Integration Site Analysis in Gene Therapy Patients: Expectations and Reality

Abstract

Integration site (IS) analysis is one of the major tools for addressing the safety of gene therapy clinical protocols based on the use of integrating vectors. Over the past years, the study of viral insertions in gene therapy–treated patients has allowed identifying insertional mutagenesis events, evaluating the safety of new viral vector platforms and tracking the in vivo clonal dynamics of genetically engineered cell products. While gene therapy is progressively expanding its impact on a broader area of clinical applications, increasingly more accessible, faster, and more reliable safety readouts are required from IS analysis. Several actors, from researchers to clinicians, from regulatory agencies to private companies, have to interface to different degrees with the results of IS analysis while developing and evaluating gene therapy products based on retroviral vectors. This review is aimed at providing a brief overview of what the current state and the future is of these studies with a particular focus on what are the main analytical constraints that should be considered upon conducting IS analysis in clinical gene therapy.

Introduction

Over the past decade, gene therapy (GT) has moved from being a niche experimental approach to become a widespread clinical reality for the treatment of a variety of hematological disorders.^1
–3 Approximately 25% of the platforms currently used in GT clinical trials utilize integrating vectors to treat inherited and acquired diseases, aiming at a long-lasting therapeutic correction (www.wiley.com/legacy/wileychi/genmed/clinical/). Integrating vectors are natural mutagens, as they insert their genetic payload semi-randomly into the host-cell genome.^4,5 As a consequence, the permanent gene transfer achieved by integrating vectors comes with a potential risk of altering the cell's natural genetic program due to the disruption of tumor suppressor genes, to the generation of new gene isoforms (from aberrant splicing or premature termination of transcripts) or to the activation of proto-oncogenes.^6,7 Since the first insertional mutagenesis events were reported in GT clinical trials,^8,9 integration site (IS) analysis became a crucial instrument for assessing the safety and efficacy of clinical GT.

The technologies for collecting IS from GT-treated patients have greatly improved in terms of output, resolution, and costs during the past decade. In general, all the techniques for IS retrieval are based on an initial in vitro enrichment for fragments containing vector-genome junctions. This is usually achieved through fragmentation of genomic DNA by different means (using restriction enzymes or sonication), followed by selective amplification with primers specific for the provirus sequences. In the early days, linker-mediated polymerase chain reaction (LM-PCR) products were shotgun cloned into competent bacteria, which generated colonies that were individually picked and sequenced using the Sanger method. This protocol has been extremely useful for identifying the overall features of viral IS distribution, as well as the causes of the early insertional mutagenesis events in GT clinical trials.^10
–12 Nonetheless, this procedure was too time-consuming and expensive for allowing real-time monitoring of GT outcome, since each colony had to be generated and processed separately, and each resulting sequence had to be analyzed manually. The advent of next-generation sequencing (NGS) ignited the development of a new protocols for IS retrieval with reduced costs and improved throughput potential.^13,14 These methods also provided, for the first time, an estimation of the relative abundance of each IS using sequencing reads count as a surrogate marker of clone size. The current state of the art for IS retrieval combines LM-PCR or linear amplification mediated (LAM)-PCR with Illumina sequencing.¹⁵ The features of past and present protocols for IS retrieval and the requirements for the proper handling of the sequencing results have been described elsewhere in detail.¹⁶ The potential and limitations of IS analysis for preclinical safety evaluations of GT have also been previously addressed.^17,18 This review, based on a summary of the author's work in light of the most relevant published literature in the field, aims at providing an overview of the current state of IS analysis in GT patients from a fresh perspective, by specifically pointing out some of the central analytical conundrums related to IS collection and by conveying a realistic scenario on the future of IS profiling in GT patients.

