Eliminating Panglossian thinking in development of AAV therapeutics

Main text

The US Food and Drug Administration (FDA) held a 2-day meeting (September 2–3, 2021) of the Cellular, Tissue and Gene Therapies Advisory Committee to consider toxicity risks of adeno-associated virus (AAV) vector-associated gene therapy products including a review of oncogenicity risks due to vector genome integration and safety issues identified during preclinical and/or clinical evaluation.

Great progress has been made with AAV vectors for hemophilia gene therapy in recent years and several late-stage clinical trials are close to completion with licensing decisions on the horizon. Recent events prompting the convening of this important FDA meeting serve as an urgent reminder that, while gene therapy is under clinical investigation, we need additional research in specified areas. Here, we renew the call for transparency from all those pursuing clinical trials in areas where information is available and for additional meaningful research on the unanswered questions around safety, variability, and durability of response (https://www.regulations.gov/comment/FDA-2021-N-0651-0009, last accessed 27 September 2021).¹ While the meeting highlighted and brought focus to several recent clinical events and potential toxicities, many questions and clear consensus on actions to be taken remain unanswered. We offer the following perspective on several open questions.

AAV gene therapy is currently the furthest advanced and the most promising curative modality for treatment of monogenic diseases such as hemophilia. However, first-generation vectors employed today are extraordinarily inefficient, with doses employed to achieve therapeutic results that are logs greater than the total number of cells within the human body and logs greater than viral loads ever encountered in the course of natural viral infections.

AAV is a virus and, like all other viruses, is regarded as a foreign invader by human host defenses, which results in activation of innate and adaptive immune responses.^2,3 Host immunity is dictated by genetics, prior exposure to similar or related antigens (AAV serotypes), and dose. Dose is the most critical, as an increasing incidence of adverse events and toxicities related to the host response is appearing as vector doses have increased. Since most AAV is sequestered in the liver following systemic dosing, liver toxicity, both acute and sub-acute, remains a significant concern with AAV dosing at E13 vg/kg and above. Targeting non-hepatic tissues with these liver-tropic vectors necessitates doses in the E14 vg/kg range and above, creating additional risk of toxicities. Furthermore, some groups that do not remove empty capsids see toxicity in the E12–E13 range, suggesting that empty capsids contribute to the immune response, as would be anticipated. Thus, there are consequences to dosing these levels of a virus.⁴

Toxicity to the dorsal root ganglia has emerged in nonclinical non-human primate (NHP) models, also dose dependent, with damage seen in the range of E13–14 vg/kg.⁵ Clinical manifestations have been largely undetected, with one possible exception.⁶ However, finding histopathologic damage is of concern. The axonal pathology may manifest as subclinical findings, but the long-term clinical implications of these findings are unknown. While it is hypothesized to be related to transgene overexpression and not capsid toxicity, this has not been well established.

Oncogenicity has so far been only a theoretical risk in humans, but preclinical studies continue to show that it is possible (https://investors.biomarin.com/2021-09-06-U-S-FDA-Placed-a-Clinical-Hold-on-BMN-307-Phearless-Phase-1-2-Gene-Therapy-Study-in-Adults-with-PKU-Based-on-Interim-Pre-clinical-Study-Findings). Contrary to the stubborn misconception, AAV vectors do integrate at a rate resulting in tens of millions of integrations with the currently used vector doses. AAV integration can trigger hepatocellular carcinoma (HCC) in mouse neonates and adults with preexisting liver damage.⁷ None of the dogs with hemophilia A followed up for 10 years after AAV dosing developed HCC, but most dogs had multiple integrations in or near cancer-associated genes, one of which (Dleu2) is also involved in HBV-associated HCC.⁸ Evidence of clonal expansion was seen in two dogs. Fittingly, HBV is also a small DNA virus that targets the liver and integrates. It is well recognized that HBV-related HCC develops decades after infections, with 20% of cases occurring without cirrhosis.⁹ This underscores the potential of viral integrations to drive malignant transformation in pernicious ways and warrants life-long follow-up of patients receiving AAV vectors and application of novel genomic methods in vector integration analyses. Registries, such as the World Federation of Hemophilia Gene Therapy Registry will endeavor to capture the majority of patients treated with hemophilia gene therapy to maximize chances of earlier detection of low incidence adverse events, such as HCC.¹⁰ Part of the current zeitgeist in the field is not taking the oncogenicity risk too seriously until we find a smoking gun. However, without being able to longitudinally follow vector rearrangements and integrations, some of which may be hit-and-run events, we may miss a smoking gun. Conversely, ignoring this risk altogether because it seems very small and remote seems like brinkmanship in the case of diseases with relatively long life expectancy and effective available treatments, such as hemophilia. The risk-benefit profile may well differ for other diseases and conditions where the prospect of early morbidity or mortality is well established and efficacious treatments are lacking.

