Human Gene Therapy

To help define the safety profile of the use of adenovirus (Ad) gene transfer vectors in humans, this report summarizes our experience since April 1993 of the local administration of E1^-/E3^- Ad vectors to humans using low (<10⁹ particle units) or intermediate (10⁹-10¹¹ particle units) doses. Included in the study are 90 individuals and 12 controls, with diverse comorbid conditions, including cystic fibrosis, colon cancer metastatic to liver, severe coronary artery disease, and peripheral vascular disease, as well as normals. These individuals received 140 different administrations of vector, with up to seven administrations to a single individual. The vectors used include three different transgenes (human cystic fibrosis transmembrane conductance regulator cDNA, E. coli cytosine deaminase gene, and the human vascular endothelial growth factor 121 cDNA) administered by six different routes (nasal epithelium, bronchial epithelium, percutaneous to solid tumor, intradermal, epicardial injection of the myocardium, and skeletal muscle). The total population was followed for 130.4 patient-years. The study assesses adverse events, common laboratory tests, and long-term follow-up, including incidence of death or development of malignancy. The total group incidence of major adverse events linked to an Ad vector was 0.7%. There were no deaths attributable to the Ad vectors per se, and the incidence of malignancy was within that expected for the population. Overall, the observations are consistent with the concept that local administration of low and intermediate doses of Ad vectors appears to be well tolerated.

In this study we analyze the adverse events and abnormal laboratory parameters following local administration of low (<10⁹ particle units) and intermediate (10⁹-10¹¹ particle units) single and repetitive doses (140 total) of E1^-E3^- adenovirus (Ad) gene transfer vectors administered to the respiratory epithelium, solid tumors, skin, myocardium, and skeletal muscle in eight gene transfer trials since April 1993. In the accompanying paper by Harvey et al., (Hum. Gene Ther. 2002; 13:15-63), we conclude that for the total group, no deaths were attributable to the Ad vectors per se, and the incidence of major adverse events likely caused by an Ad vector was 0.7%. The present study analyzes the trials as a group to evaluate risk factors for the adverse events, abnormal values among laboratory parameters, and known deaths. Ten putative risk factors were assessed, including "patient-related" (age, sex, comorbid index and pretherapy anti-Ad antibodies),"vector-related" (dose, route, transgene, and number of vector administrations), and "trial-related" (trial in which the individual was enrolled, and whether surgery was part of the trial). While assessment of each factor individually suggested several possible associations with adverse events, abnormal laboratory parameters, or deaths, multivariate analysis identified only age, comorbid index, and surgery(comorbid index for death; age and surgery for nondeath adverse events) as variables significantly associated with increased risk for a major (severity scale 3-4 of 4) adverse event for individuals enrolled in these gene transfer trials. Importantly, multivariate analysis suggested that vector-related parameters, including dose, route, transgene, or number of vector administrations at the doses and routes evaluated in these studies, do not appear to be significant risk factors for a major adverse event. With the caveat that these are phase I, uncontrolled trials, we conclude that (1) there is no definitive risk factor that will clearly predict a major adverse outcome resulting from local administration of low and intermediate doses of Ad gene transfer vectors; and (2) major adverse events in these gene transfer trials are associated primarily with the study population and/or trial procedures, not the Ad vectors themselves. This assessment is consistent with the concept that local administration of low and intermediate doses of Ad gene transfer vectors appears to be well tolerated.