Is Analysis and Clone Size Quantification

One of the main issues regarding the current resolution of IS analysis is its capability to quantify properly the relative contribution of each IS in a given sample. As mentioned above, since the advent of NGS, counting the sequence reads belonging to each IS has been the most immediate way to provide a surrogate estimation of clonal sizes. However, this quantification system might be affected by biases related to the uneven exponential PCR amplification of certain vector-genome junctions at the expense of others.^16,19 Although sequence composition and length could be important elements affecting the efficiency of PCR amplification, groups working in the field have been unable to find a clear rule explaining why certain IS are more or less likely to be captured and amplified. Thus, random selection of amplicons during the initial PCR cycles could also be a major factor influencing the final sequence counts distribution among IS. Despite this anticipated bias, when serial dilutions of clones carrying known integrations into polyclonal transduced populations were performed, a fairly good degree of concordance was often observed between sequencing reads counts and expected clonal contribution.²⁰ However, in the author's experience, the consistency of IS quantification dropped substantially when going <5% of relative clonal contribution, where significant variability among technical replicates could be observed (unpublished data).

In order to overcome the quantitative biases related to PCR amplifications, methods have been developed to exploit molecular tags introduced at the time of genome fragmentation. One system is based on sonication of genomic DNA, which creates random fragments of different sizes in proximity of each vector-genome junction.²¹ Sonication could theoretically favor a less biased collection of IS, since it does not rely on the use of restriction enzymes whose recognition sequences could be unevenly distributed along the genome. Additionally, the resulting shearing sites can be retrospectively quantified and associated to each IS at the time of sequence processing, providing an estimation of the original numbers of genomes containing each given IS before exponential amplification occurred. Unfortunately, this method is also affected by intrinsic technical constraints. First, it has been recently showed that genomic fragmentation through sonication is not purely random and that sequence dependence of DNA shearing could be much higher that it is commonly expected.²² Furthermore and most importantly, the small range of fragment sizes, purposely tuned upon sonication, and the sequence length limitations of the current NGS technologies could lead rapidly to the saturation of the shearing sites diversity.¹⁶ New protocols are now relying on the use of a stretch of random sequences as molecular identifiers to be added during linear amplification and/or genome fragmentation.²⁰ If carefully designed, random sequence barcodes could make available an extremely high molecular tag diversity for IS quantification, which is not limited by fragment lengths or sequencing biases.

New systems for IS capturing are being developed that are not PCR-based and promise to improve the quantification resolution of IS analysis further. These include methods for enrichment of vector-genome junctions by biotin-capture or Cas9-based pull down of vector sequences, which could be capable of generating with high efficiency PCR-independent IS libraries.^23,24 Furthermore, these strategies could be combined with new high-throughput sequencing technologies, which are being optimized for working on low starting material and granting improved sequencing depth.²⁵ This will most likely allow reducing the costs of IS retrieval and, at the same time, will increase the chance of detecting less represented but potentially detrimental IS events.

Still, regardless of the system employed for IS retrieval, a realistic expectation from IS analysis on GT samples is that it can only provide an estimation of the overall population composition at the time of analysis rather than an exact quantification of the precise contribution of each clone. This is primarily due to the sampling biases associated to the collection of biological material from treated individuals. For the follow-up of GT trials, the largest fraction of a patient sample is usually dedicated to a series of laboratory tests, including sorting, phenotyping, functional studies, and vector copy number evaluation. An additional portion is then usually stored as backup material for further analyses on the basis of specific clinical protocol requirements. Therefore, in a usual setting, only ≤1 μg of genomic DNA (corresponding to <150,000 cells) could remain available for IS collection. This factor, combined with the technical constraints of IS retrieval techniques allowing only for a fractional coverage of all IS contained in each given sample, substantially impacts the resolution of IS analysis. Smaller clones, constituting a minimal proportion of the total engineered cell population, will likely fall below the detection limit in most of the IS samplings, and each attempt to measure their relative contribution and/or their survival precisely is prone to be heavily affected by significant variability. For this reason, strong claims based on the absence of a given IS in a population should be avoided unless this observation is confirmed by multiple samplings over time. For instance, insertional mutagenesis events giving rise to abrupt aberrant proliferation events have in the past escaped detection until the associated clonal expansion became self-evident.^26,27 According to the current resolution of IS analyses, a rule of thumb could be that a more reliable estimation can be expected when a clone constitutes >5–10% of a given population. These types of clones are usually more likely recaptured over different follow-ups, and changes in their relative frequency can be monitored in a fairly consistent fashion. As an example, different from what happened for the acute leukemia cases, progressive and slower clonal expansions occurring in GT clinical trials could be successfully detected and measured over time through IS analysis.^28,29

