The opportunity to integrate cancers that naturally develop in pet dogs within the development path of a novel human cancer drug is further realized through the inaugural study of the COTC. In this report, we describe a targeted delivery of TNFα with an AAVP gene delivery system to tumor blood vessels of pet dogs with spontaneous cancer. Selective AAVP homing, tumor-associated vascular expression of TNFα, systemic safety, and RECIST-based objective responses were observed (same species therapeutic index). These large animal spontaneous cancer models are well suited to inform pre-clinical to clinical transitions necessary for successful drug development and compliment the use of both existing rodent models and human clinical trials.
The COTC is an active network of academic comparative oncology centers, centrally managed by the NCI's Comparative Oncology Program. The COTC designs and implements clinical trials in dogs with cancer with the goal of providing necessary translational data for novel therapies, techniques or devices for future cancer patients. Trials are executed at COTC member institutions, which currently include eighteen veterinary teaching hospitals across the United States. As described above, trials conducted by the COTC may include several biological and clinical endpoints that can be directly integrated into the design of human Phase I and II clinical trials. Although not included in the study presented here, correlative imaging strategies, such as PET/CT or dynamic MRI, with the ability to link these imaging endpoints to tumor or circulating biological surrogates are feasible through the designed infrastructure. Such complex correlative studies may be vetted in a clinically relevant setting through COTC studies, and therein provide a model to evaluate all parts of this process including tissue collection standards and techniques, timing of imaging and biopsy, and assay methodologies prior to early phase human studies.
In the current study, COTC001, we assessed the fluidity and structure of this novel preclinical infrastructure. In the process, we defined the safety and efficacy of RGD-A-TNF
targeting to tumor endothelium in pet dogs with spontaneous cancer. RGD targeted delivery to tumor αV integrins has been previously described 
and this anticipated targeting was the basis of the RGD-A-TNF
vector development 
. In previously published work in mouse xenograft models, we described targeted systemic delivery of RGD-AVVP
vectors expressing either HSVtk or TNF-α, to tumor vasculature 
. Anti-tumor activity has been seen in Kaposi sarcoma, bladder carcinoma, prostate carcinoma and melanoma 
. Although, activity of RGD-A-TNF
has been demonstrated in traditional small animal models, the comparative oncology approach provided unique information regarding the safety of RGD-A-TNF
that would not have been possible in conventional animal safety studies. Since neither tumor nor tumor vasculature are present in healthy animals (i.e. purpose-bred research dogs), a safety assessment in these animals would likely under-report adverse events related to RGD-A-TNF.
Drug-related events reported in our population of tumor-bearing pet dogs were indeed generally mild and self-limiting. In fact, a MTD was not achieved in the single dose cohorts evaluated in the dose-escalation as increasing the dose of RGD-A-TNF
was limited by manufacture process. Most adverse events that were documented were seen either during the administration or within 2 hours of administration in dogs that had received multiple doses of RGD-A-TNF.
Administration-related events included hypersensitivity-like reactions and fever. Given the per-acute nature of these events, it is plausible that such they were related either to undetectable remnant endotoxin in the treatment product or to a specific response to the vector itself. Testing of all production lots of RGD-A-TNF
failed to reveal residual endotoxin; however, one cannot entirely exclude this or other contaminants as a cause of some of the adverse events.
Serial tumor and control tissue biopsies taken before and after the administration of RGD-A-TNF
provided an opportunity to correlate drug exposure with tumor and normal tissue trafficking of RGD-A-TNF.
These experiments validated the tumor-specific targeting of the RGD-A-TNF
in the setting of dogs with spontaneous cancers. We observed vector localization in tumor vascular endothelium in post-treatment tumor biopsies taken 4–6 hours and 4 days after systemic administration of RGD-A-TNF.
Notably, there was a complete absence of the vector in normal tissue biopsies in all treated dogs. Thus, RGD-A-TNF
exploits aberrant disease-related vasculature to target the therapy of interest specifically to the tumor. Thus confirming the safety of this delivery system first observed in our small animal models 
. Consistent with this, warm necropsies from dogs euthanized due to disease progression showed that RGD-A-TNF
targeted tumor vasculature but not blood vessels within normal visceral organs. Importantly, we did not observe RGD-A-TNF
in any of the control tissues analyzed, including liver whereas the presence of RGD-A-TNF
has been previously seen in the liver and spleen of rodents treated with RGD-A-TNF 
. These large animal data are particularly valuable as the risks and benefits for AAVP delivery strategies in human cancer patients are considered. Effective targeting of RGD-A-TNF
was seen at doses from 5×1012
TU. A dose of 5×1012
TU was selected as optimal dose for the multiple-dose study, due to equivalent trafficking, targeting, and safety as well as our inability to produce 1×1013
TU in a timely manner. At this dose vector targeting also resulted in measurable expression of human TNFα. Further support for the biological relevance of the observed trafficking of RGD-A-TNF
along with targeted expression of human TNFα, was provided by the objective anti-tumor activity observed in dogs receiving multiple weekly treatments. RECIST-based responses were observed in two dogs with soft tissue sarcoma and metastatic melanoma; such objective tumor responses are particularly germane, because all dogs in this study had large bulky tumors and were not candidates for conventional loco-regional treatments such as surgery or radiation therapy. We speculate that the observed objective responses were the result of TNFα transgene expression, as no activity was seen in our mouse models from either non-targeted AAVP or targeted-null vector 
. The proposed mechanism for this activity is the induction of endothelial cell apoptosis and hyperpermeability leading to hemorrhagic necrosis in treated tumors 
. This biology will be further explored in planned canine studies using Good Manufacturing Practice (GMP) quality RGD-A-TNF,
via caspase-3 and CD31 staining. However, it is important to note that the future development strategy for this delivery system is not as a single agent. Instead it will likely involve combinational therapies, either dual transgene insertion or adjunct administration of complementary agents. Hence its single agent activity is note worthy.
In summary, the biological complexity of naturally occurring cancers in pet dogs, their size and strong similarities to human cancers, and the availability of a motivated population of pet-owners interested in treatments for their pets with cancer provide an opportunity to now develop a comparative and integrated approach to cancer drug development. 
. The first study of the COTC (COTC001) provides an example of this integration and a functional infrastructure that may deliver trial outcomes in a timely manner. Specifically, COTC001 assessed the safety, selective tumor-specific localization, and anti-tumor activity of RGD-A-TNF
in dogs with measurable malignant cancers. This report supports a new paradigm for rapid intermediate evaluation of agents prior to or after early human trials as a means to creating a more informed and optimal cancer drug development pathway.