Encouraging and Unsettling Findings in Long-Term Follow-up of AAV Gene Transfer

Main Text

Gene therapy for genetic disease requires long-term therapeutic gene expression. Adeno-associated viral (AAV) vectors are considered ideal for this task when administered to non-dividing target cells. The value of these therapies becomes obvious during long-term follow-up. A study published last year showed that substantial improvements in the vision of patients that received ocular AAV gene transfer for correction of Leber’s congenital amaurosis, a rare form of inherited blindness, were stable for at least 4 years.¹ However, very recent results raise new questions and concerns about long-term outcomes of AAV gene transfer. As noted before, two recent clinical trials with improved vectors found therapeutic, if not curative, factor IX (FIX) expression that was stably maintained for at least 3 years in patients with hemophilia B.² A 3-year follow-up of hepatic gene transfer, following intravenous high-dose delivery of a serotype 5 vector expressing factor VIII, has been encouraging.³ At 6 × 10¹³ vector genomes/kg, average FVIII activity levels were in the normal rage. These levels were stable during the first year but declined gradually after that to between ∼50% and ∼30% of normal, depending on the assay used (n = 7). Remarkably, median annualized bleeding rates and the median use of recombinant FVIII was reduced to 0, and bleeding into target joints resolved. Hopefully, patients will not experience further decline of their factor levels. Nonetheless, the decrease in systemic transgene expression after the first year differs from the experience with FIX and may relate to the high vector dose or the nature of the transgene product. It is known that FVIII can induce cellular stress responses. At the highest vector dose, mild liver toxicity was observed during the first months after gene transfer, and patients received a course of steroid treatment. Whether these early events relate to the later decline in expression is currently unknown.

At the recent annual meeting of the American Society of Hematology in Orlando, FL, in December of last year, Sabatino and colleagues presented results from a long-term follow-up (up to 10 years) in 9 hemophilia A dogs that had received AAV8 gene transfer at doses up to 4 × 10¹³/kg. The animals expressed up to ∼10% of normal, experienced a robust reduction in bleeding rates, and had no evidence for toxicity. Very surprisingly, however, 2 dogs showed steady increases in their factor levels starting at approximately 3 years into the study. Integration site analysis yielded evidence for clonal expansion of integrated vector sequences. While integration sites were distributed across the canine genome, integrations near genes that potentially affect growth control or transformation were found. However, there was no evidence of tumor formation or liver pathology. It has long been debated how the largely (but not exclusively) episomal AAV vector genomes can be so persistent in hepatocytes, which are expected to slowly turn over. Increased expression suggests expansion of transduced hepatocytes relative to non-transduced cells. Interestingly one of the two dogs had been treated with dual vectors, one expressing the light and the other the heavy chain of FVIII. Since the two chains have to assemble prior to secretion from the hepatocyte to yield a functional molecule, one would predict expansion of cells that had integrated copies of both vectors. Expansion of transduced hepatocytes may not necessarily have been caused by vector integration. One could envision a scenario in which certain hepatocytes had a growth advantage in an aging liver or due to other causes and coincidently harbored integrated vector genomes. The counter argument would be that there should have then also been examples of a decline in FVIII expression in other animals due to growth advantages of hepatocytes that lacked integration of transgene since FVIII protein is not expected to provide an advantage to hepatocytes. Should gene therapy patients with FVIII expression in the normal range experience a further increase at later times, one may not only worry about the potential for transformation but also for increased thrombotic risk. However, it is curious that several other long-term studies of similar duration in hemophilic dogs and non-human primates, published previously in Molecular Therapy and Blood, did not find an increase in expression over time. At the same hematology meeting in December, Amit Nathwani summarized an 8-year follow-up of 10 hemophilia B patients. Here, FIX expression from an AAV8 vector was quite stable, with only minor changes over time. Thus, multiple questions arise how to reconcile these various findings and what the cause of the increase of expression was in the recent canine study.

A different concern about AAV gene therapy is the potential for immune responses, including CD8⁺ T cell responses. For example, patients with α₁-antitrypsin deficiency showed >5 years transgene expression after intramuscular gene transfer with an AAV1 vector. However, muscle biopsies also revealed persistent inflammation at the site of gene transfer that was evident even at a 5-year time point.⁴ These infiltrates included CD8⁺ T cells. T cell assays showed that these responses, in rare cases, were directed against a polymorphic sequence in the transgene product, but more typically were against the viral capsid. The effect of this immune response on transgene expression over time has been surprisingly modest, which in part may be explained by recruitment of regulatory CD4⁺ T cells (Tregs) with an immune suppressive phenotype. In a new study forthcoming in Molecular Therapy, Gernoux et al.⁵ show that, in non-human primates and patients, in addition to infiltration by Tregs, an additional mechanism is at play. While capsid-specific CD8⁺ T cells may persist for years, they express PD-1 and are hypo-responsive, exhibiting properties of exhausted T cells. The PD-1/PD-L1 pathway is a powerful negative immune checkpoint and thus a target in immunotherapy against cancer, which aims to restore T cell functionality. Interestingly, these results are reminiscent of earlier papers that showed induction of an exhausted phenotype in transgene product-specific CD8⁺ T cells upon AAV gene transfer to the skeletal muscle of mice. These papers revealed the limitations of using AAV as a vaccine carrier, as immune responses could not be boosted. Instead, the T cells eventually underwent programmed cell death (interestingly, more functional CD8⁺ T cell responses can be obtained when using AAV vectors with self-complementary instead of traditional single-stranded genomes). Gernoux et al.⁵ were able to restore the responsiveness of human capsid-specific T cells by blocking the checkpoint pathway. These results suggest that the T cells are not permanently inactivated and may be able to regain function under changing conditions. A limitation of the new study is a lack of more extensive phenotyping of the CD8⁺ T cells, as not only exhausted T cells may express PD-1. For instance, we found that non-functional CD8⁺ T cell responses in AAV gene transfer in mice may reflect expression of PD-1 by not necessarily other markers of exhaustion.⁶ Moreover, T cells lacking PD-1 may emerge at late time points, rendering the response functional. Therefore, it is at the same time encouraging and concerning that downregulated T cell responses may persist for years after AAV gene transfer. In conclusion, long-term studies on AAV gene transfer yield both reassuring and potentially troubling results that warrant yet further follow-up and analyses.