Molecular evaluation and vector integration analysis of HCC complicating AAV gene therapy for hemophilia B

TO THE EDITOR:

Genotoxicity remains an unknown safety concern of gene therapy. Molecular techniques for determining the frequency and genomic localization of vector integration are central to understanding primarily integrating viral vectors (ie, retrovirus and lentivirus). Unlike these vectors, recombinant adeno-associated virus (rAAV) vectors integrate into host genomes at low frequencies. Nevertheless, the integration of rAAV sequences in oncogenic hotspots could theoretically lead to hepatocellular carcinoma (HCC).¹ This report describes the molecular characterization of the first case of HCC complicating an rAAV gene therapy trial.

Studies using rAAV in mice reported low integration levels into host chromosomal sequences associated with HCC.^2,3 Investigations suggested that HCC was driven by microRNA-341 dysregulation within the Rian locus, a hotspot for mouse genome integration. The Rian locus in mice has a human ortholog, the human long-coding RNA, MEG8, which is overexpressed in some HCCs and may interact with microRNA-367-3p in the pathogenesis and progression of some HCCs.⁴ Nevertheless, the microRNA-341 locus found to be susceptible to rAAV insertional mutagenesis in mice has no human homolog.⁵ 

Studies using mouse models have previously been performed, including work in tumor-prone mouse species and under experimental conditions that stimulate cell proliferation, which suggests specific components of vector design or administration may affect vector integration and infer greater risk of tumorigenesis (eg, use of chicken β-actin or thyroxine-binding globulin enhancer/promoter, treatment in the neonatal period, partial hepatectomy, and vector DNA–related contaminants).^5-7 In studies specific to hemophilia gene therapy, rAAV–factor IX (FIX) vectors in mice, dogs, and nonhuman primates showed no evidence of vector integration leading to HCC up to 8 years after treatment.^8-10 Recently, human biorepository analysis identified HCC with the insertion of wild-type AAV sequences, potentially implicating wild-type AAV in some cases of HCC (although not rAAV, which does not carry any viral genes).^11,12 

At least 39 human trials have used rAAV to target the liver to correct primary liver diseases, lysosomal storage disorders, liver protein synthetic deficiencies, and other metabolic disorders.¹³ Seventeen studies of liver-directed rAAV hemophilia gene therapy have been published to date, comprising >320 patients followed up for 0.5 to 15 years.^14,15 Current clinical experience identifies no added oncogenic risk from liver-directed rAAV gene therapy. However, prior exposure to contaminating viruses in plasma-derived clotting factor concentrates leads to chronic hepatitis, meaning older patients with hemophilia have an increased incidence of HCC, which is a confounding factor.¹⁶ Considering the conflicting data linking AAV specifically with HCC, the development of HCC in a recipient of liver-directed rAAV gene therapy is a highly anticipated event.

We report, to the best of our knowledge, the first event of HCC in a recipient of rAAV gene therapy, including molecular characterization of the tumor to examine the biology of the event, generate 1 of the first human rAAV vector integration site (IS) profiles, and advance understanding of rAAV vector safety.

Etranacogene dezaparvovec, comprising a liver-directed rAAV5 vector containing a codon-optimized Padua-variant human FIX transgene and a liver-selective promoter, is the first systemically delivered AAV gene therapy approved to treat hemophilia B by the US Food and Drug Administration and the European Regulatory Authorities.

The phase 3 HOPE-B trial (NCT03569891) enrolled 54 adult males with hemophilia B (FIX ≤ 2%), including participants with prior hepatitis B virus (HBV; n = 9) and hepatitis C virus (HCV; n = 31) infection, and excluding participants with cirrhosis or advanced fibrosis.¹⁷ Participants received a single intravenous dose of etranacogene dezaparvovec (2 × 10¹³ genome copies per kg). Liver-specific assessments included twice-yearly α-fetoprotein measurements and annual ultrasound over 5 years follow-up. The trial was conducted in accordance with the International Council for Harmonisation Good Clinical Practice guidelines and ethical principles originating in the Declaration of Helsinki. The protocol was approved by institutional review boards and independent ethics committees at each trial site.

Molecular analyses were conducted by GeneWerk GMBH (Heidelberg, Germany) independently from the sponsor and included the following experiments:

• Vector copy number measurement using quantitative polymerase chain reaction.

