, Bayer, CSL Behring, Novo Nordisk, and Roche, as well as honoraria for lecturing or consultancy from Alexion, Bayer, CSL Behring, Daiichi Sankyo, Octapharma, Pfizer, Sobi, and Viatris.

Peer reviewers on this manuscript have no other relevant financial relationships or otherwise to disclose.

Additional information

Funding

This paper was not funded.