Targeted Molecular Imaging in Oncology: Focus on Radiation Therapy

Abstract

Anatomically based technologies (CT, MR, etc.) are in routine use in radiotherapy for planning and assessment purposes. Even with improvements in imaging, however, radiotherapy is still limited in efficacy and toxicity in certain applications. Further advances may be provided by technologies that image the molecular activities of tumors and normal tissues. Possible uses for molecular imaging include better localization of tumor regions and early assay for the radiation response of tumors and normal tissues. Critical to the success of this approach is the identification and validation of molecular probes that are suitable in the radiotherapy context. Recent developments in molecular imaging probes and integration of functional imaging with radiotherapy are promising. This review focuses on recent advances in molecular imaging strategies and probes that may aid in improving the efficacy of radiotherapy.

Introduction

The last few decades have seen a remarkable change in our understanding of the genetic, molecular and cellular basis of disease. With this new understanding there has been a shift in the management of cancer from an organ-system approach to targeting specific molecular abnormalities. Until a decade ago our understanding of cancer has been descriptive and has relied on information gathered from 10–10,000 cells at a time, collected from a heterogeneous tumor at a single instant. Recent developments in molecular imaging allow one to follow the same number of cells but the changes can be monitored dynamically. Using parametric imaging it is possible to map the variations in the architecture and physiology of disease in a three dimensional mode. This closer look at dynamic processes, however, is limited to very few cellular events because such dynamic visualization requires a clear understanding of the molecular nature of the process and the proper tools to study it.

Developments from the human genome project promise to produce more therapeutic targets. Despite this remarkable progress there has been a steady decline in the number of new agents reaching the clinic. One explanation for this is the lack of appropriate biologically relevant screens and models. Recent developments in genetic models of cancer have been addressing this problem. Many excellent reviews on this issue exist (1–4). Another reason is the lack of accurate prognostic markers and a way to monitor them. Imaging may enable more rapid translation of promising therapeutic agents to the clinic by demonstrating that drugs that are destined to fail are identified early and inexpensively in the development process and by enabling selection of appropriate patients, i.e., those with a phenotype that may benefit from the targeted agent, on whom to initiate clinical trials.

Recent advances in small animal imaging now allow highly specific and sensitive detection of cellular and molecular events non-invasively. Molecular imaging modalities such as positron emission tomography (PET) and single photon emission computed tomography (SPECT) provide spatial resolutions on the order of 1–2 mm. With anatomic small animal imaging modalities such as magnetic resonance imaging (MRI) and computed tomography (CT) it is possible to achieve spatial resolutions on the order of 50 μm. Significant advancements have been made clinically as well, particularly in the area of hybrid technologies such as PET-CT and SPECT-CT, which allow CT to complement the low resolution molecular imaging techniques. The details of multimodality imaging devices and their clinical and preclinical applications are beyond the scope of this review but have been addressed in detail elsewhere (5–9).

While developments in generating imaging targets and in imaging technology improve our understanding of biology, there is an unmet need in the development and characterization of new imaging probes and reagents for specific biological processes. Also, there is a strong interest in using molecular imaging not only in drug development but also in targeted treatment planning and monitoring – including image-guided radiation therapy. In this article we will focus on the current paradigms for targeted multimodality molecular imaging and their relevance to monitoring radiotherapy for cancer.

Metabolism

The classical PET radiotracer used in the clinic is [¹⁸F]2-fluoro-2-deoxy-d-glucose (FDG). Many tumors are known to have high glycolytic rates compared to normal tissue (10). FDG exploits that abnormal increase in glucose metabolism to image tumors. FDG is transported into tumor cells in increased amounts as a result of the upregulation of glucose transport proteins, i.e., GLUT1 and rate-limiting enzymes, e.g., hexokinase. Once inside the cell, FDG is phosphorylated into FDG-6-phosphate but does not undergo glycolysis due to the presence of a fluorine substitution in the molecule. FDG-6-phosphate accumulates and becomes metabolically “trapped” within the cell. FDG is routinely used in the clinic for diagnosis, staging, detection of recurrent disease and monitoring therapy of cancer. Many excellent reviews on FDG use in clinical monitoring have been published in the literature (11–13). The literature indicates an 84 to 87% specificity and 88 to 93 % sensitivity of FDG-PET in various oncologic applications (13).