• Vector DNA IS analysis by shearing extension primer tag selection/ligation-mediated polymerase chain reaction (triplicate analyses: 3 × 500 ng DNA input).^18,19 

• Whole-genome sequencing (WGS) via Illumina next-generation sequencing platform (singlicate analysis: 1 μg DNA input).

• RNA sequencing transcriptome profiling, RNA fusion products, and differential RNA expression.

Detailed methodology and results are provided in the supplemental Material. Final interpretation of the results was evaluated by independent consultants experienced in clinical hepatology, hematology, and genetics, and by the HOPE-B trial independent data monitoring committee.

The patient who developed HCC received etranacogene dezaparvovec in October 2019. Medical history, prior treatments, and baseline characteristics are presented in Table 1. The patient’s history of hepatitis virus infections and advanced age were HCC risk factors.

A chest computerized tomography with angiography including visualization of the upper abdominal organs and liver ultrasound performed 1 and 3 months after treatment revealed no liver abnormalities; α-fetoprotein levels remained within normal range. There were no notable changes in transgene (FIX) expression levels after the expected increase to steady-state.

The per-protocol, 1-year (day 365) postdose abdominal ultrasound revealed a subcapsular liver lesion. A subsequent abdominal computerized tomography revealed a single 3.1-cm lesion in segment 8. A needle biopsy (day 389) specimen revealed predominantly healthy liver tissue, with a single atypical focus consistent with HCC. Immunohistochemistry was positive for glutamine synthetase, HSP-70, and glypican-3, with increased capillarization (CD34). Histology of surrounding liver tissue showed steatosis grade 1, slight lobular infection, and periportal fibrosis (no certain bridging), consistent with nonalcoholic fatty liver disease.

The patient underwent exploratory laparotomy for surgical excision (day 443). Intraoperative ultrasound revealed a secondary liver lesion 0.8 cm in diameter in segment 2/3; biopsy confirmed a 90% likelihood of HCC. Specimens from the resected secondary lesion and adjacent healthy liver tissue were analyzed for histopathology and molecular integration. The primary tumor was not excised because of the complex location, possible morbidity, and minimal impact on the multifocal HCC prognosis. Histopathology and immunohistopathology confirmed the occurance of moderately differentiated HCC. Results from adjacent tissue were consistent with the observations from the preoperative biopsy specimen.

The patient underwent transarterial chemoembolization for the lesion in segment 8 with doxorubicin (55 mg once, intra-arterial) 5 weeks after surgery. The patient subsequently received orthotopic liver transplantation on day 851 after treatment (February 2022) with a single preoperative bolus of FIX concentrate. The patient developed progressive transplant failure because of hepatic artery thrombosis and, ultimately, had to undergo a second orthotopic liver transplant. Since then, the patient has been recovering well.

Molecular analysis of rAAV sequences in HCC and HCC-adjacent samples had 3.21 and 4.11 vector copies per cell, respectively. In both samples, many vector-vector fusion sequences were detected, demonstrating that most vector genomes were episomal, with ∼1 in 10 000 integrated. There were 56 unique ISs in the HCC and 39 in the HCC-adjacent sample; <0.03% of cells (∼60 of 250 000 cells) had AAV vector integration. In both samples, no dominant ISs were found; the highest shearing count detected was 2, indicating that the insertion occurred in only 2 cells. All the remaining IS detected had only 1 shearing count.

WGS detected a deletion on chromosome 8 (Table 2), which is present in 48% of HCCs arising outside of gene therapy.²⁰ Moreover, mutations within HCC-associated genes were apparent, including TP53, NFE2L2, and PTPRK. WGS and RNA sequencing of the HCC-adjacent sample revealed a premalignant genetic signature. Transcriptome profiling and analysis identified features and genes highly associated with HCC.

If HCC development was driven by AAV vector integration, frequent IS and a dominant IS would be expected; neither was observed in this patient’s samples. Furthermore, WGS would show integration in/near known HCC oncogenes (eg, TP53 and NFE2L2), which did not occur. Instead, mutations within these, and other, characteristically HCC-associated genes were observed independent of associated AAV integration.