Since human adenoviruses replicate only in human cells, toxicology studies with adenoviral vectors are hampered by the lack of a permissive nonhuman host. Before a replication-competent adenoviral vector expressing HSV-tk (Ad.OW34) can be used in clinical studies for intratumoral injections in patients with cutaneous lesions of head and neck cancer or intralesional injection for in situ vaccination strategy in advanced metastatic melanoma patients, risks have to be estimated in animal studies. In an attempt to assess potential toxicology, dissemination, persistence and shedding, we injected Ad.OW34 subcutaneously into cotton rats. (Sigmodon hispidus), which are considered a semipermissive host for human adenoviruses. The animals underwent one or two subcutaneous injection cycles with 2.3 × 10¹² viral particles/kg each or a single course with 6.9 × 10¹³ viral particles/kg and were analyzed at defined time points for histopathological changes in the brain, heart, lungs, spleen, liver, kidneys, ovaries, and skin. Additionally, these tissues as well as urine, feces, mouth, and skin swabs were analyzed at multiple time points by real-time quantitative polymerase chain reaction for the presence of vector sequences. The only significant treatment-related histopathologic finding was dermatitis with mild acanthosis at the site of vector injection. All other tissues evaluated were within normal limits or showed changes that were most likely incidental or spontaneous in nature. Vector sequences were detected in the skin at the injection site and to a lesser extent in the liver, spleen, and lungs. In addition, small amounts of vector DNA were detected in the ovaries. The vector sequences were rapidly cleared and the absence of viral sequences in the excreta and swabs of the majority of animals suggest that there was no significant replication of the vector in this host. The administration of Ad.OW34 was also associated with mild hyperamylasemia, lymphocytosis, and granulocytosis; however, we did not observe any clinical signs of illness or death in the experimental animals over the course of the study.

We constructed a first-generation adenovirus vector (AVC3FIX5) that we used to assess the rhesus macaque as a nonhuman primate model for preclinical testing of hemophilia B gene therapy vectors. Although we succeeded in our primary objective of demonstrating expression of human factor IX we encountered numerous toxic side effects that proved to be dose limiting. Following intravenous administration of AVC3FIX5 at doses of 3.4 × 10¹¹ vector particles/kg to 3.8 × 10¹² vector particles/kg, the animals in our study developed antibodies against human factor IX, and dose-dependent elevations of enzymes specific for liver, muscle, and lung injury. In addition, these animals showed dose-dependent prolongation of clotting times as well as acute, dose-dependent decreases in platelet counts and concomitant elevation of fibrinogen and von Willebrand factor. These abnormalities may be caused by the direct toxic effects of the adenovirus vector itself, or may result indirectly from the accompanying acute inflammatory response marked by elevations in IL-6, a key regulator of the acute inflammatory response. The rhesus macaque may be a useful animal model in which to evaluate mechanisms of adenovirus toxicities that have been encountered during clinical gene therapy trials.

Adenoviruses are commonly used as vectors in human clinical gene therapy trials. High doses of intravenous adenovirus vectors have been associated with development of thrombocytopenia of undetermined origin. Viral internalization requires the presence cell surface integrins, α _v β ₃ or α _v β ₅, that can blind ligands with a arginine-glycine-aspartic acid (RGD) sequence. This sequence is found in the adenovirus penton base. Platelets express the α _v β ₃ integrin and other integrins that bind the RGD sequence of ligands such as fibrinogen, laminin, vitronectin, and von Willebrand factor (vWF). Platelet aggregation is mediated, in part, by the binding of the RGD sequence of fibrinogen to a platelet surface integrin, glycoprotein IIb/IIIa (GP IIb/IIIa). We investigated whether adenovirus particles could interfere with or potentiate agonist-induced platelet aggregation. Incubation of platelet-rich plasma with adenovirus under stirred conditions did not promote spontaneous aggregation. The addition of physiological platelet agonists, ADP, collagen, or epinephrine, induced platelet aggregation. However, the presence of adenovirus in a wide range of concentrations did not inhibit or potentiate agonist-induced aggregation. These results suggest that the adenovirus-associated thrombocytopenia observed in vivo is independent of a direct effect of the virus on platelet aggregation.

Preclinical and clinical studies with adenoviral vectors have clearly illustrated the potential advantages of this gene transfer system. However, many studies have also demonstrated potent immune responses directed at both vector and transduced cells. We examined in vitro responses of human peripheral blood mononuclear cells (PBMC) to virus exposure as a model for this host response. PBMC were isolated from normal donors and incubated with wild-type adenovirus (Ad5), Ad5 variants deleted for segments of E1 and/or E3, and empty viral capsids. Proinflammatory cytokine release was monitored for 96 hr. Induction of TNF-α by intact virions was low although stimulation by empty capsid gave a significant and sustained response. Induction of IL-6, GM-CSF, and a panel α- and β-chemokines by intact virions was prominent, often approaching results obtained with 2.5 μg/ml of lipopolysaccharide (LPS). Responses were generally independent of virion genetic composition and were only partially blunted when UV-inactivated virus was used. Dose-response data showed 100-fold increases in virion concentration produced a maximum 3-fold increase in cytokine release, suggesting saturation. Surprisingly, prominent stimulation occurred after addition of empty capsid, which typically provoked responses equivalent to those seen with LPS stimulation. We present arguments that cellular signal transduction mechanisms activated by binding of virions/capsids stimulate transcription of proinflammatory cytokine genes.

