CLINICAL GENE THERAPY is going through an uncomfortable adolescence. Many observers have already been disappointed by the lack of clear-cut successes with this strategy. To this sense of frustration has now been added a feeling of dismay because of the death of an 18-year-old with ornithine transcarbamylase deficiency, who received an intrahepatic arterial injection of an adenoviral vector that encoded a wild-type version of the defective enzyme.1 The primate study by Lozier et al2 in this issue of BLOOD, further emphasizes the limitations of many adenoviral vector systems for the treatment of human diseases. This group used an adenovector in which the E1 genes had been deleted to inhibit replication of the vector in human cells. After incorporating the human factor IX gene into this, a first-generation vector, they injected the construct intravenously into macaque monkeys. The virus had no effect at its lowest dose, but at the highest, it produced significant levels of human factor IX for approximately 3 weeks. Unfortunately, this benefit was secured at the cost of severe and likely permanent liver damage, manifest by high enzyme levels and persistent hypofibrinogenemia. The experience of Lozier et al2 underscores the difficulty of finding a “Goldilocks” dose for this gene-vector combination. Indeed, as the investigators point out, the window between an effective and a toxic dose is exceedingly narrow. It is important to emphasize that this type of adverse event is not unique to the model used by Lozier et al.2 A baboon receiving 1.2 × 1013 viral particles/kg intravenously also developed severe endothelial injury with coagulopathy and hepatocellular damage.3
Even if toxicity could be avoided, another adenovector-associated problem may become apparent. In the study of Lozier et al,2the transgenic human factor IX produced was highly immunogenic, which is in striking contrast to its lack of effects in the same monkeys when administered as a purified protein. The investigators suggest that the adenovector proteins trigger an acute-phase danger response during infusion that is manifest by secretion of inflammatory cytokines that recruit an immune response both to the adenovector and to any associated transgenic proteins. Regardless of the mechanism, the end result appears to be a cellular and humoral immune response that, within a few days or weeks, destroys the transgenic cells and may neutralize any human factor IX they have produced. Moreover, these effects are seen even in animals that have never been exposed to the adenoviruses on which the vectors are based and thus constitute a primary immune response.4-7 Because most humans already have immunity to adenoviruses, we may expect the secondary responses induced by the vectors to be correspondingly accelerated and more intense than effects generated in naive animals.
Given these limitations, one might be tempted to give up on adenovectors entirely and seek a better alternative. This would be a mistake. In the appropriate setting, and with appropriate modifications, adenovectors have characteristics that make them very desirable for certain gene therapy applications. To appreciate this potential, it is necessary to understand how adenovectors produce their adverse effects and how these may be thwarted. The toxicity of adenovectors occurs in 3 overlapping phases. During the first few hours of infection, the adenovector proteins act directly on the host's defense system, provoking an acute-phase response marked by rapid release of inflammatory cytokines, including interleukin-6 (IL-6) and IL-8, and recruitment of host cellular defenses. Indeed, in some species, this phase of infection may include the activation of mast cells and basophils, which induce IgE-independent acute anaphylaxis.8-10 Over the next 24 to 96 hours, most of the vector-associated toxicity can be attributed to cellular production of adenoviral late proteins involved in the assembly of the viral coat. Although the precise mechanisms of adenovector-related toxicity are unknown, some insights have been gained. Fiber protein, for example, can disrupt cellular endosomes, thereby disabling or killing the infected cell.9-12 Subsequently, the immune system recognizes and destroys adenoviral peptides on the surface of infected cells, mainly by cytotoxic T lymphocytes (CTL)-mediated and natural killer (NK) cell-mediated responses, and this late phase toxicity may be associated with hepatocyte hypertrophy and hepatic fibrosis.4,5
How might these different toxicities be overcome? Most of the earliest efforts focused on reducing the production of viral proteins by infected cells, with the hope of limiting both direct adenoviral toxicity and the antiadenoviral immune response. First generation vectors were deleted in the E1 region, which contains genes whose products regulate transcription of the late adenoviral structural proteins.6 This type of vector was selected by Lozier et al2 for their study of gene therapy to correct factor IX deficiency. Unfortunately, as was amply demonstrated by these investigators, E1 deletion by itself is not adequate to block late protein expression and, hence, the toxic effects of the vector.
