Hemophilia Gene Therapy

Gene therapy for the X-linked bleeding disorder hemophilia B—factor IX (FIX) deficiency—by means of in vivo gene transfer with adeno-associated viral (AAV) vectors has been in clinical trials for the past 16 years, cycling between partial successes in the clinic and further development in the laboratory. At the recent World Congress of the International Society on Thrombosis and Haemostasis (Toronto, Canada, June 2015), Paul Monahan and colleagues presented data from a patient who has achieved FIX levels of 20–25% that have been sustained for more than 6 months since undergoing hepatic gene transfer with an AAV8 vector (administered via peripheral vein). This result, representing another milestone in gene therapy for hemophilia, was in part accomplished by incorporating a missense mutation into a codon-optimized FIX sequence, which improves FIX enzymatic activity by 5- to 10-fold. However, similarly treated patients lost therapeutic expression because of immune responses to virally transduced hepatocytes or for other, yet-to-be-determined reasons.

In the past decade, gene therapy for hemophilia B has focused on expression and secretion of FIX by hepatocytes into the circulation. Sustained correction from severe to mild hemophilia (>5% of normal coagulation activity) using an AAV serotype 8 vector has been documented in recent years. Further improvements so as to sustain levels of >10% would essentially represent a cure for most patients, preventing spontaneous bleeds and limiting the need for intravenous factor infusions to surgery and treatment of severe trauma. The ”Padua” mutation was discovered when analyzing a case of X-linked juvenile thrombophilia, and has since been rigorously tested for improved gene therapy in murine and canine models of hemophilia B. For the clinical trial sponsored by Baxalta and conducted by Monahan and colleagues at the University of North Carolina–Chapel Hill, an FIX expression cassette was incorporated into a self-complementary AAV (scAAV), a vector system initially developed by McCarty and Samulski to eliminate the need for second-strand synthesis, which limits traditional AAV vectors. Seven patients have been treated in the trial thus far, and the results vary widely.

It has been known for 10 years that the immune system may limit the duration of therapeutic gene expression from AAV vectors in the human liver. The discovery of a hepatotoxic CD8⁺ T-cell response against AAV capsid was surprising at the time because none of the animal models had shown anything similar. Several articles in Molecular Therapy have since uncovered differences between murine and human, and even human and nonhuman primate, T-cell responses to capsid. More recently, a murine model was developed that mimics transaminitis and loss of FIX expression after ex vivo expansion followed by adoptive transfer of capsid-specific CD8⁺ T cells. At the highest vector dose in the current trial by Monahan et al. (3 × 10¹² vector genomes/kg), both patients showed even higher levels of expression than those mentioned above (up to >50% of normal) but subsequently lost expression concomitant with transaminitis and a T-cell response to capsid. It remains remarkable that these responses may emerge two months after vector administration. No immune response against FIX (or the Padua variant) was found. In efforts to counter the destructive T cells, elevation of liver enzyme levels has previously been established as a biomarker that warrants initiation of immune suppression, and the steroid drug prednisone has been successfully applied to stop the T-cell response to AAV8 in its tracks and preserve FIX expression.

Although the same strategy was adopted in the trial by Monahan et al., efforts to preserve gene expression upon onset of transaminitis were unsuccessful. One drawback to this approach is that drug administration must be initiated very soon after the first signs of hepatotoxicity. Additionally, transaminitis may not be a sufficiently sensitive biomarker, and steroid drugs may not be effective against T-cell responses in all patients. Hence, the field continues to wrestle with the question of whether a prophylactic immunosuppression protocol should be incorporated into hepatic AAV gene transfer, and how such a regimen should be designed. Preclinical studies suggest a requirement for innate immune sensing of the AAV genome by Toll-like receptor 9 (TLR9) for CD8⁺ T-cell activation. Development of vectors devoid of TLR9-activating CpG motifs has been proposed. Using scAAV vectors may, on the one hand, increase or accelerate responses because of stronger TLR9 signaling, which, on the other hand, could be an advantage by providing a more defined target for immune suppression. No changes in liver enzymes or capsid-specific T cells were detected at a mid-dose of 1 × 10¹² vector genomes/kg, reinforcing the conclusion that the T-cell response is vector dose–dependent. However, differences in vector production, purification, design (such as promoter), and measurement of titers complicate a direct comparison between trials. Interestingly, the three patients treated with the mid-dose experienced very diverse outcomes.

As mentioned, one patient continues to express at curative levels of >20%, whereas the others showed therapeutic levels initially but then spontaneously lost expression in the absence of any evidence for an immune response. Although minor changes in persistence of gene transfer or expression may have been amplified by the highly active Padua mutation, this observation remains unsettling because it adds another layer of complexity. Hence, AAV gene transfer to the human liver is caught somewhere between a cure, cellular immune responses, and additional factors that have yet to be determined.

Roland W. Herzog