As with other aspects of cancer imaging, FDG-PET has been the most widely employed molecular imaging technique for applications to radiotherapy. One of the main thrusts has been to use FDG-PET to delineate tumor boundaries more accurately for treatment (14, 15). This has proved particularly useful in lung tumors where a large inter-user variability is observed in some tumors due to atelectasis (16, 17). The addition of FDG-PET has a strong effect on the size of the region that is identified as tumor. Compared to using CT alone, the inclusion of PET data changes the size of delineated gross tumor volumes by approximately 20% to 40% in approximately half of non-small cell lung cancer (NSCLC) patients (15). FDG-PET has also been employed for tumor definition sites such as head and neck (18, 19), primary brain (20), cervical (21), rectal (22) and esophageal cancers (23) and lymphoma (24).

Implicit in the use of PET for target definition is the need for a robust means of delineating tumor boundaries. One widely used method is to set a voxel value threshold above which all areas are considered tumor (25). The appropriate threshold level is not known a priori, however, and recent studies have shown that the volume recovered is extremely sensitive to the exact threshold used (26). In other words, the extent of the tumor depends on the window and level used. Given those uncertainties, further development, such as the creation of uniform standards for quantification, is needed before PET can be used confidently to outline target volumes. One interesting avenue of investigation includes validation studies comparing pathological specimens with imaging. One recent study employed a careful alignment of pathological specimens from laryngeal cancers with FDG-PET, CT and MR images obtained prior to surgery (25). That study showed significant differences in tumor volumes between the modalities (with FDG-PET returning the largest volumes). Although all of the modalities overestimated the size of the tumor, each one also missed tumor extension into some regions. Further study is clearly required.

The second way that PET may be used in radiotherapy is as a predictor or measure of therapeutic response. In head and neck cancer significantly better local control and disease-free survival have been shown for lesions with lower standard uptake values (SUVs). SUV is defined merely as the radioactivity concentration in the tumor/injected amount/patient weight) (27–29). Decreases in SUV indicating adequate therapy have also been reported for NSCLC (30–33). FDG-PET has been used extensively to measure the response to radiotherapy in head and neck cancers (34, 35), as well as in lung (31, 36) and brain (37–39) tumors.

Monitoring therapy-induced changes in tumor metabolism with FDG is not without limitations. FDG accumulation can be nonspecific, particularly in cases of inflammation and infiltration of macrophages, which also readily sequester FDG due to high glucose metabolism (40). Therapy-induced cellular inflammation was shown to cause a temporary increase in FDG uptake in esophageal cancer patients treated with chemoradiotherapy (41). In addition, physiologically high normal “background” metabolic activity can render the quantification of FDG uptake difficult in some areas of the body such as brain (Figure 1). Furthermore, glucose metabolic changes may not be apparent in targeted therapies, requiring additional imaging probes to complement FDG to monitor targeted cancer treatment. On the microscopic level FDG uptake has been positively correlated with hypoxia and negatively correlated with proliferation and perfusion in an experimental model (42). It is therefore likely that FDG uptake is sensitive not only to metabolic activity but other processes as well. We explore more specific radiotracers below.

Figure 1. Comparison of FDG and FLT.

An 11-year-old male with a germ cell tumor of the basal ganglia. MRI shows subtle changes (arrow) in the right basal ganglia. On FDG-PET, the right basal ganglia lesion shows slightly decreased uptake compared with the contralateral basal ganglia but increased uptake compared with normal white matter. 3′-Deoxy-3′-[¹⁸F]fluorothymidine (FLT)-PET, however, reveals intensely increased uptake, suggesting the presence of a malignant tumor (arrow). Based on the FLT-PET results, a stereotactic biopsy was performed in the right putamen. could be performed in the right basal ganglia (63).

From: Choi SJ, Kim JS, Kim JH, et al. [18F]3′-deoxy-3′-fluorothymidine PET for the diagnosis and grading of brain tumors. Eur J Nucl Med Mol Imaging. Jun 2005;32(6):653–659.

