Revisiting the “New” Inflammatory Toxicities of Adeno-Associated Virus Vectors

In the earliest days of clinical translation of recombinant adeno-associated virus (rAAV) gene therapy, it was often said that rAAV vectors were relatively non-immunogenic. This assertion was based on the fact that rAAV vectors induced a significantly smaller innate immune cytokine response than first-generation adenovirus (rAd) vectors at similar doses. ^1,2 However, even from the earliest studies, it was obvious that rAAV gene transfer elicited neutralizing antibody responses to the AAV capsid, which limited the ability to re-dose rAAV vectors with the same serotype capsid. ³ When combined with the observation that rAAV vectors persist primarily in episomal form, the anti-capsid antibody proved to be an important limitation to rAAV-based gene therapy in rapidly dividing cells, such as in the airway epithelium. ⁴ However, in cells that are nondividing, the one-and-done nature of rAAV transduction made the anti-capsid antibody response less of a limitation. Hence, products such as Luxturna and Zolgensma, targeted to the retina and spinal cord consisting of terminally differentiated cell types, respectively, could be efficacious in spite of the development of anti-capsid antibodies after the first dose.

A series of human studies, most notably in patients with hemophilia B, subsequently showed that effector T-cell responses to rAAV capsid were also a common issue with rAAV gene therapy. ⁵ These anti-capsid T-cell responses could be effectively countered with fairly simple corticosteroid therapy. ⁶ Furthermore, certain dose–route–vector combinations elicit regulatory T-cell responses, which may naturally counter the anti-capsid effector T-cell response. ⁷ Depending on the disease being targeted by gene therapy, anti-transgene immunity has also occasionally been a limiting factor. Certain genetic diseases, such as alpha-1 antitrypsin deficiency, are commonly caused by missense mutations, rendering anti-transgene responses relatively rare. However, in other cases, such as with Pompe disease, anti-transgene product responses are quite common and may necessitate heightened immune suppressive regimens. ⁸

Beginning 2 years ago, with a series of studies in this journal and elsewhere, two “new” rAAV-related inflammatory toxicities have recently been observed. These include a syndrome of neuroinflammation in the spinal dorsal root ganglia after high-dose intrathecal administration and an acute syndrome of thrombocytopenia, hepatic and renal toxicity after high-dose intravenous administration, reminiscent of complement activation. Both syndromes were first described in nonhuman primates and neonatal piglets. ^{9 –11} Both have since been observed in a number of cases in patients. ¹² While the details of the clinical cases have yet to be published, the emergence of these two new forms of dose-limiting toxicity is affecting the clinical development of AAV gene therapies for central nervous system disorders.

The more complete picture of rAAV immune and inflammatory responses is depicted in Table 1. The implications of the first four forms of adaptive anti-vector immunity have been extensively studied, whereas the implications of the newer forms of toxicity have yet to be fully explored. While these toxicities may now present a dose-limiting toxicity, effectively setting an upper limit on intrathecal and intravenous dosing, it is not clear whether these limitations could be partially overcome with adjunctive transient anti-inflammatory or immune suppressive medications. It is also unclear whether methods of vector manufacturing and improvements in vector preparations, such as elimination of empty capsids and some other viral impurities, could improve the safety margin for higher vector doses. In any case, the rAAV gene therapy field will undoubtedly need to address these issues with further studies seeking to understand their mechanisms. Only after a deeper understanding of the mechanisms of high-dose toxicities has been obtained will optimal solutions for these problems be devised.