IS analysis for safety monitoring of GT patients

The need for taking advantage of IS analysis for clonal quantification relates to the demand for tools capable of tracking the fate of engineered cells once they are infused into an individual. In the case of ex vivo GT for inherited hematological disorders, clonal analyses allow the risk of insertional mutagenesis events and the stability of the hematopoietic reconstitution to be monitored. When studying the effects of engineered cells against tumors, as in the case of chimeric antigen receptor (CAR) T-cell therapies, clonal tracking is useful to estimate the efficacy of immune response and the persistence of immune surveillance for potential disease relapses. In both cases, the capability of detecting clonal expansions is key for the assessment of the clinical outcome. Although thresholds have been proposed to define clonal dominance events, there is currently a missing frame of reference for the calculation of the relative contribution of clones. For example, for evaluating the safety of GT trials, regulatory agencies often consider arbitrary thresholds of relative frequency to call for further monitoring of treated patients (e.g., IS abundance >10%; www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Public_assessment_report/human/003854/WC500208201.pdf). As simple and straightforward as it might seem, such definition could lead to substantial misinterpretation of IS data. Indeed, an often-overlooked factor is the choice of the total population over which the frequency of clones is calculated. For example, a given clone could constitute >10% of a specific cell fraction (e.g., T cells), but its frequency would obviously be much lower if normalized for the total number of cells in a blood sample. Before considering this scenario too trivial, one should be reminded that in a great number of cases the estimation of clonal abundance for GT safety follow-ups is performed on peripheral blood mononuclear cells (PBMCs), thus excluding the contribution of granulocytes, which constitute the largest fraction of human blood cells.

Second, when interpreting clonal abundance data, oligoclonality is often considered as one culprit of IS-driven clonal dominance. However, several factors other than insertional mutagenesis could lead to oligoclonality. For instance, in the early phases of hematopoietic reconstitution, it is common to detect oligoclonal populations and dominant clones. This is generally due to an unequal activation of clones sustaining the first hematopoietic wave, during which stressed hematopoiesis physiologically select for cells with the highest fitness and the faster differentiation potential.³⁰ For this reason, it is difficult through IS analysis to identify true insertional mutagenesis events on samples collected in the first 6 months after infusion of engineered cells. Furthermore, oligoclonality can be observed even many years after GT, independently from positive growth advantages induced by insertional mutagenesis. Indeed, according to several factors, as the level of engraftment of infused cells and the degree of selective pressure to which they undergo related to the disease that is being treated, the diversity of the pool of genetically engineered cells can drop substantially over time, even in the absence of vector-induced clonal stress. In the context of hematopoietic stem-cell (HSC)-GT, it has been shown that a few thousand clones, out of the millions originally infused, become responsible for the long-term maintenance of the engineered hematopoiesis at steady state.^30,31 In CAR T-cell therapies, heavy selective pressure leading to clonal dominance and oligoclonality could be actually expected when an effective antitumor response is mounted, in the absence of vector genotoxicity.³²

The in vitro manipulation of the infused cell product upon transduction could also impact the in vivo diversity of IS repertoire. In the case of HSC-GT, the in vitro culture time is typically limited to 3–4 days, and this in principle allows for preserving a very polyclonal repertoire of engineered cells. On the other hand, in protocols involving lymphocyte engineering, where expansion of transduced PBMCs could last for 10–15 days, the diversity of the infused cell pool could drop during the culture time, and as a result, unbalanced repertoires could be detected in treated patients independently from other in vivo selective pressures. Finally, even when engraftment levels are high and the disease background does not induce clonal imbalances, mathematical models predicts that, by means of purely random selection and neutral drifts, few clones with uneven sizes might eventually take over the entire pool of infused cells,^33,34 thus giving rise to an oligoclonal repertoire of IS in the long term.

Overall, the definition of clonal dominance events (in the absence of self-evident abnormal proliferation) remains based on arbitrary thresholds, which often do not take into account the physiological nature of clonal dynamics. Relying on the information from current and future IS-based clonal tracking studies will be crucial for generating instructed models of hematopoietic cell behavior and for ultimately providing a more reliable system for identifying aberrant deviations from physiological population drifts.

