In this issue of Blood, Konkle et al report that 7 of 8 participants in a phase 1/2 trial of adeno-associated virus (AAV) vector (BAX335) for factor IX (FIX)-Padua gene transfer in patients with hemophilia B did not maintain expression despite steroid intervention, which the authors hypothesize is a result of the innate immune stimulatory effect of CpG motifs enriched within their vector cassette.1 Their study demonstrates that the cellular immune response to AAV vectors does not always respond to steroids and provides insight into mechanisms that may contribute to the AAV immune response with implications for the future design of AAV vectors.
Safe, reliable, and durable factor expression that can ameliorate or eliminate bleeding is the holy grail of hemophilia gene therapy. The field has converged around the use of intravascular delivery of AAV vectors to target hepatocyte expression of FVIII or FIX. After more than 2 decades of clinical trials, hemophilia gene therapy has demonstrated repeated proof-of-concept success. Although achieving the ultimate goal of hemophilia gene transfer seems to be within striking distance, there are outstanding questions in need of answers to actualize its widespread potential clinical benefit. The precedent for unexpected outcomes in hemophilia gene transfer clinical trials began with the first intravascular delivery of an AAV vector in which participants surprisingly only transiently expressed FIX.2 Subsequent studies demonstrated that the loss of FIX expression may be explained by a vector dose-dependent CD8+ cellular immune response to MCH-I–presented AAV capsid peptides on transduced cells (ie, a capsid immune response).3 The next clinical trial iteration incorporated the use of oral steroid administration to rescue FIX expression in the setting of a capsid immune response, which resulted in the first successful hemophilia gene therapy trial.4 The use of steroids to treat a capsid immune response was reproduced by a subsequent hemophilia B trial.5
Concurrently, 3 hemophilia B trials (NCT02618915, NCT01620801, and NCT01687608 reviewed in Samelson-Jones and Arruda6 ), including the first published observation outlined here, had only transient transgene expression despite steroid administration, which demonstrated that the cellular immune response to AAV is not universally abrogated by steroids. Human immune toxicity to AAV is multifactorial, incompletely understood, and inadequately replicated by available animal models, which precludes comprehensive preclinical study. Nevertheless, available information suggests that toll-like receptor 9 (TLR9) innate immune system sensing of AAV is required for the activation of CD8+ cellular immunity. TLRs are responsible for recognizing pathogen-associated molecular patterns of microbial pathogens and stimulating innate and adaptive immunity to the offending pathogen. Preclinical investigation supports that vectors deplete of CpG motifs (that when unmethylated, trigger TLR9 innate immunity) may minimize or circumvent an AAV capsid immune response.7 Although the capsid immune response has not presented a major safety concern in hemophilia, use of systemic AAV vectors in other disease states that use approximate log-fold or greater vector doses suggest that the capsid immune response is capable of marked hepatotoxicity.8,9 Collectively, efficacy limitations and potential safety concerns of immunoreactivity toward the viral capsid, incomplete mechanistic understanding of AAV immune toxicity, and the lack of adequate animal models em phasize that thoughtful translational study of clinical trial observations is required to rationally improve AAV vector design and clinical outcomes.
Even though the trial described by Konkle et al ultimately did not progress beyond phase 1/2 clinical development, data from the small cohort of enrolled participants supported ongoing clinical trial work and generated hypotheses worthy of additional exploration. The investigators were the first to therapeutically exploit FIX-Padua, which is a naturally occurring missense mutation resulting in a five- to tenfold increase in specific activity relative to the wild-type protein.10 Despite transient expression, the lack of an immunologic response to FIX-Padua supported the safety of this approach for hemophilia B gene therapy. The use of the FIX-Padua mutation has universally been adapted to all currently enrolling hemophilia B gene therapy trials. In addition, the investigators generated initial data to support their hypothesis that their CpG-enriched vector may, in part, explain the steroid-resistant capsid immune response observed in 7 of 8 participants. Comparison of vectors with varying CpG content demonstrated a decreased humoral immune response in mice treated with a CpG-deplete vector. However, a quantitative relationship between the observed humoral and cellular immune response to AAV in humans has not been described; thus, the clinical implications of these murine studies remain unclear. Whole-exome sequencing of the single participant who had sustained FIX-Padua expression demonstrated a heterozygous interleukin-6 receptor (IL-6R) missense variant that is predicted to disrupt function, which suggests the hypothesis that relative haploinsufficiency of IL-6R function may help explain the patients’ diminished immune response. This working hypothesis outlines a series of future investigations that may be undertaken to improve understanding of AAV immune toxicity, which may be fertile for clinical translation, given the availability of approved anti-IL-6R targeted therapy agents (eg, tocilizumab).
Finally, the authors’ AAV integration analysis of the single patient who developed a tonsillar carcinoma after receiving the vector set a precedent for performing integration studies on de novo tumor development after AAV gene transfer. Although AAV vectors are predominantly nonintegrating, hepatocellular carcinoma (HCC) genotoxicity after AAV integration has been identified as a theoretical risk of AAV gene transfer. The prevalence of iatrogenic acquired hepatitis C virus, an established risk factor for HCC development in the hemophilia population, may suggest that hemophilia is a provocative model for studying theoretical HCC risk after AAV-mediated gene transfer. This underscores the importance of integration analysis if HCC occurs after AAV gene transfer in a patient with hemophilia.
The outlined observations and hypothesis generated by Konkle et al highlight the importance of thoughtful analysis of clinical trial data, irrespective of clinical success. There are a remarkable number of hemophilia gene therapy trials underway, including licensing studies. Realizing the potential of gene therapy to alter the paradigm of hemophilia care will undoubtedly be predicated on a commitment to investigate and publish unexpected clinical trial observations.
Conflict-of-interest disclosure: L.A.G. has been a consultant for Pfizer and Bayer, serves on the data monitoring committee of Avrobio, and is a clinical investigator for phase 1/2 hemophilia A and B gene therapy trials sponsored by Spark Therapeutics and Pfizer, respectively.