Harvey Herschman and Ralph Weissleder led a two-day meeting of experts in new imaging technologies to discuss the current state of molecular imaging applications in clinical practice, clinical research and basic research, and to consider the directions most appropriate for future research and clinical utility of molecular imaging in the diagnosis, staging, and therapy of cancer. For structural convenience, the meeting was divided into four topical themes: 1) clinical applications of molecular imaging, 2) imaging applications to cancer research in small animal models, 3) the development of new imaging probes for optical, magnetic and radionuclide imaging modalities, and 4) the next generation of imaging instruments for research and clinical applications in cancer. In practice, as is common in the Forbeck Forums, the discussions in each session – and for each presentation – were wide ranging and often extended into a number of these topics. Animal cancer models in which tumor initiation, progression, metastasis and response to therapy can be monitored repeatedly and non-invasively permit researchers to perform experiments more easily and efficiently, and – in some cases – facilitate studies that cannot be performed by other means. Anton Berns described development of transgenic mice in which conditionally activated luciferase reporter genes can be induced in a temporal and tissue- specific manner, along with the targeted activation of oncogenes that lead to the tumor development.

He demonstrated the utility of transgenic reporter mice to monitor the onset of tumors in “spontaneous” cancer models, to observe metastases, and to compare alter- native combinations of therapeutic drugs and their order and timing of presentation. Comparisons of therapeutic regimens for doxorubicin and cyclacell 2000 in treating an Rb deficient pituitary tumor model illustrated the value of this approach. Many proteins and molecular pathways are regulated by targeted protein degradation. In particular, ubiquitin ligase mediated protein degradation by the proteasome plays a major role in pathways and cell cycle functions important in cancer. Coupling proteins to luciferase, as described by Bill Kaelin, provides a method to monitor non-invasively protein degradation in murine models, in response to altered oncogenes and in response to therapeutic drugs that target ubiquitin ligase and proteasomal functions. A fusion of luciferase with the cell cycle regulatory protein p27 was used to illustrate the ability of repeated bioluminescent imaging to monitor ubiquitin-proteasome mediated degradation in vivo. A genomics/proteomics approach to studying proteasome/ubiqutin ligase biology is underway with libraries of cDNA fusion proteins to both GFP and firefly luciferase, using high- throughput cell assays to identify proteins targeted for proteasomal degradation by specific ubiquitin ligases. Activation of the endogenous immune system and the use of activated, targeted immune cells as therapeutic agents are both under active investigation for cancer therapy.

Owen Witte discussed the potential to monitor migration and expansion of anti-tumor lymphoid cells in animal modeling of endogenous immune responses with reporter imaging genes and the use of both HSV1-TK (for PET) and luciferase (for bioluminescence) marked targeted T cells for immunotherapy, in a murine sarcoma model system. Remarkably, using microPET imaging of HSV1-TK with positron labeled substrate and systemic glucose metabolism with FDG, the tumor (HSV1-TK) and the adjacent lymph node could be distinguished. This observation provoked a discussion on the need to re-evaluate FDG-PET scanning data, to consider the possibility that activated immune cells as well as tumors are being detecting in clinical studies. Cyclooxygenase 2 (COX-2) is activated in a wide range of cancers, and is thought to be causal in the progression of many tumors (e.g., colon, breast). Early COX-2 expression is proposed as a “biomarker” to distinguish benign tumors that will progress to malignant forms. A “knock-in” mouse in which the firefly luciferase coding region replaces the COX- 2 coding region was reported by Harvey Herschman. COX-2 gene activation in the skin of heterozygous mice can be repeatedly monitored by bioluminescence in response to wounding and inflammatory stimuli. In a carcinogen-tumor promoter skin tumor induction model, COX-2 gene activation can be monitored repeatedly, non-invasively and quantitatively, suggesting the relationship between COX- 2 expression and papilloma-to-carcinoma progression will now be amenable to investigation and elucidation. Marcus Schwaiger emphasized the importance of optimizing currently available clinical imaging technologies to facilitate staging of patients and to both select and monitor alternative therapies in the clinic. He discussed the role of positron labeled tracers such as FDG, amino acids, transporters, choline and nucleotide analogues as probes for tumor response to therapy by PET. Examples of correlations of FLT (a thymidine analogue) with Ki67 histochemistry for measurement of proliferation and a comprehensive summary of the ability of alterations in FDG uptake illustrated the current value of PET in distinguishing responder and non- responders in a variety of tumor types were presented. The importance of developing adequate infrastructures for incorporating molecular imaging into clinical trials was repeatedly emphasized during this presentation. The Von Hippel Lindau (VHL) gene is an ubiquitin E3 ligase that marks the HIF- 1alpha transcription factor for proteasome degradation, as a result of prolyl hydroxylation. In hypoxic conditions, HIF1-alpha is not hydroxylated, and thus not subjected to VHL-dependent, proteasome-mediated degradation.