Interestingly, non-specific FDG uptake may actually be useful in some clinical situations. A recent example is a study of breast cancer lumpectomy patients in the context of adjuvant partial breast irradiation where it is often difficult to identify the lumpectomy bed on CT alone. This study showed an increased FDG uptake in the lumpectomy cavity, presumably due to inflammation (Figure 2) (43). The FDG-PET distribution can aid in the definition of the lumpectomy bed. The volume of the lumpectomy cavity as defined on PET-CT was larger than that defined on CT. A radiation treatment plan based on CT alone would not adequately treat the larger PET volume. The plan can be redesigned to cover the larger PET volume, however, while still respecting the dose constraints to normal tissues.

Figure 2.

Representative axial slices from CT (left) and FDG-PET (right) scans of a breast cancer patient after lumpectomy surgery. The lumpectomy cavity was delineated for further treatment with the partial breast irradiation technique. Contours based on FDG-PET-CT (blue) were generally larger than those defined on CT alone (green). (From Ford et al., 2007).

In addition to glucose metabolism, cancer cells also have an accelerated protein metabolism and upregulated amino acid transport, which could be monitored by labeled amino acids (44–46). Radiolabeled methionine and tyrosine analogs O-(2-fluoroethyl)-l-tyrosine (FET), 6-fluoro-l-m-tyrosine (FMT), l-3-iodo-α-methyltyrosine (IMT) have been used (47) for imaging cancer. Both methionine and IMT have demonstrated very high sensitivity and specificity in detection of various tumors, particularly brain (48, 49). One of the problems associated with amino acid-based imaging is the metabolism of these compounds, with generation of a plethora of radioactive metabolites that render radiotracer kinetic modeling difficult. Imaging with radiolabeled amino acids is advantageous during radiotherapy monitoring – perhaps more so than FDG-PET – due to less uptake by inflammatory tissue. That is particularly the case for brain tumors where it is critical to differentiate recurrent tumor from inflammatory and other changes due solely to radiation.

Proliferation

Cellular proliferation may be measured in many ways, including use of radiolabeled lipid precursors (50) and nucleosides (51, 52). While lipid precursors tend to have more than one metabolic pathway, nucleosides, particularly thymidine and its analogs, are quite unique because they incorporate into DNA and provide a readout of DNA synthesis. Preclinical and clinical studies (53–55) of thymidine labeled with carbon-11 (20.2 min) showed rapid catabolism generating considerable amounts of recirculating radioactive metabolites making [¹¹C]thymidine impractical for routine clinical use (56–59).

Nucleosides with a substitution of fluorine at the 2′- or 3′-position of the pentose sugar are more resistant to degradation by thymidine phosphorylase. One promising agent is 3′-deoxy-3′-[¹⁸F]fluorothymidine (FLT). FLT is resistant to degradation and is trapped intracellularly after phosphorylation by tightly regulated S-phase cytosolic thymidine kinase (TK1) (60, 61). Studies have demonstrated correlation of FLT uptake with cellular proliferation in cells, animals and humans. FLT has been used to detect and monitor changes in proliferation using PET in various cancers including colorectal (62), brain (63), and lung (64). For example, Kenny et al. have demonstrated that early changes in proliferation in patients with breast cancer can be monitored by FLT uptake just after one week of chemotherapy and preceding any observable changes in tumor volume (65).

Sugiyama et al. have demonstrated that uptake of FLT and changes in FLT uptake after radiotherapy were more pronounced than FDG and correlated well with proliferating cell nuclear antigen (PCNA) labeling index (66). In another study Apisarnthanarax et al. demonstrated the use of FLT in detecting changes in proliferation after chemoradiotherapy in an experimental model of esophageal carcinoma (67). Yang et al. have also employed FLT to assess early response to radiation in a xenograft model and compared it to changes in FDG uptake (68). This study found a marked FLT response to single doses of radiation at early time points (1–2 days post-therapy) which were larger than those measured with FDG-PET.

Although FLT is very promising as an agent for imaging proliferation, it is not readily incorporated into DNA and therefore does not reflect the total DNA synthesis. That may be a limitation to monitor the therapeutic effectiveness of cytostatic agents. Other thymidine analogs that incorporate into DNA and therefore demonstrate the efficacy more directly such as (1-(2′-deoxy-2′-[¹⁸F]fluoro-β-d-arabinofuranosyl)thymine) (FMAU) (69) and 1-(2′-deoxy-2′-[¹⁸F]fluoro-1-beta-d-arabinofuranosyl)-5-bromouracil (FBAU) (70) are being investigated.

