Gene Therapy for Hemophilia

I am delighted to participate in the preparation of a special issue of the Journal in honor of Dr. George Stamatoyannopoulos. George and I worked together over the years in scientific collaborations, organization of meetings, and intellectual exchanges. For example, we organized biannual hemoglobin switching conferences beginning in 1978 with a meeting in the Battelle Institute in Seattle. ¹ These conferences encouraged work in the field, which is gradually coming to fruition after a period of nearly 50 years, with clear potential for clinical application. I consider George to be a very wise and helpful friend.

Much of this review will focus on the studies for the treatment of hemophilia B as work has progressed most rapidly for that disorder. ² However, efforts are now being refocused at the development of vectors suitable for the treatment of hemophilia A. Both hemophilia A and B can be treated with recombinant factor replacement with significant improvement in morbidity and mortality. However, such treatment is extremely costly and is still often marred by clinical complications, including bleeding, particularly bleeding in the CNS despite factor replacement. The need for frequent administration of recombinant factor concentrates infused 2–3 times per week in order to maintain minimal therapeutic levels is demanding and highly invasive. Some improvement in outcome has been achieved in highly developed countries ³ but the majority of subjects throughout the world lack resources for optimal treatment. ² These considerations prompted efforts to develop novel approaches for treatment of hemophilia using viral vector-mediated gene transfer.

Work on this project began shortly after Amit C. Nathwani arrived from the University College London to work as a postdoctoral fellow in my lab (July 1997). We were just beginning to work with recombinant adeno-associated viral (rAAV) vectors. Dr. Nathwani's initial project involved attempts to transduce primitive hematopoietic cells with an rAAV vector. We soon learned that this project was doomed to failure because of the fact that rAAV vectors do not integrate efficiently and are rapidly lost from the dividing cells. ⁴ Dr. Nathwani came to me and asked if he could work on gene therapy for hemophilia B, a disorder for which he had developed an interest during his prior training at University College London. I agreed with enthusiasm. Dr. Andrew Davidoff, a fully trained pediatric surgeon, was recruited to the project because of the belief that a portal vein injection in mice was needed to insure liver transduction. This turned out not to be true as we subsequently found equivalent transduction after portal versus peripheral vein infusion. ⁵ However, the collaboration between Drs. Nathwani and Davidoff was born and has flourished now for several years. They directed all of the preclinical studies and are leaders in our clinical trial.

rAAV vectors have been derived from the many serotypes of AAV that exist in nature. ⁶ Serotype rAAV2 was used most extensively in early studies because it had been the first serotype that was characterized. AAV viruses are thought to be nonpathogenic. The AAV genome is single stranded and 4.7 kb in length. It includes inverted terminal repeats (ITRs) of approximately 145 bp. ⁷ The ITRs are the only elements that must be retained in the vector genome to allow packaging of the rAAV vector genome. AAV viruses depend on a helper virus for replication, usually adenovirus. Production of viral vector occurs in packaging cells that have been transduced with a multiplasmid system in which a number of adenoviral proteins are expressed as well as the AAV rep and capsid proteins. Initial studies attempting to exploit rAAV vectors for treatment of hemophilia B relied on the use of these single-stranded rAAV vectors. Two clinical trials were completed with limited results.

The first trial involved intramuscular injection of rAAV encoding hFIX in a classic dose-escalation study. ^8,9 Vector doses ranged from 2 × 10¹¹ to 1.8 × 10¹² vg/kg. Eight men were treated. Each received injections of vector at multiple sites. These injections were well tolerated. Muscle biopsies verified that myocytes at the injection sites were transduced and that the vector genome could be detected. However, efficacy, in terms of expression, was very limited, with only 1 participant having detectable hFIX. He was in the low-dose cohort. ⁸ FIX expression has persisted for more than 10 years in this individual. ¹⁰

A second clinical trial focused on liver-targeted delivery. ¹¹ The liver is the natural site for hFIX synthesis and expression of the transgene there was judged to be less likely to induce a neutralizing antibody response to hFIX when compared to intramuscular injections. Seven patients with severe hemophilia B received an rAAV2 vector via hepatic artery injection. Only individuals treated at the highest dose (2 × 10¹² vg/kg) had measurable FIX levels initially. ¹² However, over several weeks, the levels of FIX declined coincident with the appearance of cytotoxic T-lymphocytes in the blood stream, which were judged likely to have destroyed the hFIX-expressing hepatocytes. ^12,13 Systematic studies have resulted in the identification of multiple AAV serotypes. ⁶ Serotype AAV8 was particularly intriguing in that it does not cross react with antibodies directed against other AAV serotypes. Because AAV8 rarely infects humans, the incidence of neutralizing antibodies is relatively low with only approximately 20% of individuals having neutralizing antibodies. Recent studies suggest a significantly higher incidence of anti-AAV8 antibodies in humans. ¹⁴ Furthermore, relatively low titers of antibody abrogate liver transduction. Two studies have shown that transient immunosuppression may facilitate transduction of the liver in vivo in animal models. ^15,16