Michal Safran fused the HIF1-alpha “degron” to firefly luciferase, and created a transgenic mouse in which ODD-luciferase is expressed from a ubiquitous promoter. Both tissue hypoxia and the efficacy of inhibitors of HIF1-alpha prolyl hydroxylation can be monitored in vivo. When crossed to VHL deficient mice, kidney luciferase expression is elevated three to five fold. Preliminary data suggest that developing tumors undergoing an “angiogenic switch” can be monitored by transient hypoxia and consequent elevated luciferase activity. Neal Rosen emphasized the importance of imaging “to find out what is happening in patients”; to determine whether targeted therapies are – in fact – inhibiting their targets and demonstrate appropriate specificity. He illustrated his point by showing how PET imaging of proliferation (with the positron-emitting thymidine analogue FLT) can distinguish MEK 7 inhibitor responses of Raf oncogene activated tumors from lack of response to this agent in Ras oncogene activated tumors. 17-AAG, an inhibitor of HSP90 regulated protein folding, is currently in clinical trials. Dr. Rosen also demonstrated, in murine xenografts, the ability of microPET imaging with a 68Ga positron-labeled antibody to monitor the loss of Her2. Introduction of multimodality imaging and – in particular – CT-PET has provided substantial improvement in the use of molecular imaging for staging patients and for monitoring therapeutic responses. Steve Larson reported on the improvements in the use of positron labeled metabolic probes (FDG), proliferation probes (FLT) and receptor targeted ligands (FDHT for androgen receptor) to image tumor responses to alternative therapies, both in cancer patients and in murine xenograft tumor models. An extensive discussion on the reasons why FDG metabolism changes in response to so many therapies, despite a relatively small effect on hexokinase activity in response to these treatments, followed Dr. Larson’s presentation. Glioblastomas frequently have mutations leading to PTEN loss and amplification and truncation of the EGF receptor.

Ingo Mellinghoff demonstrated that EGF receptor tyrosine kinase inhibitors, effective in only 20% of a clinical cohort of stage IV glioblastoma patients, were far more effective in patients that had both the EGFRvIII mutation and a wild-type PTEN locus; six out of seven patients with this genotype showed tumor regression. In contrast, patients with the EGFRvIII mutation, but with an accompanying PTEN mutation, were refractory to EGFR TK inhibition therapy. Additional mutations in candidate tyrosine kinases are being sought by target sequencing of tumors from several tissues. Phage display is a powerful technique to identify new imaging ligands for specific molecular targets upregulated in cancer. Kimberly Kelly described the use of phage display to identify peptide ligands for the VCAM-1 protein on endothelial cells, for “cancer targets” in pancreatic and colon cancer tumors, and for subpopulations of macrophages, associated with “inflammatory cancers” or macrophage recruitment during therapy induced apoptosis. Specific biological screens were illustrated by using flowing phage to identify internalizing peptides and in vivo screening to identify peptides that bind to atherosclerotic plaques. Discussions included topics that included how to choose among available phage libraries, what makes a good target, identifying appropriate “hits”, identifying the targets, and optimizing imaging agents from phage peptides. Ninety seven percent of clinical PET examinations use FDG. Michael Welch described the preparation of a variety of additional positron- emitting isotopes (60Cu, 61Cu, 64Cu, 124I, 76Br, 77Br, 66Ga, 68Ga and 86Y), and the program established at Washington University to provide such isotopes to other laboratories. The use of Cu(ASTM) as an agent to measure tissue hypoxia, to distinguish responders and non-responders for radiation therapy, and a bromine labeled Sigma 2 receptor probe that preferentially binds to proliferating cells were described as examples of the use of positron-emitting probes, other than 18F, with potential for clinical application.

Dr. Welch emphasized the difficult barrier presented by FDA approval in getting new imaging agents into the clinic, both for trials and for standard-of-care use. The development of lymphotropic magnetic nanoparticles for MRI imaging has revolutionized cancer staging in patients. Current efforts are under way to design newer, highly specific tumor targeted and multimodality agents detectable by MRI and intraoperative imaging.