Elevated choline metabolites such as phosphocholine and phosphatidylcholine, observed in tumors are a result of increased activities of choline transporters and choline kinase associated with increased cell membrane synthesis and tumor cell proliferation (71). The highly active phospholipid metabolism of tumor cells has been targeted with [¹¹C]choline as a marker of proliferation (72). [¹¹C]Choline imaging has been applied for the detection of several cancers, particularly brain and prostate (73, 74). To increase the effectiveness of lipid metabolism imaging, efforts have been made to develop [¹⁸F]-labeled choline analogs (75, 76). Similarly, Inhibitors of choline kinase and choline transporters such as hemicholinium-3 are being investigated as potential imaging agents (77). Despite promising initial results obtained with [¹¹C]choline, there are conflicting reports on its correlation with measures of proliferation (78, 79).

Significant progress has been made in using MR technology to detect the changes in tumor phospholipid metabolism (80, 81). Several excellent reviews on the use of ¹H-, ¹³C- and ³¹P-MR for choline metabolite detection in cancer are available (71, 82). Knowledge of phospholipid metabolism has been extended to characterize the tumor progression and radiotherapy in gliomas (83, 84).

Hypoxia

Solid tumors, due to imperfect and chaotic vasculature, develop regions of low oxygen concentration regions that are hypoxic. In the majority of solid tumors hypoxia is the basis for angiogenesis, metastasis and therapy resistance – particularly radiation therapy. The ability to image hypoxia has potential therapeutic applications. One hope has been that radiation therapy can be tailored to irradiate hypoxic tumor regions preferentially (85). That could theoretically give enhanced tumor control even if the tumor only has modest hypoxic subfractions (86). One difficulty with this approach, however, is that the hypoxic regions appear to change in response to radiation even at early time points as seen in a model system (87). Work is ongoing to understand these effects in more detail.

Another possible means of exploiting hypoxic conditions is the use of a prodrug activating gene that can be specifically triggered by hypoxia-specific promoters or bioreductive drugs that could be specifically activated at low tumor oxygen concentrations (88–93). Because hypoxia is a marker for poor prognosis, using hypoxia as a diagnostic marker may also be helpful for better chemotherapeutic and radiation treatment outcomes (94).

[¹⁸F]Fluoromisonidazole, or 3-[¹⁸F]fluoro-1-(2′-nitro-1′-imidazolyl)-2-propanol (FMISO), accumulates in regions with pO₂ < 10 mm Hg and is retained in a nitroreductase dependent manner. FMISO has been evaluated in preclinical (95) and clinical settings and shows correlative variation in uptake with hypoxic status in various tumors (93, 96–98). The nitrosoimidazole-based SPECT imaging probe, ^99mTc-labeled 4,9-diaza-3,3,10,10-tetramethyldodecan-2,11-dione dioxime [^99mTc]HL91, has also been useful in predicting hypoxia status and changes during radiotherapy (99, 100).

Another hypoxia marker that has been extensively investigated is a metal chelating agent Cu-diacetyl-bis(N4-methylthiosemicarbazone) (Cu-ATSM) (101). Uptake of Cu-ATSM has reliably detected hypoxic regions within NSCLC (102) and cervical tumors and used in image-guided intensity-modulated radiation therapy (IMRT) of advanced cancers (103). However, comparison studies in murine squamous cell carcinoma and other models showed better correlation between FMISO uptake changes and hypoxia levels than Cu-ATSM (101, 104). Cu-ATSM, labeled with ⁶⁴Cu, due to its favorable β-particle emissions, has also been used as a therapeutic agent (105–107). Studies have indicated that the Cu-ATSM uptake is particularly sensitive to the time at which images are acquired, probably due to the passive transport across cell membranes (108). A new generation of nitrosoimidazoles and labeled azomycin analogs such as [¹⁸F]fluoroazomycin arabinoside (FAZA) and iodo-azomycin-galactosides (IAZG) are under investigation as PET and SPECT imaging probes to determine the predictive value of hypoxia directed treatments and chemoradiotherapy (109, 110).

