Clonal analyses of retroviral vector integrations in ADA-SCID patients treated with stem cell gene therapy

Gene transfer into hematopoietic stem/progenitor cells (HSC) by gammaretroviral vectors is an effective treatment for patients affected by severe combined immunodeficiency (SCID) due to adenosine deaminase (ADA)-deficiency. Recent studied have indicated that gammaretroviral vectors integrate in a non-random fashion in their host genome, but there is still limited information on the distribution of retroviral insertion sites (RIS) in human long-term reconstituting HSC following therapeutic gene transfer. We performed a genome-wide analysis of RIS in transduced bone marrow-derived CD34+ cells before transplantation (in vitro) and in hematopoietic cell subsets (ex vivo) from five ADA-SCID patients treated with gene therapy combined to low-dose busulfan. Vector-genome junctions were cloned by inverse or linker-mediated PCR, sequenced, mapped onto the human genome, and compared to a library of randomly cloned human genome fragments or to the expected distribution for the NCBI annotation. Both in vitro (n=212) and ex vivo (n=496) RIS showed a non-random distribution, with strong preference for a 5-kb window around transcription start sites (23.6% and 28.8%, respectively) and for gene-dense regions. Integrations occurring inside the transcribed portion of a RefSeq genes were more represented in vitro than ex vivo (50.9 vs 41.3%), while RIS <30kb upstream from the start site were more frequent in the ex vivo sample (25.6% vs 19.4%). Among recurrently hit loci (n=50), LMO2 was the most represented, with one integration cloned from pre-infusion CD34+ cells and five from post-gene therapy samples (2 in granulocytes, 3 in T cells). Clone-specific Q-PCR showed no in vivo expansion of LMO2-carrying clones while LMO2 gene overexpression at the bulk level was excluded by RT-PCR. Gene expression profiling revealed a preference for integration into genes transcriptionally active in CD34+ cells at the time of transduction as well as genes expressed in T cells. Functional clustering analysis of genes hit by retroviral vectors in pre- and post-transplant cells showed no in vivo skewing towards genes controlling self-renewal or survival of HSC (i.e. cell cycle, transcription, signal transduction). Clonal analysis of long-term repopulating cells (>=6 months) revealed a high number of distinct RIS (range 42–121) in the T-cell compartment, in agreement with the complexity of the T-cell repertoire, while fewer RIS were retrieved from granulocytes. The presence of shared integrants among multiple lineages confirmed that the gene transfer protocol was adequate to allow stable engraftment of multipotent HSC. Taken together, our data show that transplantation of ADA-transduced HSC does not result in skewing or expansion of malignant clones in vivo, despite the occurrence of insertions near potentially oncogenic genomic sites. These results, combined to the relatively long-term follow-up of patients, indicate that retroviral-mediated gene transfer for ADA-SCID has a favorable safety profile.