Gene Therapy Advances But Struggles to Interpret Safety Data in Small Animal Models

Abstract

Two recent studies will generate additional controversy in the therapeutic application of gene transfer to treat human diseases. Although these studies point to incremental and important advances, they highlight continued uncertainty about the relevance of studies in murine species with respect to human applications. In the first study, Manuel Grez and colleagues (Ott et al., Nat. Med., 2006) report the successful use of a gamma retrovirus vector to treat chronic granulomatous disease (CGD). CGD is caused by mutations in genes encoding proteins of the NADPH oxidase pathway. Neutrophils of CGD patients lack the ability to generate sufficient reactive oxygen radicals to kill certain types of bacteria and fungi. The disease is life-threatening and leads to a uniform reduction in life expectancy. Current treatments include aggressive use of antibiotics and anti-fungals, pharmacologic agents to enhance neutrophil killing and, in severe cases, stem cell transplantation. In the absence of appropriate stem cell donors, transplant procedures have high morbidity and mortality and are considered experimental. In this setting, Ott et al. utilized gene therapy in two adult patients with life-threatening infections resistant to standard therapies.The approach included subtle, but important, protocol modifications. First, informed by the success of the X-linked severe combined immunodeficiency (X-SCID) and adenosine deaminase deficiency SCID trials in Europe, this trial utilized low doses of busulfan for conditioning by myelo-reduction prior to infusion of gene-modified CD34+ cells. These cells were derived from granulocyte colony stimulating factor (G-CSF) “mobilized” peripheral blood, allowing relatively large numbers of CD34+ cells (∼5 × 106/kg) to be infused. In addition, the investigators utilized a retrovirus vector in which the transgene, GP91phox, was expressed from the spleen focus-forming virus (SFFV) long terminal repeat (LTR). Expression of GP91phox in hematopoietic cells has been shown to be particularly high with this vector (Becker et al., Hum. Gene Ther., 1998, and Sadat et al., Gene Ther., 2003). Transduction efficiency (∼40-50%) of target hematopoietic cells was in the range observed in several previous human trials.After infusion of genetically modified cells, both subjects demonstrated a remarkably similar kinetics of gene marking in the myeloid lineage. In both, between 120 and 170 days post infusion, the percentage of gene marked leukocytes increased from ∼20% to 40–50%. Both patients demonstrated functional reconstitution of biochemical activity and killing capacity in their neutrophil lineage. Although both study subjects showed resolution of active bacterial or fungal infections, the degree to which full innate immunity was restored is not clear. These data provide solid evidence that the protocol provided relevant clinical benefit. In addition, extensive analysis of the integration sites in both patients revealed additional molecular data that may lead to new understanding of myelopoiesis following stem cell engraftment and contributes important new information regarding vector design as it relates to therapeutic outcomes. Increased contribution from gene marked cells after day 120 was concurrent with a relative narrowing of the distribution of vector insertion sites. These over-represented integration sites appeared to represent dominant myeloid progenitor clones. Insertion site identification demonstrated emergence of cells with integrations in one of three genetic loci: MDS1-EVI1, PRDM16 or SETBP1. The proportion of cells derived from insertions in these three loci eventually reached >80%. There was a significant tendency in both patients for the insertions in the MDS1-EVI1 loci to be increasingly represented over time. The insertions near these genes led to increased expression of the neighboring gene. For instance there was up to a 36-fold increase in MDS1-EVI1 expression by RNA analysis. However, the shift to myelopoiesis derived from stem or progenitor cells harboring integrations near these genes, all of which have been implicated in either human or murine myelodysplasia or leukemogenesis, was never associated with abnormally high white blood cell counts or any indication of leukemia. Transduced cells never became factor-independent in vitro and progenitor-derived colonies exhibited no abnormal replating capacity, both indications of hematopoietic transformation.Thus, integration into these loci appeared to influence hematopoiesis in vivo without any evidence of transformation or malignant proliferation. The follow-up is still short and the molecular data raise concern about the long-term risk associated with activation of these genes in dominant clones. However, as previously suggested by Kustikova et al. (Science, 2005), there is at least some theoretical arguments that the integration-driven myeloid expansion contributed to the therapeutic outcome in this trial. Overall, the data strongly suggest that vector design and disease targets are critically linked. The findings of this study have important implications for evaluation of risk:benefit ratio in human gene therapy trials, vector configuration and the reliance on murine preclinical studies in evaluating safety issues in human gene therapy trials.The reliance on mouse studies for evaluation of safety issues in gene therapy trials is also raised by a study published by Woods et al. (Nature, 2006). It has been speculated that abnormal expression of IL2Rγ could directly contribute to leukemia associated with vector insertional activation of specific additional genes. There is no evidence to date suggesting the leukemogenic potential of IL2Rγ overexpression (and many appear to have tried), although Davé et al.