Apoptosis

Apoptosis, or programmed cell death, is a tightly regulated and genetically defined cellular process in vivo. The externalization of phosphatidylserine (PS) to the cell surface is a hallmark of the caspase-3 activated apoptotic process (111, 112). Externalized PS can be detected using a 40kDa vesicle protein called annexin, which has been extensively used in molecular imaging for imaging apoptosis (113–116). ^99mTc-labeled annexin uptake after chemotherapy has been shown to be a predictor of tumor response and prognosis in some human cancers (117). Radiation induced apoptosis also correlates with the uptake of radiolabeled annexin in follicular lymphoma patients undergoing radiotherapy (118). Annexin labeled with near infrared optical imaging probes has also been used for tumor detection and treatment response in animal models (119, 120).

Another molecule that binds to externalized PS is C2 domain of synaptotagmin (121). Synaptotagmin conjugated with SPIO nanoparticles has been used to detect apoptosis in murine lymphoma models using MRI (122).

The preceding sections have dealt with imaging agents and related mechanisms (metabolism, proliferation and hypoxia, and apoptosis) that have been studied to at least some extent in the context of radiation therapy. We now consider processes that are also amenable to the development and use of targeted imaging agents but have not yet been studied with respect to the radiation therapy, namely: receptor-based targets, gene expression and angiogenesis. These processes are clearly relevant for radiotherapy and could be used either to predict response to therapy, provide an early measure of outcome, or possibly allow tailored therapy via dynamic measurements.

Receptor-based imaging

Receptors play a pivotal role in signal transduction and proliferation (123, 124). Expression levels of estrogen receptor (ER) are regularly used for prognosis in patients with breast cancer and to predict who will respond favorably to endocrine therapy (125). [¹⁸F]Fluoroestrodiol (FES), a ligand that binds ER, has been shown to predict response to endocrine therapy (126). Dehdashti et al. found that FES-PET imaging was useful in predicting the ultimate response to tamoxifen treatment within 7 to 10 days of initiating this form of endocrine therapy (127). Similarly, androgen-dependent prostate carcinomas can be screened for androgen receptor status with the PET imaging agent, [¹⁸F]fluoro-dihydroxytestosterone (FDHT)(128), with promising initial clinical results.

Growth factor receptors such as epidermal growth factor receptor (EGFR) are also becoming viable imaging targets. EGFR is overexpressed in NSCLC, bladder, cervical, ovarian, kidney and pancreatic cancers. Several SPECT-, NIRF- and PET-based EGFR inhibitors and antibodies are being investigated. Promising results were demonstrated from immunoimaging with an anti-EGFR antibody in patients with NSCLC (129–131). Small molecule inhibitors of EGFR such as gefitinib and others are also being functionalized for imaging (132–134).

HER2/neu receptor is significantly overexpressed in breast, ovarian, gastric, lung, bladder, kidney and several other cancers (135). HER2/neu receptor is a predictive marker for therapy response in breast cancer patients (136). HER2/neu-based imaging agents, mainly radiolabeled antibodies, are being investigated for classification of receptor status (137). The first patient study with [¹¹¹In]trastuzumab allowed identification of the HER22/neu positive tumors in patients with metastatic breast cancer (138).

Neuroendocrine tumors express a high density of somatostatin receptors. Scintigraphic somatostatin receptor imaging of neuroendocrine tumors with radiolabeled somatostatin analogs such as [¹¹¹In]DTPA-d-Phe¹-octreotide is a routine clinical diagnostic tool (139). Currently ⁶⁸Ga- and ⁶⁴Cu-labeled analogs are also being investigated as PET radiotracers (140–142) in order to take advantage of the inherently superior sensitivity of PET.

MR imaging of tumor receptors is rather challenging due to the combination of low concentration of such receptors and the inherent limitation of sensitivity for MR. However, novel strategies to improve signal intensity for MR-based molecular imaging have been developed to overcome this limitation. For example, tumors expressing an engineered transferrin receptor have been imaged using a superparamagnetic iron oxide (SPIO)-labeled transferrin receptor ligand (143). In another study, Artemov et al. have targeted the HER2/neu receptor overexpressed in NT-5 breast cancer cells with a two-step avidin-biotin system. In this approach animals were pretreated with a biotinylated HER2/neu antibody. After clearance of the unbound antibody streptavidin-conjugated gadalonium-chelates were administered. That approach allowed controlled signal amplification for specific detection of the HER2/neu positive tumors (144, 145).