We choose serotype 8 for our studies because of its efficacy in transducing hepatocytes. ⁶ Another favorable factor for the AAV8 serotype is that the genome is rapidly uncoated, allowing for prompt expression. ¹⁷ Another technical advance that we incorporated into our vector design was the use of a self-complementary vector. ^18,19 Self-complementary vectors have each strand of the hFIX coding genome in an inverted orientation between which there is a 3′ ITR that has been mutated, preventing digestion of the mutated ITR. Self-complementary genomes quickly self-anneal when cells are transduced, enhancing the rate of gene expression as well as the level of gene expression achieved. ¹⁸ We designed our vector to have a factor IX expression cassette that was small enough to accommodate the self-complementary genome by modifying transcriptional control elements (enhancer, promoter, and polyadenylation site) to minimize size and therefore to allow efficient packaging of the self-complementary genome. ^18,19 The FIX coding sequences were codon optimized by synthesis with the most frequently used codons in natural cellular transcripts. We observed a 20-fold improvement in hFIX expression in mice compared to ssAAV vectors. Expression of hFIX was very high despite the use of much lower vector doses than had been applied in earlier studies with ssAAV vectors. ¹⁹

Extensive preclinical studies were performed, initially in mice. ²⁰ Vector was given by tail vein administration as early studies had indicated equivalent liver uptake by peripheral injection compared to portal vein injection. ⁵ scAAV2/8 vector particles were found to be much more efficient at transducing hepatocytes than ssrAAV2/8. A dose-dependent increase in hFIX was observed in mice, with animals receiving the highest dose achieving 100% of the hFIX levels that are found in normal human plasma. Preclinical studies were extended to evaluate transduction of nonhuman primate hepatocytes. ²¹ This was achieved very successfully with a peripheral vein infusion. The vector genome was found predominantly in the liver with lesser amounts in the spleen. Expression of the vector-encoded transcript was found exclusively in the liver as predicted based on the construction of the liver-specific enhancer–promoter combination that was used. Noteworthy is the fact that no toxicity was observed in any of the animals that received vector as part of the preclinical studies. Despite the superiority of AAV8 in mice, the two serotypes seemed more equivalent in nonhuman primates and in the clinical trials to date. Presumably, this reflects the presence of a receptor or other elements that facilitate uptake of the serotype 8 in mouse liver.

Our clinical trial was designed as a phase I/II dose-escalation study in the classic design. ^22,23 The initial dose was 2 × 10¹¹ vg/kg with an intermediate dose of 6 × 10¹¹ vg/kg and a higher dose of 2 × 10¹² vg/kg. Overall 12 subjects have participated in this trial. This report provides detailed results on the first 10, each of whom has been followed 3 years or more after the single vector infusion (Table 1). All 10 have had measurable levels of factor IX. The two patients treated most recently also have stable production of hFIX. Among the initial six patients who received the highest dose, the average hFIX level was 5.1% ± 1.7%, with each having significant production of ≥2.8%. hFIX expression resulted in a dramatic reduction in requirements for factor IX infusions. Several of the patients had a mild elevation in transaminase levels. After observing this phenomenon in the first high-dose patient, we resolved to begin prednisolone as soon as there was a 50% or greater increment in transaminases even if the values remained within the normal range. With this treatment approach, the transaminase elevations resolved over a period of a week or two and steroids could be withdrawn after four weeks with no recurrence of the transaminitis. Fortunately, the prompt treatment permitted maintenance of the hFIX levels in the individual participants.

OPEN IN VIEWER

Table 1.

Clinical data

Patient No.	Duration of observation in months	Factor IX level, vg/kg	Vector dose, vg/kg
1	71	2.1 ± 0.78	2 × 1011
2	66	3.9 ± 2.1	2 × 1011
3	64	2.4 ± 1.5	6 × 1011
4	63	2.8 ± 1.6	6 × 1011
5	61	4.8 ± 2.4	2 × 1011
6	60	7.7 ± 2.4	2 × 1012
7	48	5.7 ± 1.4	2 × 1012
8	44	5.7 ± 1.4	2 × 1012
9	42	4.9 ± 1.6	2 × 1012
10	38	4.8 ± 3.2	2 × 1012

Despite the clinical success achieved to date, significant limitations remain, preventing broad application of this approach. These limitations include the need to produce the vector in a transient transfection system that limits the number of particles that can be prepared. Needed is a packaging cell line in which the viral protein-encoding sequences are incorporated as a stable transgene. Studies exploring the use of a gain-of-function factor IX variant (R338L) support its use in a clinical trial. ^24,25 Indeed, the clinical trial is ongoing with a number of patients having been enrolled and treated. ²⁶

Hemophilia A is a far more common disorder than is hemophilia B. Accordingly, we have great interest in developing an effective rAAV vector suitable for treating hemophilia A. Development of such a vector presented several challenges. ²⁷ The human factor VIII coding sequence is 7 kb in length, too large to be packaged into an AAV capsid with its limiting packaging capacity. hFVIII has three functional domains. The B domain can be deleted, and in the construction of our vector, a 226 amino acid spacer was included in place of the B domain. Amino acid triplets that function as glycosylation sites that are normally part of the B domain were included within this spacer. The resulting 5.2 kb vector was efficiently packaged after its coding sequences were codon optimized. Super physiological levels of hFVIII were achieved in mice and nonhuman primates. ²⁷ An immune response mediated by antibodies targeted to FVIII were eliminated with transient immunosuppression.

Thus, early results from our clinical trial of liver-targeted, AAV-mediated FIX gene transfer for hemophilia B suggest that this approach is safe and efficacious. Now that the challenges of the large size of the FVIII protein have been overcome, a similar, AAV-mediated gene transfer trial for hemophilia A will be opening this year. Further successful experience with this approach for the treatment of hemophilia will likely result in a paradigm shift toward gene therapy for the treatment of this group of diseases.