Adeno-Associated Viral Vectors for Hereditary Emphysema;Progressing Along the Clinical Development Highway

A deno-associated viral (rAAV) vectors were introduced into clinical trials some 16 years ago. In the first trial (Flotte et al., 1996, 2003), an AAV vector with a capsid from serotype 2 was introduced into the lungs of subjects with cystic fibrosis (CF). In keeping with the fact that this was the first clinical test of any AAV vector, the starting clinical trial dose with the AAV2-CF vector was at a low level of 6×10⁴ DNase-resistant particles or viral genomes (VG). In this issue, Flotte and colleagues (2011) report on a phase 2 trial for another lung disease, α₁-antitrypsin (AAT) deficiency, in which an rAAV1-AAT vector was administered at doses that are 10 orders of magnitude higher, and indeed are the highest dose administered to date in any AAV vector clinical trial. This trial is particularly noteworthy and exemplifies how several AAV vector technology advances can be harnessed in stepwise fashion to address a clinical indication, which likely requires a high vector dose.

The AAV vectors for the first CF trial, and other early clinical trials, used an AAV2 capsid and were manufactured in a rudimentary process using DNA transfection of human cell lines for the upstream production, and downstream purification relying largely on ultracentrifugation in CsCl density gradients. The technology of rAAV now has advanced remarkably, leading to an enormous expansion of rAAV clinical trials for many clinical indications, using multiple routes of delivery. Observations in these clinical trials are, in turn, driving further advances in AAV vector technology (summarized by Wright, 2011).

The advances in AAV vectors reflect two main themes: modification of the capsid and improved manufacturing processes. Vectors using the AAV2 capsid have now provided clinical proof of concept in retinal degenerative disease, but they target many tissue types inefficiently and may be particularly susceptible to host immune responses because type 2 is the most ubiquitous AAV in humans. Progress to overcome these hurdles is focused on using capsids from many new AAV serotypes or clades, as well as synthetically modified capsids, and these are now being introduced into clinical trials. More sophisticated, scalable manufacturing methods are being developed to enhance vector quality, increase production yields, and permit much higher doses. Thus, downstream purification by orthogonal processes involving various types of chromatography and filtration is substantially improving the quality of vectors. DNA transfection for the upstream process is still used and has been substantially improved, but it is limited in scalability and eventual commercial application. Instead, a number of cell line-based or hybrid virus-mediated upstream production systems have been developed that are much more readily scalable and generally yield at least 10-fold more vector per unit biomass than do transfection procedures.

α₁-Antitrypsin (SERPINA1), or AAT, is a 52-kDa protein that is synthesized and secreted from hepatocytes, and functions in the lung to prevent the neutrophil elastase from degrading alveolar wall components. AAT deficiency (hereditary emphysema), due to mutations in AAT, causes chronic obstructive lung disease and is the second most common monogenic lung disease with a carrier frequency of 5% in North America. The disease can be treated by protein replacement therapy with human AAT (Prolastin), but this requires weekly intravenous infusion. An AAV vector-mediated gene therapy might deliver a prolonged and constant level of the protein, but this is a substantial challenge for gene therapy because AAT is one of the most abundant human serum proteins with a steady state level of 1000–2500 μg/ml.

The therapeutic target serum level of AAT for humans is about 11 μM (570 μg/ml), but this level of expression may require high vector doses as suggested by animal studies. An rAAV2-human AAT vector using the strong chicken β-actin transcription promoter and administered to rodents, either intramuscularly or by portal vein, did yield sustained expression of human AAT at therapeutic levels of 400–800 μg/ml; however, the dose of vector required was at least 10¹³ VG per mouse, which may imply scaling to high doses for human application.

Flotte and colleagues have conducted a series of clinical trials in subjects with AAT deficiency, with AAV vectors administered by intramuscular injection. The subjects with AAT deficiency enrolled in these trials generally express the common Z (missense) mutant form of AAT, which can be distinguished from the vector expressing wild-type M-AAT by assay of serum. They first tested the rAAV2-AAT vector in a phase 1 trial in 12 adults with AAT deficiency (Brantly et al., 2006). There were no vector-related serious adverse events and anti-AAV2 capsid antibodies rose after administration, but no T cell responses to AAV capsid were detected. Even at the highest dose of vector administered (6×10¹³ VG per patient), detection of any M-AAT expression was ambiguous at best. These investigators then conducted a similar phase 1 trial (Brantly et al., 2009) using the AAV-AAT vector pseudotyped in an AAV1 capsid, which, in animal models, generally gives much higher expression after intramuscular injection than an AAV2 vector. As expected, anti-AAV1 capsid antibody responses were observed. In addition, AAV1 capsid-specific T cell responses were detected in interferon (IFN)-γ enzyme-linked immunospot (ELISPOT) assays, although the subjects were negative for such responses before vector delivery. In spite of these anti-capsid T cell responses, sustained expression of AAT was detected in serum for up to 1 year after vector administration. Thus, switching to an AAV1 capsid did lead to detectable expression; however, at the highest dose (6×10¹³ VG per patient, or 8×10¹¹ VG/kg), the level of M-AAT in serum was about 2.6 μg/ml (50 nM), which was still more than 200-fold below the target therapeutic level. Given that there was detectable AAT expression, it was logical to ask whether an increase in vector dose could provide elevated expression.