Gene expression

The future of gene therapy lies in safe and efficient delivery and controlled expression of targeted genes at the tissue of interest. Gene delivery remains a considerable challenge because of poor specificity, inadequate transfer efficiency and lack of correlation between expression and function of the genetic material introduced (146). To achieve a better therapeutic response, quantitative and noninvasive monitoring of distribution and knowledge of the magnitude and duration of gene expression are essential. To improve the assessment of gene delivery, several imaging platforms employing various reporter gene strategies and reporter probes have been explored. Some of the most common reporter gene systems based on either enzymes, receptors or transporters in nuclear imaging are 1) the sodium/iodide symporter (NIS) system with ¹³¹I (147, 148), 2) the wild type and mutant dopamine receptor (D₂R) system with 3-N-(2-[¹⁸F]fluoroethyl)spiperone (¹⁸F-FESP) as the imaging probe (149, 150), 3) wild type herpes simplex thymidine kinase (HSV1-tk) system with radiolabeled analogs of 1-(2′-deoxy-2′-fluoro-1-beta-d-arabinofuranosyl)-5-iodouracil (FIAU) (151, 152) and 4) mutant herpes simplex thymidine kinase (HSV1-sr39tk) system with 9-(4-[¹⁸F]fluoro-3-hydroxymethylbutyl)guanine, [¹⁸F]FHBG (153, 154). By tagging an appropriate reporter gene, many cellular processes such as initiation of transcription (155), transcriptional regulation (156), translation of mRNA (157), protein-protein interactions (158), and antitumor immune response (159) etc., have been imaged.

The first imaging-based readout of human gene therapy was reported by Jacobs et al. in glioma patients undergoing stereotactic gene therapy. Imaging of the HSV-tk expression was undertaken with [¹²⁴I]FIAU-PET (206). Only one out of the five patients showed an indication of radiotracer accumulation due to transgene expression. In another study Peñuelas et al. monitored HSV-tk expression in patients with hepatocellular carcinoma after an intratumoral injection of a recombinant adenovirus. The gene expression levels monitored using [¹⁸F]FHBG showed a viral dose-dependent accumulation of the radiotracer (160).

Instead of using an exogenous reporter gene system, Fu et al. utilized the viral thymidine kinase expression in Epstein-Barr virus and other gammaherpesvirus-associated tumors in a preclinical model system. The authors induced the viral TK expression using bortezomib and monitored the expression levels with [¹²⁵I]FIAU-SPECT (Figure 3). This mechanistic approach may be useful in targeting radiolabeled and therapeutic agents to virus-associated tumors (161).

Figure 3. Viral TK induction.

Time course of uptake of [¹²⁵I]FIAU by Burkitt’s lymphoma xenografts [EBV(+) Akata] following treatment with bortezomib as assessed by planar gamma scintigraphy in vivo. Large arrows, tumors. The dark area (A, small arrow) represents lead shielding of bladder to improve the dynamic range of the images. Each animal has one tumor placed in the hind limb. A, no tumor uptake is evident in animals pretreated with PBS only (control). B, tumors are visualized at later time points in the pretreated animals (2 μg/g bortezomib).

From: Fu DX, Tanhehco YC, Chen J, et al. Virus-associated tumor imaging by induction of viral gene expression. Clin Cancer Res. Mar 1 2007;13(5):1453–1458.

Hypoxia induces a number of genes and is a significant contributor to the radiation resistance observed during cancer treatment (162). Greco et al. have successfully achieved a therapeutic gain in vitro by employing vectors containing hypoxia responsive elements and radiation-responsive CArG elements from the Egr1 early growth response gene. This novel strategy utilizes the adverse gene expression triggered in response to hypoxia and/or ionizing radiation stimuli (89, 163). The effectiveness of these combined therapies in vivo are yet to be determined.

MR-based gene expression imaging is still in its infancy with only a few endogenous reporter genes in development (164, 165). Recently Gilad et al. have developed a chemical-exchange saturation transfer (CEST)-based lysine-rich protein that could be turned on or off by selectively saturating at the exchangeable proton-resonance frequency. The authors demonstrated the ability to image CEST reporter gene stably expressed in brain xenografts using MRI (166).