The patient’s prior history of hepatitis virus infection and advanced age were risk factors for HCC. Although most hepatitis-related HCCs occur in individuals with advanced fibrosis or cirrhosis,²² up to 20% of HCCs occur in noncirrhotic livers.^23,24 Generally, the risk of HCC is higher in the population with hemophilia vs that in the general population, partly because patients with hemophilia have higher rates of hepatitis virus infection (with/without sustained viral response) and HIV infection.²⁵ HCV and HIV coinfection accelerates liver deterioration in patients with hemophilia.²⁶ Patients with hemophilia with HBV/HCV infection are generally also older than those with HBV/HCV infection but without hemophilia, with age-related HCC risk factors such as nonalcoholic fatty liver disease or nonalcoholic steatohepatitis, obesity, and excessive alcohol use.²⁵ 

In conclusion, the molecular and vector integration analysis of the index case of HCC after liver-directed rAAV-based gene therapy established no relationship to rAAV administration, and provides a model for investigating neoplasms in future clinical application of gene therapy with integrating vectors. Collection of real-world data on HCC occurrence through gene therapy–specific registries will also be critical to understanding the long-term risk of malignancies associated with AAV vector integration.

Acknowledgments: The authors acknowledge and thank the following individuals from Amsterdam University Medical Centers, University of Amsterdam, Amsterdam, The Netherlands: Joris Erdmann for expert surgical management of liver tissue collections, Joanne Verheij for expert pathologic biopsy preparation, and Daniela Stols-Gonçalves and Liesbeth van Huizen for coordination of sample management and interinstitutional transfer. The authors thank Mandy Lauw, Hematology, Erasmus Medical Center, Rotterdam, The Netherlands for coordination of collection and transfer of liver transplantation tissues. The authors thank Brahm Goldstein and Samantha Lucas, Research and Development, Hematology, CSL Behring Biotherapies for Life, for critically reviewing this manuscript and providing feedback. Jackie Read provided editorial support on behalf of GK Pharmacomm Ltd, and Lesley Taylor provided editorial support on behalf of Bioscript Science (Macclesfield, United Kingdom) with input from all authors and supported by CSL Behring, Haywards Heath, United Kingdom. The authors are deeply saddened by the passing of Manfred Schmidt and acknowledge his pioneering contributions to gene therapy and the study of viral vectorology and safety.

Contribution: M.S., G.R.F., S.W.P., R.D., and L.H. designed the research; M.S., M.C., H.T., L.H., and S.W.P. performed the research; M.S. contributed vital new reagents or analytical tools; S.W.P. and M.C. collected data; M.S., H.T., G.R.F., S.W.P., L.H., R.D., and P.E.M. analyzed and interpreted data; M.S. and H.T. performed statistical analysis; and M.S., G.R.F., S.W.P., L.H., M.C., and P.E.M. wrote the manuscript.

Conflict-of-interest disclosure: M.S. was the CEO and founder of GeneWerk GmbH. G.R.F. has received consultancy and speaker fees from AbbVie, BioMarin, Gilead, GlaxoSmithKline, MSD, and uniQure. M.C. has received financial support from Bayer, CSL Behring, Roche, and Novo Nordisk, and has received honoraria for lecturing or consultancy from Alexion, Bayer, CSL Behring, Daiichi Sankyo, Sobi, and Viatris. H.T. is an employee at GeneWerk GmbH. R.D. and L.H. are employees at uniQure. P.E.M. is an employee at CSL Behring. S.W.P. has received consultancy fees from Apcintex, ASC Therapeutics, Bayer, BioMarin, CSL Behring, GeneVentiv, HEMA Biologics, Freeline, LFB, Novo Nordisk, Pfizer, Regeneron/Intellia, Roche/Genentech, Sanofi, Takeda, Spark Therapeutics, and uniQure; has received research funding from Siemens; and holds membership in a scientific advisory committee for GeneVentiv.

Manfred Schmidt died on 13 January 2022.

Correspondence: Steven Pipe, Pediatrics and Pathology, University of Michigan, 1500 E Medical Center Dr, Ann Arbor, MI 48109-5718; e-mail: ummdswp@med.umich.edu; and Paul Monahan, CSL Behring, 1020 1st Ave, King Of Prussia, PA 19406; e-mail: paul.monahan@cslbehring.com.