Subsequent generations of adenovectors lacked more than one set of adenoviral genes, eg, E1 and E4,13,E2a9 and E4,12 or E1and E3,10,11 and so on (reviewed in Hitt et al14). Although showing less immunogenicity and toxicity and more durable transgene expression in some studies, these modified vectors produced little or no benefit in others.10-14 Indeed, it was anE1/E4-deleted vector that was associated with the death of the young patient with enzyme deficiency.1 The quintessential attenuated adenovector is the so-called helper-dependent or gutless vector, in which virtually all of the adenoviral genes have been removed and replaced with the gene of interest and its promoter, together with irrelevant DNA to allow packaging in the viral envelope.15,16 These vectors can only be made with the assistance of a helper adenovector, which must then be separated from the deleted vector.17 Helper-dependent vectors have shown a much higher therapeutic index than conventional adenovectors in several different models.14,16,18-20 Importantly, they also seem to be much less immunogenic, so that transduced postmitotic cells (eg, muscle or liver) may secrete vector-derived proteins over many months,8,18 a prime consideration in the treatment of many deficiency disorders. On the other hand, these vectors are proving quite difficult to manufacture in adequate quantities for human trials. They likely remain nonintegrating and do not replicate as episomes, so that transduction of rapidly dividing tissues, such as hematopoietic stem cells, becomes feasible only when short-term gene expression is desired.
Aside from gene removal, modification of cell targeting could increase the safety and efficiency of adenovectors. Adenoviruses bind to at least 2 molecules on their target cells: the Coxsackie adenovirus receptor (CAR) and cell surface integrins (usually αvβ3or αvβ5).14,21-23 Binding is mediated by domains on the adenoviral knob protein. Because the sequence and crystal structure of this protein is known, one could select new ligand sequences and incorporate them into positions that would disrupt pre-existing patterns of binding and establish new ones.21-23 In this way, it would be possible to reduce the initial doses of infused virus, thereby decreasing immediate toxicity, and target the circulating virus to organs that are more resistant to the toxicity of adenovectors than are lung, liver, and vascular endothelium, currently the major tissues affected by this route of administration.
Given the liabilities of adenovectors, how can they be optimally applied in the clinic? The narrow therapeutic index, the short duration of expression, and the immunogenicity of E region-deleted vectors, illustrated so clearly by Lozier et al,2 probably mean that they are not going to be suitable for protein replacement therapy in genetic diseases, although continued refinement of late-generation constructs may well change this assessment. In particular, the ability of helper-dependent adenovectors to induce secretion of high levels of the gene product for a prolonged period may make them well suited to treatment of deficiency disorders. Currently available adenovectors are well suited to applications in cancer gene therapy. Because they transduce many different cell types with relatively high efficiency, they can be relied on to transfer toxic or immunomodulatory genes to a broad spectrum of tumors, both ex vivo and locally in vivo. The availability of conditionally replication-competent adenovectors, which will only divide in malignant cells with genetic abnormalities, such as a nonfunctioning p53gene, further increases the utility of such treatment.24Because the aim of cancer gene therapy is to kill tumor cells, the limited duration of transgene expression by adenovectors is not necessarily a problem. Instead, the induction of a destructive immune response against the transduced cell and the transgene product can be used to advantage.
Adenovectors have been used to transfer the thymidine kinase gene to a range of different tumors, including gliomas and prostate or ovarian carcinomas, where the kinase phosphorylates the pro-drug ganciclovir into an active agent, resulting in significant antitumor responses.25,26 The utility of this strategy for metastatic disease is limited by the need to inject the vector locally. Success has also been reported from the use of adenovectors encoding a wild-type p53 protein in a range of tumor types; this reagent is now entering phase III clinical trials.27 Similarly, a conditionally replication-competent adenovirus has produced remarkable tumor regression in relapsed head and neck tumors and is now also entering wider clinical testing.28 Finally, the use of adenovectors to transfer immunostimulatory genes, such as IL-2 and granulocyte-macrophage colony-stimulating factor (GM-CSF), to generate a tumor vaccine has been accompanied by antitumor responses in both adult and pediatric malignant diseases.29,30 None of these approaches has produced significant toxicity. Although few, if any, of the patients treated in this manner are likely cured of their disease, the availability of cancer therapies that appear to be non–cross-reactive with cytotoxic drugs or radiation would significantly increase the number of treatment options. As more is learned about which genes should be transferred to cancer cells, and with continued improvement in vector technology and the ability to incorporate gene therapy into other modalities, we can anticipate substantial contributions from adenovector-driven cancer treatments, extending to hematologic malignancies. In the longer term, the maturation of gene therapeutics from early adolescence to adulthood should see the application of highly modfied adenovectors to a broad range of ailments, which may well include hematologic disorders such as factor IX deficiency.
Address reprint requests to Malcolm Brenner, MD, Baylor College of Medicine, 1102 Bates, Suite C1140, Houston, TX 77030.