(Science, 2004) found statistical evidence suggesting that insertional activation of IL2Rγ is pro-leukemic. Woods et al. report data from mice transplanted with lentivirus vectors expressing IL2Rγ from a hybrid cellular/viral enhancer-promoter. The frequency of thymomas in mice expressing the IL2Rγ transgene may be more than in animals engrafted with cells transduced with an empty vector or with mock transduced cells. The frequency of thymomas may be less in IL2Rγ mice than mice expressing transgenic LMO2, the known oncogene that has been previously implicated in the French X-SCID trial. While clearly provocative and certain to increase the controversy about the use of so-called “unregulated” viral vectors in this disease, there are significant weaknesses in the presented data. For instance, the number of mice in each arm was fifteen or less, and there are no molecular details to show the transgene copy numbers in each group, apart from reference to the fact that transduction efficiency was very high (perhaps extremely high). In spite of the lentivirus vector design, this may be an important confounding issue, since it is not yet clear whether integrations from such enhancer-deleted vectors are unable to activate neighboring genes. In addition, transgene expression was several fold higher than normal in bone marrow and spleen, and thus significantly overexpressed. Data from two different donor cell types were pooled. There was reported to be no common integration sites in “lymphomic” tissues between different mice, but “similar” vector integrations in hematopoietic tissues of the same mouse with thymomas, but these data are not shown and the significance of this finding is not clear. The strain of mouse utilized, C57BL/6, is known to develop thymomas with age. In addition, one case is mentioned in which a tumor arose without evidence of gene transfer.Thus, a cautious interpretation of the somewhat preliminary data in this report is warranted. At the very least, there is very little direct relevance of these data to the safety of the apparently more moderate levels of transgene expression in the X-SCID clinical trials, particularly when one considers the outcomes of the alternative experimental therapies (or no therapy at all) for this group of patients. Irregardless, this study has implications for the planning and implementation of human trials. Taken at face value, and according to the conclusions of the authors, these data suggest that long-term murine studies need to be carried out before human trials begin. Currently, FDA approval in some cases allow concurrent murine studies. Whether the study by Woods et al. warrants a change in this approach is open to question.Adding to the uncertainty, in late April it was reported that one of the patients in the CGD study has died from sepsis. While there was no evidence that death had any direct relationship to the vector insertion-related myelopoiesis noted above, the potential role of the patient's post-gene therapy immune reconstitution, and thus the relationship to the gene therapy trial, is not yet clear. These two studies will energize the field and lead to additional discussion about the most appropriate methods to evaluate the safety and efficacy of vectors prior to opening of human trials. Two recent studies will generate additional controversy in the therapeutic application of gene transfer to treat human diseases. Although these studies point to incremental and important advances, they highlight continued uncertainty about the relevance of studies in murine species with respect to human applications. In the first study, Manuel Grez and colleagues (Ott et al., Nat. Med., 2006) report the successful use of a gamma retrovirus vector to treat chronic granulomatous disease (CGD). CGD is caused by mutations in genes encoding proteins of the NADPH oxidase pathway. Neutrophils of CGD patients lack the ability to generate sufficient reactive oxygen radicals to kill certain types of bacteria and fungi. The disease is life-threatening and leads to a uniform reduction in life expectancy. Current treatments include aggressive use of antibiotics and anti-fungals, pharmacologic agents to enhance neutrophil killing and, in severe cases, stem cell transplantation. In the absence of appropriate stem cell donors, transplant procedures have high morbidity and mortality and are considered experimental. In this setting, Ott et al. utilized gene therapy in two adult patients with life-threatening infections resistant to standard therapies. The approach included subtle, but important, protocol modifications. First, informed by the success of the X-linked severe combined immunodeficiency (X-SCID) and adenosine deaminase deficiency SCID trials in Europe, this trial utilized low doses of busulfan for conditioning by myelo-reduction prior to infusion of gene-modified CD34+ cells. These cells were derived from granulocyte colony stimulating factor (G-CSF) “mobilized” peripheral blood, allowing relatively large numbers of CD34+ cells (∼5 × 106/kg) to be infused. In addition, the investigators utilized a retrovirus vector in which the transgene, GP91phox, was expressed from the spleen focus-forming virus (SFFV) long terminal repeat (LTR). Expression of GP91phox in hematopoietic cells has been shown to be particularly high with this vector (Becker et al., Hum. Gene Ther., 1998, and Sadat et al., Gene Ther., 2003). Transduction efficiency (∼40-50%) of target hematopoietic cells was in the range observed in several previous human trials. After infusion of genetically modified cells, both subjects demonstrated a remarkably similar kinetics of gene marking in the myeloid lineage. In both, between 120 and 170 days post infusion, the percentage of gene marked leukocytes increased from ∼20% to 40–50%. Both patients demonstrated functional reconstitution of biochemical activity and killing capacity in their neutrophil lineage. Although both study subjects showed resolution of active bacterial or fungal infections, the degree to which full innate immunity was restored is not clear. These data provide solid evidence that the protocol provided relevant clinical benefit. In addition, extensive analysis of the integration sites in both patients revealed additional molecular