While the above risks remain largely nonclinical, elevated liver transaminases are a real concern because they are the most common adverse event, which occurs in all hemophilia gene therapy trials, and signifies underlying liver toxicity. In hemophilia trials, these elevations are generally considered mild and transient, 1- to 3-fold above the upper limit of normal, but in some participants they rose as high as 10–20 times over the upper limit of normal or persisted for months to over a year. Our understanding of the mechanism of these elevations is still limited and hindered by differences between FIX and FVIII gene therapy trials. In some cases, they correlate with transgene expression loss and cytotoxic T cell response,² but this correlation is weaker in FVIII trials. Elevations variably respond to immunosuppression, with less predictable responses observed in FVIII trials, suggesting additional causes of hepatocyte damage, such as ER stress and/or unfolded protein response to FVIII expression (or overexpression) in hepatocytes, which normally is synthesized in the sinusoidal endothelium. Our limited understanding of these observations points to how little we know about what happens between vector entry into the target cell and the release of the transgenic protein.⁵ Events in between are largely a black box. With escalating AAV doses and contaminating empty capsids, this black box may lead to toxicity and death, thus it is essential to understand (https://www.biospace.com/article/patient-dies-following-treatment-with-astellas-pharma-s-experimental-gene-therapy/).11, 12, 13

There has been a call by some in industry and industry organizations to ignore rodent and NHP toxicities and march ahead with clinical studies. This is foolhardy and ignores the advances in the drug development paradigm over the past 50 years, which relies upon animal data to support advancement of experimental therapeutics into the clinic. Many experimental products have been canceled based upon preclinical studies or early clinical studies magnifying toxicities found preclinically. Should these findings be ignored and should humans be the relevant and primary test case for adverse events resulting from a new experimental therapy?

Undoubtedly some questions will not be fully answered in the preclinical period; however, where there are known areas of concern, we should not simply adopt a wait-and-see approach. Life-long registries¹¹ and post-vector infusion surveillance protocols will be important and essential tools but cannot be a replacement for investigations that could and should be done at an earlier stage. Accelerated approval pathways being utilized by many new and innovative therapies, including hemophilia gene therapy, are designed to allow for earlier approval of drugs that treat serious conditions and that fill an unmet medical need based on a surrogate endpoint. Under this pathway, there generally will be fewer, smaller, or shorter clinical trials than is typical for a drug receiving traditional approval, which may mean there is less information about the occurrence of rare or delayed adverse events. This tension between rapid approval and unknown future events requires careful consideration and balance. Full transparency of results is essential, but is not consistently practiced in this industry, where negative results and adverse events may not be published, and trials may be presented in a Panglossian fashion.

Recommendations

• •Study mitigation of the host immune response to AAV in clinical trials
• •Research on causes and mitigants of variability, and lack of predictability and durability following dosing
• •Increase efforts at liver de-targeting for vectors that target other tissues or organ systems in an effort to decrease overall systemic genome dose
• •Be fully transparent on number of empty AAV capsids, ratio of empty to full capsids, and adventitious agents. Minimize empty capsids in clinical preparations
• •Incorporate molecular analyses of DNA quality and quantity into product release as a part of quality control, and share transparently
• •Standardize AAV vector and anti-AAV antibody titer assays
• •Support life-long follow-up via registries: regulators, industry, health care providers, and patients