Viruses may affect human health by altering host immunity and the genome, contributing to chronic disease and cancer. The purpose of this study was to examine evidence of oncogenicity accompanying integrations of non-hepatitis viruses into the liver cancer genome through analysis of a cohort of 147 Thai liver cancer patients comprised of 47 hepatocellular carcinoma (HCC) cases and 100 intrahepatic cholangiocarcinoma (iCCA) cases. For each case, we generated paired RNASeq data from cancerous tissue and adjacent non-cancerous tissue. Initial analysis with PathSeq comparing viruses with RefSeq genomes revealed that only Adeno-associated viruses (AAV) were differentially and predominantly present in the non-cancer samples (p < 9.4e-6, two-sided Fisher test). However, J. Zucman-Rossi and colleagues (Nault et al., Nat. Genet. 2015; 47:1187-1193) reported that integrations of AAV resembling the strains AAV2 or AAV13 near human genes such as CCNA2, CCNE1 are oncogenic. Understanding when AAV integration is oncogenic is important because AAV2-derived vectors are used in gene therapy. To investigate the apparent discrepancy, we performed AAV-targeted viral capture DNA sequencing of the cancerous tissue and adjacent non-cancerous tissue from each patient with AAV. One patient has an integration of AAV2/13 into CCNA2, expanded in the non-cancer sample and further expanded in the cancer sample that has 10,006 chimeric AAV/CCNA2 DNA reads. We validated that this integration substantially increased mRNA expression of CCNA2, replicating the previous findings. In contrast, we found ten patients with AAV integrations into other loci only in the non-cancer samples and two other patients with AAV integrations into other loci only in the cancer samples. These single-sample AAV integrations did not significantly increase the expression of the target gene, if any, suggesting that they are not oncogenic. All previously reported AAV integrations have been demonstrated only in HCC. Similarly, in our analysis, the only oncogenic integration was in a patient with HCC. Among the ten non-oncogenic integrations in non-cancer samples, one is HCC and nine are iCCA. Both non-oncogenic integrations in cancer samples were in iCCA. Analysis of all previously published AAV integrations and the new AAV integrations we identified showed that almost all the integrations reported as oncogenic overlap the poorly characterized X gene, while the non-oncogenic integrations need not overlap the X gene. This finding explains why the oncogenic AAV integrations are necessarily from AAV2 or a similar strain since other sequenced AAV strains do not contain the X gene. This would suggest re-engineering AAV vectors to make the X gene non-functional, to retain the useful properties of these vectors and concomitantly mitigate the risk of liver cancer in AAV-therapy recipients.

Citation Format: Alejandro A. Schaffer, Dana Dominguez, E. M. Gertz, Lesley M. Chapman, Anuradha Budhu, Marshonna Forgues, Jittiporn Chaisaingmongkol, Siritida Rabibhadana, Benjarath Pupacdi, Xiaolin Wu, Curtis C. Harris, Mathuros Ruchirawat, Eytan Ruppin, Xin W. Wang. Oncogenicity of adeno-associated virus (AAV) integration into the genomes of liver cancer patients depends on the overlap between the AAV integration site and the AAV X gene [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2658.