Angiogenesis

Angiogenesis, the development of new blood vessels from preexisting blood vessels, is fundamental to tumor growth and transition from a benign into malignant state (167, 168). As a result angiogenesis has become an important therapeutic target and several novel therapies based on either the existing tumor blood vessels (vascular disrupting agents) or tumor blood vessel development (antiangiogenic agents) have been developed (169). Since the effect of the antiangiogenic drugs seems to be primarily cytostatic, new molecular measures of biological response based on membrane proteins selectively expressed by angiogenic vessels have been explored (170). In addition, other physical parameters such as tumor vascular permeability, perfusion, blood volume, and microvessel density have been investigated with MRI, CT and PET (171–173). Some of the molecules germane to angiogenesis that have gained particular attention are vascular endothelial growth factor (VEGF), VEGF receptors (VEGFR), α_vβ₃ integrin, matrix metalloproteinases (MMPs) and thrombospondin-1 receptor (170, 174).

VEGF, a mitogen, plays a critical role in stimulating endothelial migration, proliferation and angiogenesis (175). Thus VEGF-targeted imaging may provide valuable information on tumor angiogenesis and antiangeogenic treatments. Several VEGF isoforms (VEGF_121, VEGF ₁₆₅) were either labeled directly or tagged with chelating agents to label with positron (176) or gamma emitters (177–179) to visualize the VEGFR expression in preclinical models. Some of the most valuable information was derived from PET imaging studies of the ¹²⁴I-labeled humanized mouse monoclonal anti-VEGF antibody, HuMV833. This study in advanced cancers revealed a marked variation in the pathophysiologic behavior of tumors, even within the same patient, to antibody distribution, clearance and biological response demonstrating the need for a more tailored approach to using cytostatic antiangiogenic agents (180).

Integrins, a family of cell adhesion molecules, play a key role in endothelial cell migration and survival during angiogenesis. Of the family of integrins, α_vβ₃ integrin is significantly upregulated on activated endothelial cells during angiogenesis and interacts with extracellular matrix proteins with an exposed RGD (arginine-glycine-aspartic acid) motif such as vitronectin, fibronectin and fibrinogen (181, 182). Several novel radiotracers containing RGD sequences labeled with ¹⁸F, ⁶⁴Cu, ^99mTc, ¹¹¹In, or ⁹⁰Yr to target specifically tumor integrins have shown high tumor/background ratios in preclinical models (183). Cyclic glycosylated pentapeptides labeled with ¹⁸F have been successfully used to image α_vβ₃ integrin expression in cancer patients (Figure 4) (184, 185).

Figure 4. Correlation of tracer accumulation and α_vβ₃ expression.

(A–C) patient with a soft tissue sarcoma dorsal of the right knee joint. (A) The sagittal section of a [¹⁸F]galacto-RGD PET acquired 170 min p. i. shows circular peripheral tracer uptake in the tumor with variable intensity and a maximum SUV of 10.0 at the apical-dorsal aspect of the tumor (arrow). (B) The image fusion of the [¹⁸F]galacto-RGD PET and the corresponding computed tomography scan after intravenous injection of contrast medium shows that the regions of intense tracer uptake correspond with the enhancing tumor wall, whereas the non-enhancing hypodense center of the tumor shows no tracer uptake. (C) Immunohistochemistry of a peripheral tumor section using the anti-αvβ3 monoclonal antibody LM609 demonstrates intense staining predominantly of tumor vasculature. (D–F) patient with malignant melanoma and a lymph node metastasis in the right axilla. (D) The axial section of a [¹⁸F]galacto-RGD PET acquired 140 min p. i. shows intense focal uptake in the lymph node (arrow). (E) Image fusion of the [¹⁸F]galacto-RGD PET and the corresponding computed tomography scan after intravenous injection of contrast medium. (F) Immunohistochemistry of the lymph node using the anti-αvβ3 monoclonal antibody LM609 demonstrates intense staining predominantly of tumor cells and also blood vessels.