The effects of intravenous administration of a first-generation adenoviral vector expressing β-galactosidase were compared in two baboons receiving a high dose or lower dose of vector, 1.2 × 10¹³ or 1.2 × 10¹² particles/kg, respectively. The high-dose baboon developed acute symptoms, decreased platelet counts, and increased liver enzymes, and became moribund at 48 hr after injection, while the lower-dose baboon developed no symptoms. Expression of the β-galactosidase transgene was prominent in liver, spleen, and endothelium of the arterial vasculature in the high-dose baboon, but was much more limited and spared the endothelium in the lower-dose baboon. Injury to the vascular endothelium was the most prominent abnormality in the high-dose baboon. Extensive histological studies provide a detailed picture of the pathology associated with a lethal dose of first-generation adenoviral vector in a primate.

A study was conducted in normal healthy C57BL/6 mice to determine the effect of method of blood collection on clinical pathology parameters and to provide value ranges for these parameters. Males and females were used and were randomly assigned to treatment groups based upon phlebotomy method. The blood was collected using one of four methods: intracardiac (IC), a single attempt at collection from the caudal vena cava (VC), collection from the caudal vena cava with collection of any extravasated blood from the peritoneum (MC), or retroorbital phlebotomy (RO). Evaluation of blood and serum samples was conducted for a number of serum biochemistries including liver function tests and complete blood count with differentials and platelet counts. Female mice demonstrated higher values for red blood cell number, hemoglobin(p < 0.002), hematocrit, alkaline phosphatase, albumin, total protein, and creatinine. Males demonstrated higher values for platelet counts, specific white blood cell numbers (total, neutrophil, lymphocyte, and eosinophil counts), globulin, amylase, and the BUN/creatinine ratio. Overall, the VC method was associated with the least variation in both sexes and appeared slightly better than the IC method for the parameters evaluated. The largest difference between groups was noted for the transaminase levels. While alanine aminotransferase (ALT) values were similar between the IC and VC groups, aspartate aminotransferase (AST) values were associated with less variation for the VC method. Transaminase levels for the MC and RO groups were associated with relatively large ranges and variation. ALT results from the RO method, the only method amenable to repetitive sample collection used in this evaluation, indicate that this is an acceptable method. The results demonstrate the substantial impact that phlebotomy method has on the assay results and that the VC or IC methods provide the most consistent results. The ranges by collection method and sex provided here can be used to select the preferred method of collection when designing a study and for comparison of data obtained with reference ranges. The authors recommend establishment of normal ranges based upon methods employed within an investigator's laboratory.

Ornithine transcarbamylase deficiency (OTCD) is an inborn error of urea synthesis that has been considered as a model for liver-directed gene therapy. Current treatment has failed to avert a high mortality or morbidity from hyperammonemic coma. Restoration of enzyme activity in the liver should suffice to normalize metabolism. An E1- and E4-deleted vector based on adenovirus type 5 and containing human OTC cDNA was infused into the right hepatic artery in adults with partial OTCD. Six cohorts of three or four subjects received 1/2 log-increasing doses of vector from 2 × 10⁹ to 6 × 10¹¹ particles/kg. This paper describes the experience in all but the last subject, who experienced lethal complications. Adverse effects included a flulike episode and a transient rise in temperature, hepatic transaminases, thrombocytopenia, and hypophosphatemia. Humoral responses to the vector were seen in all research subjects and a proliferative cellular response to the vector developed in apparently naive subjects. In situ hybridization studies showed transgene expression in hepatocytes of 7 of 17 subjects. Three of 11 subjects with symptoms related to OTCD showed modest increases in urea cycle metabolic activity that were not statistically significant. The low levels of gene transfer detected in this trial suggest that at the doses tested, significant metabolic correction did not occur.