Analytical Workflows for Is Retrieval and Analysis

In light of the constraints discussed above, a fair approach to IS analysis should acknowledge that whatever technology is employed for IS retrieval, it will never be able to overcome fully some inherent technical obstacles. This should open the debate on whether and how much it is worth investing in the development of new protocols for improving the capturing of vector-genome junctions, in the context of GT safety and efficacy monitoring. As mentioned above, the early studies based on shotgun cloning and Sanger sequencing, although based on few hundred IS from each individual, were already capable of unveiling the main features of viral insertional preferences^35
–37 and were extremely useful for the identification of the IS associated to the first events of insertional mutagenesis in humans.^10
–12 Thus, in terms of providing the basic information on viral IS distribution, these protocols have been working just fine. Nonetheless, the step toward adapting the technology to NGS platforms was natural to take, as it dramatically reduced the time and costs associated to IS collection. The latest substantial advancement of IS retrieval protocols has been the use of molecular identifiers for refining the IS quantification potential. It should now honestly be recognized that a technological plateau is fast approaching for further improving the resolution of IS collection. Efforts are currently in place to retrieve a higher amount of IS from each sample by modifying the vector sequences capturing techniques and/or by increasing the sequencing depth. Nonetheless, as pointed out above, the inherent limitations in the size and nature of available clinical samples are likely to cripple any attempt to achieve the ultimate coverage of viral IS.

In the author's opinion, the technological efforts should now be directed toward generating simpler and less expensive protocols for IS retrieval. As the number of clinical trials employing integrating viral vectors is rapidly increasing, the technology for IS collection should become more broadly accessible, particularly to the main clinical sites where GT is conducted and where patients follow-up is performed. This would cut costs related to sample preparation, storage, and shipping while at the same time allowing IS collection to be performed more rapidly and on fresh patient material. Importantly, it would also reduce the chance of contamination among unrelated samples, as IS retrieval would not be performed in a center collecting vials from different clinical sources. Additionally, such an approach would make the clinical laboratories involved in patient follow-up more conscious of the technical aspects related to IS retrieval and more capable of putting the results of IS analysis in relation to the practical facets of laboratory procedures. For instance, deviations from the protocols of sample collection, cell purification, and genome extraction, which are usually performed in clinical laboratories other than the ones in charge of IS retrieval, might occasionally be the reasons for odd results emerging upon IS analysis. Being able to trace back the technical steps that could have led to a specific outcome upon IS collection is crucial for a proper discrimination between technical errors or results that should raise safety and/or efficacy concerns. Nonetheless, in order for a clinical lab to be able to follow through with all the steps of IS enrichment, a series of technical issues must be addressed, and compliance should be carefully assessed, particularly with regard to the control of inter-sample cross-contamination. A wise strategy should envisage a counselling by experts in the field for the proper design of the laboratory areas dedicated to IS collection and for the setup and troubleshooting of the protocol that will be employed for IS retrieval.

This leads to a second consideration regarding the handling and interpretation of IS data. Once IS-containing amplicons undergo NGS, the following steps include sequence cleaning, mapping, and data storage. The current state of the technology for handing IS data sets, including dealing with contamination issues, have been addressed elsewhere.¹⁶ Few groups have acquired over the years the capability of managing high amounts of IS sequence data (mostly from Illumina platform), learning in the process how best to meet the changing requirements for a progressively faster and more precise IS identification. Even when the “wet” laboratory procedure is made simple and accessible, these analytical steps would require specific computational/data-storage capabilities and a standardized analytical pipeline. Thus, it is difficult to envisage that this processing will eventually be performed by the same clinical laboratories collecting IS amplicons. Moreover, the downstream analyses (e.g., estimation of clonal diversity, identification of insertional hotspots, and correlation of IS distribution with other genomic features) rely, even more than the mere NGS data handling, on the experience of laboratories on managing IS data sets with high proficiency and on their familiarity with the study of IS profiles in GT patients.

