Comparison and cross-validation of long-read and short-read target-enrichment sequencing methods to assess AAV vector integration into host genome

Evaluation of host integration profiles by adeno-associated virus (AAV) is an important component of de-risking novel AAV gene therapies. Targeted enrichment sequencing (TES) is a cost-effective and comprehensive method for assessing integration. Most published TES datasets have been generated using short-read sequencing, which enables quantitation of integration sites (ISs) and identifies patterns such as hotspots or clonal expansion. Characteristics such as IS length and recombination require longer reads to measure. The present study compared short-read to long-read TES using samples from monkeys treated with AAV and used <i>in vitro</i> lentiviral-treated samples, a stable cell line, and an engineered spike-in as controls. Both methods showed stochastic integration by both AAV and lentivirus, with most vector domains identified in ISs. More ISs were identified by short-read TES, as deeper coverage per base was achieved from a single sequencing run. AAV-treated samples showed minimal evidence of clonal expansion, in contrast to <i>in vitro</i> treated and stably transduced lentiviral cell line samples. Long-read TES revealed vector rearrangement in 4%-40% of ISs in AAV-treated animals. In summary, both methods yielded similar conclusions about relative numbers of ISs and overall patterns. Long-read TES identified fewer ISs but enabled measurement of IS length and recombination patterns.