data that may lead to new understanding of myelopoiesis following stem cell engraftment and contributes important new information regarding vector design as it relates to therapeutic outcomes. Increased contribution from gene marked cells after day 120 was concurrent with a relative narrowing of the distribution of vector insertion sites. These over-represented integration sites appeared to represent dominant myeloid progenitor clones. Insertion site identification demonstrated emergence of cells with integrations in one of three genetic loci: MDS1-EVI1, PRDM16 or SETBP1. The proportion of cells derived from insertions in these three loci eventually reached >80%. There was a significant tendency in both patients for the insertions in the MDS1-EVI1 loci to be increasingly represented over time. The insertions near these genes led to increased expression of the neighboring gene. For instance there was up to a 36-fold increase in MDS1-EVI1 expression by RNA analysis. However, the shift to myelopoiesis derived from stem or progenitor cells harboring integrations near these genes, all of which have been implicated in either human or murine myelodysplasia or leukemogenesis, was never associated with abnormally high white blood cell counts or any indication of leukemia. Transduced cells never became factor-independent in vitro and progenitor-derived colonies exhibited no abnormal replating capacity, both indications of hematopoietic transformation. Thus, integration into these loci appeared to influence hematopoiesis in vivo without any evidence of transformation or malignant proliferation. The follow-up is still short and the molecular data raise concern about the long-term risk associated with activation of these genes in dominant clones. However, as previously suggested by Kustikova et al. (Science, 2005), there is at least some theoretical arguments that the integration-driven myeloid expansion contributed to the therapeutic outcome in this trial. Overall, the data strongly suggest that vector design and disease targets are critically linked. The findings of this study have important implications for evaluation of risk:benefit ratio in human gene therapy trials, vector configuration and the reliance on murine preclinical studies in evaluating safety issues in human gene therapy trials. The reliance on mouse studies for evaluation of safety issues in gene therapy trials is also raised by a study published by Woods et al. (Nature, 2006). It has been speculated that abnormal expression of IL2Rγ could directly contribute to leukemia associated with vector insertional activation of specific additional genes. There is no evidence to date suggesting the leukemogenic potential of IL2Rγ overexpression (and many appear to have tried), although Davé et al.(Science, 2004) found statistical evidence suggesting that insertional activation of IL2Rγ is pro-leukemic. Woods et al. report data from mice transplanted with lentivirus vectors expressing IL2Rγ from a hybrid cellular/viral enhancer-promoter. The frequency of thymomas in mice expressing the IL2Rγ transgene may be more than in animals engrafted with cells transduced with an empty vector or with mock transduced cells. The frequency of thymomas may be less in IL2Rγ mice than mice expressing transgenic LMO2, the known oncogene that has been previously implicated in the French X-SCID trial. While clearly provocative and certain to increase the controversy about the use of so-called “unregulated” viral vectors in this disease, there are significant weaknesses in the presented data. For instance, the number of mice in each arm was fifteen or less, and there are no molecular details to show the transgene copy numbers in each group, apart from reference to the fact that transduction efficiency was very high (perhaps extremely high). In spite of the lentivirus vector design, this may be an important confounding issue, since it is not yet clear whether integrations from such enhancer-deleted vectors are unable to activate neighboring genes. In addition, transgene expression was several fold higher than normal in bone marrow and spleen, and thus significantly overexpressed. Data from two different donor cell types were pooled. There was reported to be no common integration sites in “lymphomic” tissues between different mice, but “similar” vector integrations in hematopoietic tissues of the same mouse with thymomas, but these data are not shown and the significance of this finding is not clear. The strain of mouse utilized, C57BL/6, is known to develop thymomas with age. In addition, one case is mentioned in which a tumor arose without evidence of gene transfer. Thus, a cautious interpretation of the somewhat preliminary data in this report is warranted. At the very least, there is very little direct relevance of these data to the safety of the apparently more moderate levels of transgene expression in the X-SCID clinical trials, particularly when one considers the outcomes of the alternative experimental therapies (or no therapy at all) for this group of patients. Irregardless, this study has implications for the planning and implementation of human trials. Taken at face value, and according to the conclusions of the authors, these data suggest that long-term murine studies need to be carried out before human trials begin. Currently, FDA approval in some cases allow concurrent murine studies. Whether the study by Woods et al. warrants a change in this approach is open to question. Adding to the uncertainty, in late April it was reported that one of the patients in the CGD study has died from sepsis. While there was no evidence that death had any direct relationship to the vector insertion-related myelopoiesis noted above, the potential role of the patient's post-gene therapy immune reconstitution, and thus the relationship to the gene therapy trial, is not yet clear. These two studies will energize the field and lead to additional discussion about the most appropriate methods to evaluate the safety and efficacy of vectors prior to opening of human trials.