General Updates | Understanding potential malignancy risk following AAV gene therapy in hemophilia

The publication was motivated by a common oversimplification in the gene therapy field, which is that adeno-associated virus, or AAV vectors, are often described as non-integrating. Now, it is true that most vector genomes remain in the cell as episomes rather than integrating into chromosomes in the way that retroviral vectors do. However, AAV vectors can integrate into the human genome at a low but non-negligible frequency. Now, estimates range from approximately 0.5 to 16 integration sites per 1,000 cells, and that may sound like a very small number, but at the doses used clinically, again, it can translate into millions of individual integration events in one patient. So, the concern is that, in principle, an integration event could occur near a gene involved in cell growth or cancer development and thus trigger and drive tumor genesis. We know that this is mechanistically possible because AAV-related hepatocellular carcinoma has been observed in mouse models, particularly when vectors were administered early in life or in the setting of liver injury. The biology in mice is not directly transferable to humans, and other animal models have not shown the same clear signal. And at the time of the review, when I was writing it, no human case of malignancy caused by AAV integration was known.

So the central question in the paper was, what have we actually learned from cancers reported in people who received clinical AAV gene therapy? And at the time of the review, there had been 10 reported malignancies, eight in people treated for hemophilia and two in children treated for spinal muscular atrophy. Importantly, none of those cases provided evidence that AAV integration had driven the cancer. But an equally important message is that the strength of that conclusion varied considerably from case to case. The clearest example of a strong analysis was the hepatocellular carcinoma reported after etranacogene dezaparvovec gene therapy for hemophilia B. The patient was 69 years old and already had several established risk factors for liver cancer, including previous hepatitis B and C infections and metabolic dysfunction-associated steatotic liver disease. High-quality tumor material allowed investigators to perform a very comprehensive molecular assessment. They examined vector copy numbers, mapped integration sites, looked for clonal expansion. They even performed whole genome and RNA sequencing and assessed gene expression patterns. And vector integrations were found in both tumor and non-cancerous tissue, but not at sites known to drive hepatocellular carcinoma. And there was no evidence that a single integrated clone had expanded to form the tumor. In contrast, the tumor contained typical genetic and expression changes commonly seen in hepatocellular carcinoma. So taken together, that provided a very strong evidence that AAV integration was unlikely to have caused that cancer.

At the other side of the spectrum was a spinal cord tumor reported after AAV gene therapy for spinal muscular atrophy in one of the two children. In that case, the amount and quality of DNA recovered from the tumor were inadequate. The analysis probably underestimated the number of integrations, and there was not enough material to validate the findings using complementary methods. And, well, therefore, the absence of a demonstrated relationship could not be considered nearly as conclusive as in the other case, the case of the HCC. And that distinction is one of the most important messages of the publication. Saying that a cancer was unlikely to be related to gene therapy can mean very different things. It may mean that a high-quality multidimensional analysis found strong evidence for an unrelated biological cause. Alternatively, it may mean only that the available sample was insufficient to provide a meaningful answer, and we should communicate those situations differently so well the paper therefore makes several practical recommendations when a malignancy occurs there needs to be rapid communication among the treating physician pathologist gene therapy center and manufacturers so that appropriate tissue can be collected without interfering with cancer care. Whenever possible, the assessment should include adequate tumor tissue, matched non-cancerous tissue, and several complementary analytical methods, not simply a single assay for the presence or absence of vector DNA. We also need to interpret every case in its clinical context. People with hemophilia, particularly older individuals, may already have an elevated risk of hepatocellular carcinoma, which was the case in that individual I described earlier. And there is a high burden of historical hepatitis B or C infection in the population. And metabolic liver disease is also increasingly common, both common both in the general population as we know and in people with hemophilia so those background risks can make causality assessment particularly challenging.

My overall conclusion is reassuring but cautious, we currently do not have evidence from the cases reviewed that AV integration caused malignancy in people with hemophilia. However, that does not establish that the risk is zero. And importantly, just two months ago, a report came out in the New England Journal showing evidence that AAV integration drove a central nervous system tumor in a child with Hurler syndrome. That’s the smoking gun case if you will because it establishes that a remediated tumor genesis is mechanistically possible in humans so that risk is no longer purely theoretical and because these cancers may be very rare and may take years or even decades to emerge, we need standardized molecular investigations, transparent reporting, and coordinated global long-term follow-up through mechanisms such as the World Federation of Hemophilia Gene Therapy Registry. The aim is not to create unnecessary alarm. It is to make sure that when a cancer does occur after gene therapy, we collect the right material, perform the right analyses, and communicate clearly how certain or uncertain we are about the relationship to treatment.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.