Ralph Weissleder presented novel procedures for the parallel synthesis of nanoparticle libraries decorated with multiple small molecules, and high-throughput screening to identify specific nanoparticle preparations enhanced for properties such as macrophage uptake (or escape), endothelial cell uptake and selective binding to and/or uptake into cancer cells. Several new imaging agents have been identified for pancreatic cancer, a tumor which currently lacks reliable imaging agents. Discussions centered around target choice, signal amplification, enzyme conversions, activatable agents, multivalency, pre-targeting, optimized pharmacokinetics, background reduction and enhanced affinities. Response to radiation is critically dependent on the state of hypoxia in tumors. Benjamin Williams described the use of electron paramagnetic resonance (EPR) as a non-invasive method to image O2, NO, pH and temperature. His presentation emphasized the use of EPR oximetry to monitor the concentration of molecular oxygen in vivo continuously, over several hours without invasiveness. Dr. Williams discussed potential applications of EPR oximetry, by repeated measurements of pO2, for guiding and monitoring radiation therapy to optimize this therapy by application at periods of relatively high oxygenation. Targeted delivery of a wide variety of “payloads” – imaging agents, therapeutic agents – etc is a major goal in cancer research. Arginine-rich “cell penetrating peptides” are under active investigation as ligands to facilitate delivery of such payloads.

Roger Tsien has constructed hairpin polypeptides in which such polyanionic domains are coupled by cleavable peptides to inhibitory cationic domains. Cleavage of the linker, typically by proteases, releases the targeting domain and permits “transduction” of its cargo to cells in the immediate vicinity. Several murine model systems, including a murine transgenic mammary tumor and human tumor xenografts that express matrix metalloproteases demonstrated the in vivo applicability of this targeting procedure. Targeting of technetium and gadolinium cargos apparently accumulates in cells in a protease-dependent fashion. Many promising molecular targets for new therapeutics and imaging agents are not suited to bind small, cell-permeable organic molecules, and are said to be “undruggable.” The laboratory of Greg Verdine focuses on developing new chemistry platforms that address undruggable targets –drugging the undruggable. They have returned to using molecules with stereochemical centers to develop libraries with greater diversity. Dr. Verdine also described a method of “molecular stapling” to force polypeptides into alpha helix configurations for the development of polypeptide drugs and probes. He illustrated examples for targeting Myc protein and development of a stapled peptide that mimics the BID protein and inhibits apoptosis. Importantly, these stapled peptides are highly cell membrane permeable (2-3 orders of magnitude compared to parent sequence). He also reported preliminary results with small RNAs that target non-Watson-Crick base pairing, to inhibit formation of undesirable RNA molecules. One limitation of many radiopharmaceuticals is their relatively low target-to- background ratio in vivo. Pretargeting strategies such as Neorex’s avidin-biotin system have previously been developed to improve target-to-background ratios but have been abandoned due to other concerns.

Claude Meares introduced an alternative amplification strategy based on developing imaging agents with “infinite affinity” to targets of choice. The basic principle of achieving this is by promoting covalent reactions that would not otherwise be favored. This was illustrated with an engineered monoclonal antibody system. By coupling this reagent to two single chain antibodies, a multivalent, high affinity imaging agent could be created. MicroPET, microCT, and similar instruments have brought these non-invasive imaging techniques to small animal cancer models.

Simon Cherry discussed tradeoffs in sensitivity, spatial resolution, speed of imaging, quantification, high-throughput capability, cost and other factors that are considered in the design of next-generation imaging instruments, and asked the biologists present to consider these alternatives in their evaluation of how to proceed in instrumentation design. He also discussed the question of “Why can’t PET be better”, and clarified the theoretical and practical limits for resolution and sensitivity. Dr. Cherry pointed out the often unappreciated – by biology end-point users – radiation doses received by mice in microCT or combined microPET-microCT studies, and the potential effects of these doses on the biological processes under investigation.

Vasilis Ntziachristos evaluated applications and limitations of current optical fluorescence imaging systems. He differentiated reflectance imaging (photography), postprocessed reflectance imaging and quantitative tomographic imaging methods. He presented the principles of optical tomography for in vivo fluorescence imaging, using techniques that involve laser excitation, subject rotation, time gating and reconstruction algorithms that currently provide sub millimeter resolution with murine models. Fluorochrome detection threshold in the near infrared is currently in the femtomole range. He also presented newer developments to image GFP quantitatively, demonstrating a detection threshold of about 20,000 tumor cells in the esophagus of a mouse. Discussion continued on issues of resolution, sensitivity, etc. in the context of biological questions. Optical imaging has another unique application in medicine by bringing microscopic resolution to endoscopic and intraoperative procedures

Remy Brossel described the development of several generations of “fiber-optic confocal fluorescence microscopes”, to image at cellular resolution in inner organs and body cavities. He illustrated the potential utility of the real-time confocal microscopy technique in mice measuring GFP expression and/or autofluorescence. He also presented first clinical data showing that the fiber optic method can indeed be combined with clinical endoscopy and reveal malignant and subtle premalignant “spots.” Dr. Brossel postulated use of the technology for detecting early cancers, follow- up of “suspicious” lesions, monitoring tumor angiogenic properties and imaging with fluorescent imaging probes currently being developed.

‍