From: Noninvasive Visualization of the Activated αvβ3 Integrin in Cancer Patients by Positron Emission Tomography and [18F]Galacto-RGD Haubner R, Weber WA, Beer AJ, Vabuliene E, Reim D, et al. PLoS Medicine Vol. 2, No. 3, e70 doi:10.1371/journal.pmed.0020070

Other imaging modalities have also been extensively used for integrins; MR imaging with paramagenetic nanoparticles (186), antibody coated liposomes (187) and Gd-perfluorocarbon nanoparticle-conjugated antibodies (188) have been used for targeted tumor α_vβ₃ imaging. Near infrared-based optical imaging agents (189) and α_vβ₃ targeted microbubbles for ultrasound have also been used to image tumor angiogenesis (190).

Matrix metalloproteinases, a class of Zn⁺²-dependent endopeptidases, play a crucial role in cancer cell extravasation and invasion (191, 192). Of the 25 known MMPs, gelatinases (MMP-2 and 9) are preferentially detected in tumors, are capable of collagen IV degradation and are of particular interest as imaging targets. Several inhibitors of MMPs, including tissue inhibitors of MMPs and MMP-specific small molecule inhibitors labeled with several PET and gamma emitting radionuclides, have been used for imaging (193, 194). So far, studies in preclinical and clinical imaging with MMP inhibitors have not shown convincing MMP specific tumor uptake (195–197). In contrast to experience with radiolabeled probes, near infrared optical imaging “smart” probes designed to be activated by specific MMPs have shown MMP specific tumor uptake (198, 199). Other endothelial markers such as endostatins and E-selectin are under development and show some promise (200, 201).

Future of targeted molecular imaging for radiation therapy

Molecular imaging is emerging as a powerful technique in the management of cancer and radiation treatment planning. Although the collaborative effort between various science and engineering disciplines is greater now than ever before, more integration is necessary. With the exception of FDG, the imaging agents described above are largely unstudied in the context of radiation therapy. Early studies indicate that some of these compounds may be useful either for monitoring or predicting the response to radiotherapy or possibly for improving treatment planning (e.g. better tumor delineation). More studies of the properties of these agents are needed that address the radiation response and other issues relevant to radiation therapy.

Immediate clinical impact may derive from extending current chemoradiotherapy approaches to new targeted therapies, such as the use of EGFR inhibitors (202), to achieve a synergistic effect. Also, development of radiation sensitive gene therapy approaches as described in hypoxia section above, and radiation-triggered release of chemotherapeutic agents or radiation-sensitive prodrugs could pave the way for more localized treatment delivery.

From a probe development standpoint, it may be time for chemists to concentrate more on developing dual or multiple use agents to move into “theragnostic” imaging. Although radiation treatment planning using molecular imaging techniques could be considered theragnostic imaging, another meaning could incorporate true dual-use probes – that identify tissues to be treated and treat them concurrently. Such an agent may be for gene therapy, where the gene, which contains both an imaging “beacon” and therapeutic moiety, is selectively expressed within the target lesion. The use of DNA (aptamers) as a direct imaging agent has showed some promise but additional work is required (203). There is still a great need for the development and validation of targeted imaging agents and ultimately their use as treatment endpoints for therapy assessment.

Allowing the use of microdosing to assess the pharmacokinetics and pharmacodynamics of novel agents (204) and the National Cancer Institute recommending the use of “phase 0” trials for the new drug/imaging agents (205) are facilitating the rapid translation of the most promising agents. As evidenced by the proliferation of preclinical and in some cases clinical, imaging programs within the pharmaceutical industry we can anticipate further streamlining of imaging agents to the clinic. Soon we will have data to show how much money, if any, molecular imaging can save in the drug/therapeutic development process. In cases where direct imaging is not feasible, alternative strategies to image the effector pathways, as gained from preclinical experience, will be helpful in assessing the therapeutic effect of the new agents while they are still in phase 0 trials.

For molecular imaging to realize its full potential in facilitating personalized patient care, the localization of specific imaging probes will need to be correlated carefully with established prognostic markers and linked to gene and tissue arrays derived from patient specimens. Perhaps nowhere more in cancer therapy than in radiation treatment planning can molecular imaging provide guidance. Although of increasing sophistication, the anatomic imaging methods currently used will soon give way to sensitive molecular imaging techniques for providing the utmost in specific, and in many cases targeted, radiation therapy.