Overall, a sensible approach for the safety and efficacy follow-up of GT patients could in the future foresee a decentralized IS collection performed with a straightforward and reproducible protocol including all steps up to NGS. Then, each clinical laboratory would transfer the raw sequence data to facilities that will be responsible for providing standardized analytical pipelines for IS processing, final reporting, and data storage in an anonymous and controlled-access fashion. This workflow would not only have the advantage of granting a faster and more efficient IS collection, but also would allow comparative studies to be performed among different GT trials using uniform and centralized analytical tools.

The Future of Is Analysis

On the basis of the considerations reported in this review, one could also wonder why keep performing IS analysis in GT-treated individuals. After all, the predictive value of IS profiling is still debated, and IS analysis shares the common limitation of other clonal tracking systems in terms of sampling bias, quantification potential, and data handling. Still, there are a number of reasons why IS analysis remains a paramount tool for the follow-up of GT patients.

First, there is not yet a clear picture of the safety and risks of retroviral GT. In one of the author's earliest reviews, some questions were raised regarding the interpretation of IS profiles in GT trials, most of which remain unanswered.¹⁷ Additionally, it was highlighted lately how there is an open debate on the co-factors that might have played a role in producing vastly different safety outcomes in otherwise relatively similar GT approaches.¹⁸ As GT has become mainstream only recently, the cohort of GT individuals from whom useful follow-up data can be gained is still limited. Additionally, only a few of them have reached a mid/long-term follow-up, and just a fraction of these individual have undergone thorough IS profiling. The recent report of a vector-related leukemia case, which was detected 15 years after retroviral GT,³⁸ shows how limited our understanding still is of the influence of vectors perturbations on engineered cells. In this regard, it would be important to achieve a consensus on what is considered a IS event, which should raise concern on the use of a given engineered cell product or on the follow-up of an already treated patients. For example, recent works have shown that chimeric transcripts can be generated due to the presence of splice sites contained in the vector sequences.⁷ This aspect should be carefully evaluated at the preclinical level, in particular for those GT applications employing vector constructs containing multiple donor/acceptor splice sites and based on viral platforms that have an intrinsic tendency of integrating in proximity or inside coding regions.

In the author's opinion, the existence of these many open issues calls for a systemic long-term safety monitoring of GT patients, including the ones treated with lentiviral or self-inactivating gammaretroviral vectors, which are deemed to be safer platforms than the ones used in the earliest clinical trials.^7,39,40 Such information would be particularly relevant as GT approaches move from the treatment of lethal disorders to broader applications on milder disease forms, where the risk–benefit balance significantly changes. Furthermore, the safety data that will be generated in the next decade will be crucial as baseline information on the in vivo dynamics of transduced cells, before moving forward toward the use of gene-editing platforms on which the standard IS-based tracking of engineered cells will no longer be applicable.

IS analysis also remains important for the in vitro evaluation of the GT product, while new protocols envisaging different in vitro cell product expansions and manipulations are being developed.^41

–44 In this context, IS analysis is one the easiest and more straightforward tests to assess the clonal diversity of the infused cells, which might ultimately affect the efficacy and safety of GT. Of note, information on IS distribution can be easily combined with the results of other in vitro assays for a better evaluation of potential vector-induced perturbations.¹⁸ Moreover, with relatively small adaptations of the standard protocols for IS collection, it is possible to detect off-target activities in GT strategies envisaging the use of site-specific gene addition platforms.⁴⁵

In addition to remaining a paramount tool for the follow-up of GT patients, with its improved resolution potential, IS analysis has become a great instrument for bringing to light previously inaccessible information regarding the dynamics of blood cells in humans. It has been shown that despite the above-mentioned technical constraints, IS analysis is an extremely valuable tool for tracing the fate of engineered hematopoietic progenitors or mature T cells once infused in patients.^30,46
–48 It should be acknowledged that IS-based clonal tracking is, to date, the only method for studying human hematopoietic cell functions and dynamics that does not rely on surrogate in vitro assays or on transplantation into humanized animal models, both carrying inherent limitations on reproducing the conditions of the human body environment. Moreover, other systems based on the use of natural genome rearrangements as clonal markers for in vivo in human studies (e.g., VDJ recombination)⁴⁹ are limited to the tracking of mature lymphocytes populations and cannot be extended, for instance, to the study of myeloid cells or hematopoietic progenitors. The efficacy of IS analysis for clonal tracking was recently reviewed in comparison to other platforms, highlighting limitations and strengths of insertional profiling for tracing the fate of infused cells.¹⁶

