AAV Vectors: Are They Safe?

Dear Editor,

Ever since the very first human clinical trial with recombinant adeno-associated virus (AAV) serotype 2 (AAV2) vectors performed by Flotte et al. nearly a quarter of a century ago, ¹ which documented the safety of these vectors, it has generally been assumed that the nonpathogenic nature of the naturally occurring AAV is also retained in its recombinant counterpart. However, in view of the recently reported deaths of two patients in a gene therapy trial for X-linked myotubular myopathy using an AAV8 vector by Audentes Therapeutics, ² it is incumbent upon all of us in the gene therapy community in general, and AAV gene therapy in particular, to acknowledge the need for serious introspection.

While the preliminary investigations implicate sepsis as being the immediate cause of death, the use of the remarkably high vector dose of 3 × 10¹⁴ vg/kg must not be overlooked, since no serious adverse events were observed in any of the six patients who received a lower dose of the same vector.

Soon after taking the reins of the Human Gene Therapy family of journals, and just prior to the approval of the very first AAV as a drug, Luxturna, by the Food and Drug Administration (FDA) in the United States in December 2017, Terence Flotte and Guangping Gao, as Editor-in-Chief and Deputy Editor-in-Chief, respectively, wrote a prescient editorial aptly entitled “AAV Is Now a Medicine: We Had Better Get This Right.” ³ As they so elegantly pointed out the virtues of AAV as a vector having “just the right combination of long-term persistence of transgene expression and lack of pathogenicity,” they also highlighted “the potential risk of insertional mutagenesis” and “the potential for immune responses to both transgene products and AAV capsid epitopes.”

Although the potential risk of AAV vectors causing cancer has been the subject of intense debate in the past, ^{4 –10} more recent data are beginning to emerge from 10–15 years of follow-up studies in dogs ¹¹ and in humans, ¹² suggesting that such a risk is extremely low as long as extremely high vector doses are not used.

More worrisome is the risk of the host immune response to AAV vectors, especially when used at extremely high doses. Ironically, we have been aware of this issue for nearly a decade and a half, when the results of the first liver-directed gene therapy clinical trial for hemophilia B with the first generation of AAV2 vectors were reported. ¹³ It was abundantly clear from this trial that an adult human body of ∼70 kg could easily tolerate systemic infusion of AAV2 vectors at a dose of 4 × 10¹¹ vg/kg (a total of ∼28 trillion AAV2 particles), but at a fivefold higher dose of 2 × 10¹² vg/kg (a total of ∼140 trillion AAV2 particles), there was a CD8⁺ T-cell response against the vector capsid proteins, which led to the loss of hepatocytes that were expressing the therapeutic transgene.

What is puzzling is that in a more recent clinical trial for gene therapy of spinal muscular atrophy type 1 with AAV9 vectors, as high a dose as 2 × 10¹⁴ vg/kg (∼1 quadrillion AAV9 particles) was used in infants (average age 3.4 months; average weight 5.7 kg), and yet other than elevated levels of serum aminotransferase in four patients, which were successfully treated with prednisolone, no significant adverse effects have yet been reported (this AAV vector, Zolgensma, was approved by the FDA in May 2019).

These observations raise two important questions. First, are AAV9 vectors safer than AAV2 and AAV8 vectors? Second, are older children and adults more prone to an AAV vector-induced immune response than infants? In this context, it is also important to note that infection by the wild-type AAV2 occurs in infants and young children with no sequalae, ¹⁴ and that AAV vectors at a relatively low dose of 2 × 10¹¹ vg/mouse are only mildly immunogenic. ¹⁵

Whereas there is little doubt that answers to these questions will be forthcoming from additional clinical trials in the future, it is also important to keep in mind the following two points: (1) the naturally occurring AAV did not evolve to be used as a vector for the purposes of delivery of therapeutic genes, ¹⁶ and therefore, the first generation of recombinant AAV vectors composed of naturally occurring capsids are effectively targeted by the host immune system, and (2) an average adult human body contains only ∼3 × 10¹³ (∼30 trillion) cells. ¹⁷ In other words, it is quite a testament to the remarkable safety of even the first generation of AAV vectors that adults are easily able to tolerate roughly the same number of AAV particles as there are number of cells in the body. However, when the vector dose exceeds 100 trillion AAV particles, it is a reasonable expectation that an immune response would ensue, as was originally observed in the first hemophilia B trial nearly a decade and a half ago. ¹³

Although the use of prophylactic immune suppression is increasingly evident in most, if not all, clinical trials in which high AAV vector doses are administered, it should also be pointed out that this strategy is not necessarily without complications, as was recently editorialized by Yamada and Herzog. ¹⁸ Thus, to ensure safety in all future gene therapy clinical trials with the first generation of AAV serotype vectors in older children and adults, all attempts should be made not to exceed the total dose of 30 trillion AAV particles. ¹⁹ Efforts should also be made to develop further the next generation of AAV serotype vectors that are more efficient at significantly reduced vector doses, ^{20 –23} as well as less immunogenic, ²⁴ in addition to the use of AAV serotype vectors, when available, with selective tropism for specific human cells ^{25 –28} and organs, ^{29 –32} rather than extrapolating from data obtained with animal cells and organs.

So, the answer to my own question—are AAV vectors safe?—is an emphatic yes, as long as astronomically high doses are not administered, especially in older children and adults.

As for the Flotte and Gao editorial—AAV Is Now a Medicine: We Had Better Get This Right? ³ —I could not agree more.