In this issue, Flotte and colleagues report interim results of a phase 2 trial with the same rAAV1-AAT vector at higher doses in nine subjects with AAT deficiency. The vector was injected intramuscularly on a single occasion at a dose of 6×10¹¹, 1.9×10¹², or 6×10¹² VG/kg (n=3 subjects per dose). The highest dose is an increase of 8-fold over the previous trial and represents a total dose per patient of about 3.3–4.3×10¹⁴ VG. Thus far, the subjects have been monitored for 90 days, and there are a number of important observations.

What was the outcome of the increased dose? Importantly, there was a clear linear dose–response relationship between vector particle number and protein expression, and this appears to be the first such demonstration in gene therapy trials. Encouragingly, at the highest dose there was about a 10-fold increase in M-AAT level in serum to 21–36 μg/ml (412–694 nM); however, this is still some 20-fold below the therapeutic target level.

Were there any additional safety issues? The AAV vector doses administered are the highest of any clinical trial, but the observations continue the trend of the generally benign safety profile of these vectors. The most frequent adverse events were mild injection site reactions in eight of the nine subjects, and there were no severe adverse advents. As before, all subjects developed neutralizing antibodies to the AAV1 capsid and ELISPOT T cell responses to AAV1 capsid peptides. There were no antibody responses to AAT. In one mid-dose subject there was a T cell response to an AAT peptide, which was far distant from the missense Z mutation, but no evidence of untoward clinical effects. There were dose-dependent, transient elevations in serum creatine kinase that resolved and no other changes in clinical or hematology parameters. Histology of muscle biopsies at 90 days revealed some evidence of inflammatory infiltrates but no clinical symptoms of myositis.

Overall, this phase 2 trial supports the safety and feasibility of gene therapy for AAT deficiency; however, the design and/or delivery of the vector must be enhanced. The current trial results provide a rational starting point, but what can be done moving forward? There are a number of alternative strategies. First, we should consider two noteworthy elements of the production and administration of the vector used in this trial.

Transfection production processes are quite inadequate to generate the amounts of vector needed for the dose levels in the phase 2 trial, so a more readily scalable system was required. The authors chose to use a system in which baby hamster kidney (BHK) cells are coinfected in suspension with two replication-defective herpes simplex viruses (HSVs) containing, respectively, the AAV vector genome cassette and the complementing AAV rep2 and cap1 genes (Clement et al., 2009). Also, the downstream purification employed anion-exchange chromatography and affinity chromatography. In this system, 100 liters of crude lysate provided 10¹⁶ VG with a product recovery of 23% during purification. This is the first time that vector produced in this system has been used in a clinical trial, and an important feature in the product characterization was demonstration of absence of detectable replication-competent HSV.

The vector was used at a concentration of 5×10¹² VG/ml, and this is about the maximal concentration feasible because AAV vectors tend to aggregate at higher concentrations. Consequently, to administer the highest dose, the vector was injected at 100 sites over 10 muscle groups, with 1.35 ml injected at each site. This, in turn, required conscious sedation of the subjects with intravenous midazolam.

How might delivery be enhanced? Even though the production system can likely be scaled further, additional increases in dose simply by the intramuscular injection method used to date probably is not feasible. For instance, simply increasing the dose by 10- to 20-fold may require administration of more than 1 liter of vector. The authors note that in animal models, regional intravenous delivery may achieve substantially higher serum concentrations than intramuscular delivery.

How might vector design be enhanced? Additional modification of the vector capsid is possible and other serotypes such as AAV8 or AAV9 have also been shown to be highly efficient in muscle; however, it is not clear whether any of them increase inefficiency by another order of magnitude. An alternative strategy may be to modify both capsid and route of delivery. In this respect, preliminary results from Nathwani and colleagues (2010) with an AAV vector in subjects with hemophilia B may be highly instructive. This study is using a self-complementary AAV8 (scAAV8) vector to target the liver via peripheral intravenous delivery to express factor IX at sustained therapeutic levels. In addition, this is being achieved at substantially lower vector doses than those used in the phase 2 AAT trial. It is possible that a similar approach might be useful for AAT deficiency, although the level of AAT required is much higher than for factor IX and the current CBD-AAT expression cassette may be too large for an scAAV vector.

All in all, Flotte and colleagues have made a remarkable and sustained effort to develop a gene therapy for AAT deficiency. Their efforts have now led to a clear definition of how far they have advanced along the road and how far they may yet need to go. This provides a solid base from which judicious modifications of vector design and delivery that could increase efficiency of expression by one order of magnitude may place the target therapeutic level for AAT tantalizingly within reach.