If it is relatively easy to envisage how the results of IS-based clonal tracking could benefit therapeutic applications such as bone-marrow transplant, it should be reminded that the fallout of these studies extend also to cancer research. In this regard, a natural application of IS-based tracking would be on monitoring, in treated individuals, the fate, expansion, and survival of clones genetically engineered to activate immune responses against tumors. Less obvious information that might also derive from IS analysis is related to tumor latency. In the GT patients where vector-related adverse events occurred, viral integration could have acted as one of the first triggers for oncogenic transformation. Knowing when the patients' cells were transduced, for the first time, direct information is available on when the first genetic perturbations were introduced. This could in principle, by IS-based tracking, allow how the nature of the genetic alterations, the composition of vector-marked cell populations, the conditioning regimens, and the response of immune system influenced the timing that elapsed from the first genetic trigger to the final tumor outbreak to be addressed. Nonetheless, these types of studies should be carefully designed in order to consider pre-existing factors predisposing to leukemic transformation and to rule out, as much as possible, that IS-driven perturbations are not representing secondary events if not even neutral epiphenomenon upon oncogenic transformation.

Conclusions

In conclusion, IS analysis has played a fundamental role in bringing GT to the center stage, being the main tool for the assessment of safety of genetic engineering with integrating viral vectors. However, although the technology for IS retrieval has greatly advanced over the past years, it should be recognized that inherent biases and limitations will always exist upon analyzing IS profiles of GT-treated patients. Understanding and acknowledging these analytical conundrums would benefit a better interpretation of IS data and would generate more realistic expectations regarding IS analysis in the actors involved, at different stages, on the evaluations of GT products. The development of standardized and simple workflows for samples handling and IS collection would extend the accessibility and reproducibility of IS analysis to a broader clinical arena, while GT applications are rapidly expanding and researchers involved in IS profiling are embarking into addressing new and more challenging biological questions.

Footnotes

Acknowledgments

The work of Luca Biasco is supported by the Gene Therapy Program of Dana-Farber/Boston Children's Cancer and Blood Disorders Center, Boston, MA, and by the UCL Great Ormond Street Institute of Child Health, Faculty of Population Health Sciences, London, United Kingdom. The author is a scientific consultant of GlaxoSmithKline, for the safety evaluation of a lentiviral gene therapy trial for Wiskott–Aldrich syndrome conducted by the San Raffaele Telethon Institute for Gene Therapy (SR-TIGET), Milan, Italy.

Author Disclosure

LB is a consultant to GlaxoSmithKline (GSK) for the safety of a lentiviral gene therapy trial for Wiskott Aldrich Syndrome conducted at Telethon Institute for Gene Therapy (TIGET) in Milan, Italy.

References

Naldini

. Gene therapy returns to centre stage. Nature, 2015; 526:351–360.

Williams

. Broadening the indications for hematopoietic stem cell genetic therapies. Cell Stem Cell, 2013; 13:263–264.

Fischer

, Hacein-Bey-Abina

, Cavazzana-Calvo

. Gene therapy of primary T cell immunodeficiencies. Gene, 2013; 525:170–173.

, Li

, Crise

, et al. Transcription start regions in the human genome are favored targets for MLV integration. Science, 2003; 300:1749–1751.

Mitchell

, Beitzel

, Schroder

ARW

, et al. Retroviral DNA integration: ASLV, HIV, and MLV show distinct target site preferences. PLoS Biol, 2004; 2:E234.

Uren

, Kool

, Berns

, et al. Retroviral insertional mutagenesis: past, present and future. Oncogene, 2005; 24:7656–7672.

Cesana

, Sgualdino

, Rudilosso

, et al. Whole transcriptome characterization of aberrant splicing events induced by lentiviral vector integrations. J Clin Invest, 2012; 122:1667–1676.

Hacein-Bey-Abina

, Von Kalle

, Schmidt

, et al. LMO2-associated clonal T cell proliferation in two patients after gene therapy for SCID-X1. Science, 2003; 302:415–419.

Ott

, Schmidt

, Schwarzwaelder

, et al. Correction of X-linked chronic granulomatous disease by gene therapy, augmented by insertional activation of MDS1-EVI1, PRDM16 or SETBP1. Nat Med, 2006; 12:401–409.

10.

Aiuti

, Cassani

, Andolfi

, et al. Multilineage hematopoietic reconstitution without clonal selection in ADA–SCID patients treated with stem cell gene therapy. J Clin Invest, 2007; 117:2233–2240.

11.

Deichmann

, Hacein-bey-abina

, Schmidt

, et al. Vector integration is nonrandom and clustered and influences the fate of lymphopoiesis in SCID-X1 gene therapy. J Clin Invest, 2007; 117:2225–2232.

12.

Schwarzwaelder

, Howe

, Schmidt

, et al. Gammaretrovirus-mediated correction of SCID-X1 is associated with skewed vector integration site distribution in vivo . J Clin Invest, 2007; 117:2241–2249.

13.

Gabriel

, Eckenberg

, Paruzynski

, et al. Comprehensive genomic access to vector integration in clinical gene therapy. Nat Med, 2009; 15:1431–1436.

14.

Paruzynski

, Arens

, Gabriel

, et al. Genome-wide high-throughput integrome analyses by nrLAM-PCR and next-generation sequencing. Nat Protoc, 2010; 5:1379–1395.

15.

Schmidt

, Schwarzwaelder

, Bartholomae

, et al. High-resolution insertion-site analysis by linear amplification–mediated PCR (LAM–PCR). Nat Methods, 2007; 4:1051–1057.

16.

Scala

, Leonardelli

, Biasco

. Current approaches and future perspectives for in vivo clonal tracking of hematopoietic cells. Curr Gene Ther, 2016; 16:184–193.

17.

Biasco

, Baricordi

, Aiuti

. Retroviral integrations in gene therapy trials. Mol Ther, 2012; 20:709–716.

18.

Rothe

, Schambach

, Biasco

. Safety of gene therapy: new insights to a puzzling case. Curr Gene Ther, 2014; 14:429–436.

19.

Brugman

, Suerth

, Rothe

, et al. Evaluating a ligation-mediated PCR and pyrosequencing method for the detection of clonal contribution in polyclonal retrovirally transduced samples. Hum Gene Ther Methods, 2013; 24:68–79.

20.

Leonardelli

, Pellin

, Scala

, et al. Computational pipeline for the identification of integration sites and novel method for the quantification of clone sizes in clonal tracking studies. Mol Ther, 2016; 24:S212 (abstract).

21.

Berry

, Gillet

, Melamed

, et al. Estimating abundances of retroviral insertion sites from DNA fragment length data. Bioinformatics, 2012; 28:755–762.

22.

Poptsova

, Il'icheva

, Nechipurenko

, et al. Non-random DNA fragmentation in next-generation sequencing. Sci Rep, 2015; 4:4532.

23.

Miyazato

, Katsuya

, Fukuda

, et al. Application of targeted enrichment to next-generation sequencing of retroviruses integrated into the host human genome. Nat Publ Gr, 2016; 1–10.

24.

Ramani

, Shendure

. Smash and DASH with Cas9. Genome Biol, 2016; 9–11.

25.

Larkin

, Henley

, Jadhav

, et al. Length-independent DNA packing into nanopore zero-mode waveguides for low-input DNA sequencing. Nat Nanotechnol, 2017 Sep 11 [Epub ahead of print]; DOI:10.1038/nnano.2017.176.

26.

Hacein-bey-abina

, Schmidt

. A serious adverse event after successful gene therapy for X-linked severe combined immunodeficiency. N Engl J Med, 2003; 348:255–266.

27.

Braun

, Boztug

, Paruzynski

, et al. Gene therapy for Wiskott–Aldrich syndrome—long-term efficacy and genotoxicity. Sci Transl Med, 2014; 6:227ra33–227ra33.

28.

Stein

, Ott

, Schultze-Strasser

, et al. Genomic instability and myelodysplasia with monosomy 7 consequent to EVI1 activation after gene therapy for chronic granulomatous disease. Nat Med, 2010; 16:198–204.

29.

Cavazzana-Calvo

, Payen

, Negre

, et al. Transfusion independence and HMGA2 activation after gene therapy of human β-thalassaemia. Nature, 2010; 467:318–322.

30.

Biasco

, Pellin

, Scala

, et al. In vivo tracking of human hematopoiesis reveals patterns of clonal dynamics during early and steady-state reconstitution phases. Cell Stem Cell, 2016; 19:107–119.

31.

Aiuti

, Biasco

, Scaramuzza

, et al. Lentiviral hematopoietic stem cell gene therapy in patients with Wiskott–Aldrich syndrome. Science, 2013; 341:1233151.

32.

Scholler

, Brady

, Binder-Scholl

, et al. Decade-long safety and function of retroviral-modified chimeric antigen receptor T cells. Sci Transl Med, 2012; 4:132ra53.

33.

Dingli

, Pacheco

. Stochastic dynamics and the evolution of mutations in stem cells. BMC Biol, 2011; 9:41.

34.

Klein

, Simons

. Universal patterns of stem cell fate in cycling adult tissues. Development, 2011; 138:3103–3111.

35.

Cattoglio

, Facchini

, Sartori

, et al. Hot spots of retroviral integration in human CD34+ hematopoietic cells. Blood, 2007; 110:1770–1778.

36.

Felice

, Cattoglio

, Cittaro

, et al. Transcription factor binding sites are genetic determinants of retroviral integration in the human genome. PLoS One, 2009; 4:e4571.

37.

Biasco

, Ambrosi

, Pellin

, et al. Integration profile of retroviral vector in gene therapy treated patients is cell-specific according to gene expression and chromatin conformation of target cell. EMBO Mol Med, 2011; 3:89–101.

38.

Six

, Gandemer

, Magnani

, et al. LMO2 associated clonal T cell proliferation 15 years after gamma-retrovirus mediated gene therapy for SCIDX1. Mol Ther, 2017; 25:1–363 (abstract).

39.

Hacein-Bey-Abina

, Pai

S-Y

, Gaspar

, et al. A modified γ-retrovirus vector for X-linked severe combined immunodeficiency. N Engl J Med, 2014; 371:1407–1417.

40.

Montini

, Cesana

, Schmidt

, et al. The genotoxic potential of retroviral vectors is strongly modulated by vector design and integration site selection in a mouse model of HSC gene therapy. J Clin Invest, 2009; 119:964–975.

41.

Zonari

, Desantis

, Petrillo

, et al. Efficient ex vivo engineering and expansion of highly purified human hematopoietic stem and progenitor cell populations for gene therapy. Stem Cell Reports, 2017; 8:977–990.

42.

Goessling

, Allen

, Guan

, et al. Prostaglandin E2 enhances human cord blood stem cell xenotransplants and shows long-term safety in preclinical nonhuman primate transplant models. Cell Stem Cell, 2011; 8:445–458.

43.

Tumaini

, Lee

, Lin

, et al. Simplified process for the production of anti-CD19-CAR engineered T cells. Cytotherapy, 2014; 15:1406–1415.

44.

Gattinoni

, Restifo

. Moving T memory stem cells to the clinic. Blood, 2013; 121:567–568.

45.

Gabriel

, Lombardo

, Arens

, et al. An unbiased genome-wide analysis of zinc-finger nuclease specificity. Nat Biotechnol, 2011; 29:816–823.

46.

Wang

, Berry

, Malani

, et al. Dynamics of gene-modified progenitor cells analyzed by tracking retroviral integration sites in a human SCID-X1 gene therapy trial. Blood, 2010; 115:4356–4366.

47.

Biasco

, Scala

, BassoRicci

, et al. In vivo tracking of T cells in humans unveils decade-long survival and activity of genetically modified T memory stem cells. Sci Transl Med, 2015; 7:273ra13.

48.

Oliveira

, Ruggiero

, Teresa

, et al. Tracking genetically engineered lymphocytes long-term reveals the dynamics of T cell immunological memory. 2015; 7:1–14.

49.

Robins

, Campregher P

, Srivastava

, et al. Comprehensive assessment of T-cell receptor beta-chain diversity in alphabeta T cells. Blood, 2009; 114:4099–4107.