How molecular imaging is speeding up antiangiogenic drug development

Cancer drug discovery is a relatively long process. Rational predefined steps have been streamlined for drug development in recent years, thanks to the development of many new technologies (Fig. 1). Many imaging techniques have been routinely used in the drug discovery process to directly monitor the drug in blood, normal, and tumor tissues and to evaluate the effects of the drug in the context of tumor (1, 2). Anatomic/functional imaging modalities, such as computed tomography, magnetic resonance imaging, and ultrasound, have been used to assess tumor size and structure. They can also provide valuable information on tumor perfusion, integrity of the blood-brain barrier, vessel density, vessel permeability, blood oxygenation, blood volume, blood flow, blood velocity, and flow resistance. However, with the recent shift in drug discovery from conventional cytotoxic drugs to novel agents against specific molecular targets, these conventional imaging modalities are usually no longer adequate. Cytostatic therapies are much less toxic, and disease stabilization may not lead to shrinkage of tumors in a short period of time. Molecular imaging recently emerges with increasing popularity as it can be used to monitor the changes at the molecular level in vivo, and it can help in evaluating treatment efficacy much earlier.

Molecular imaging refers to the characterization and measurement of biological processes at the molecular level (3). For a representative comprehensive review, the readers are referred to ref. 3. Molecular imaging techniques include positron emission tomography (PET), single-photon emission computed tomography (SPECT), molecular magnetic resonance imaging, magnetic resonance spectroscopy, optical bioluminescence, optical fluorescence, and targeted ultrasound (3). Molecular imaging can give whole body readout in an intact system, help to decrease the workload and speed up the drug development process, provide more statistically relevant results because longitudinal studies can be done in the same animals, aid in lesion detection and patient stratification, and help in individualized anticancer treatment monitoring and dose optimization.

Several excellent review articles have been published on the general role of molecular imaging on drug development (2, 4–6). For the remaining of this review article, we will mainly focus on expounding the role of molecular imaging in antiangiogenic drug development and briefly summarize the recent advances in tumor angiogenesis imaging.

Angiogenesis, the formation of new blood vessels from preexisting blood vessels, is a fundamental process occurring during tumor progression (7). Tumor growth depends on the balance between proangiogenic and antiangiogenic molecules. Molecules regulating angiogenesis include, but are not limited to, growth factor receptors, tyrosine kinase receptors, G-protein–coupled receptors for angiogenesis modulating proteins, integrins, and matrix metalloproteinases (MMP; ref. 8). Here, we will focus on three of the most intensively studied angiogenesis-related molecular targets: vascular endothelial growth factor (VEGF) and VEGF receptors (VEGFR), integrin α_vβ₃, and MMPs (Fig. 2). During tumor angiogenesis, these molecules interact closely with each other. Integrin α_vβ₃ can recruit and activate MMP-2, which degrades components of the basement membrane and interstitial matrix to facilitate tumor progression (9). It can also regulate the production of VEGF in certain tumor cells (10). Several other molecular targets that are also involved in tumor angiogenesis are thus far understudied and will only be briefly mentioned, such as endoglin (CD105) and E-selectin (Fig. 2).

VEGF is a potent mitogen in embryonic and somatic angiogenesis. It plays a central role in both normal vascular tissue development and tumor neovascularization (8). The VEGF family is comprised of seven members with a common VEGF homology domain: VEGF-A, VEGF-B, VEGF-C, VEGF-D, VEGF-E, VEGF-F, and placental growth factor. VEGF-A is a dimeric, disulfide-bound glycoprotein that exists in at least seven homodimeric isoforms, consisting of 121, 145, 148, 165, 183, 189, or 206 amino acids (11). These isoforms differ not only in their molecular weight but also in biological properties, such as the ability to bind to cell surface heparin sulfate proteoglycans.

The angiogenic actions of VEGF are mainly mediated via two closely related endothelium-specific receptor tyrosine kinases: Flt-1 (VEGFR-1) and Flk-1/KDR (VEGFR-2; ref. 12). Both are largely restricted to vascular endothelial cells and are overexpressed on the endothelium of tumor vasculature, whereas they are almost undetectable in the vascular endothelium of adjacent normal tissues. All of the VEGF-A isoforms bind to both VEGFR-1 and VEGFR-2. It is generally agreed that VEGFR-1 is critical for physiologic and developmental angiogenesis, and the function of VEGFR-1 differs with the stages of development, the states of physiologic and pathologic conditions, and the cell types in which it is expressed (12). VEGFR-2 is the major mediator of the mitogenic, angiogenic, and permeability-enhancing effects of VEGF. Overexpression of VEGFR or VEGF-A has been implicated as poor prognostic markers in various clinical studies of cancer (13). Agents that prevent VEGF-A binding to its receptors, antibodies that directly block VEGFR-2, and small molecules that inhibit the kinase activity of VEGFR-2 thereby blocking VEGF/VEGFR signaling are all under active development (14–16). The critical role of VEGF-A in cancer progression has been highlighted by the recent approval of the humanized anti-VEGF monoclonal antibody bevacizumab (Avastin; Genentech, South San Francisco, CA) for first-line treatment (17). Successful development of VEGF- or VEGFR-targeted molecular imaging could serve as a paradigm for the assessment of cancer therapeutics targeting tumor angiogenesis.

Recombinant human VEGF₁₂₁ has been labeled with ¹¹¹In for identification of ischemic tissue in a rabbit model, where unilateral hind limb ischemia was created by femoral artery excision (18). However, virtually no difference was observed between the ischemic hind limb and the contralateral hind limb. VEGF₁₂₁ has also been labeled with ^99mTc through an “Adapter/Docking” strategy (19). The tracer was used to image 4T1 murine mammary carcinoma, and very low tumor signal (<3 %ID/g) was observed. Recently, this tracer was also tested for the imaging of tumor vasculature before and after different types of chemotherapy (20). [¹²³I]VEGF₁₆₅ has also been reported as a potential tumor marker (21). Despite the high receptor affinity of this tracer, biodistribution in A2508 melanoma tumor-bearing mice indicated poor tumor-to-background ratio, most likely due to the low metabolic stability of the compound. Nonetheless, biodistribution, safety, and absorbed dose of [¹²³I]VEGF₁₆₅ was studied in patients with pancreatic carcinoma (22). Following i.v. administration, sequential images were recorded during the initial 30 minutes after injection. Although a majority of primary pancreatic tumors and their metastases were visualized by [¹²³I]VEGF₁₆₅ scan, the organ with the highest absorbed doses was the thyroid, indicating severe deiodination of the probe. A recombinant protein composed of VEGF₁₆₅ fused through a flexible polypeptide linker (GGGGS)₃ to the n-lobe of human transferrin was also reported for imaging angiogenesis, and the tumor contrast was modest (23). In all the above reports, radiolabeled VEGF isoforms were used for SPECT imaging. PET has several advantages over SPECT, including 1 to 2 orders of magnitude greater sensitivity, and the increasing implementation of clinical PET and PET/CT (computed tomography) scanners can facilitate the translation of novel PET tracers to the clinic.

A few radiolabeled anti-VEGF antibodies have been reported for PET imaging applications. VG76e, an IgG1 monoclonal antibody that binds to human VEGF, was labeled with ¹²⁴I for PET imaging of solid tumor xenografts in immunodeficient mice (24). Whole-animal PET imaging studies revealed a high tumor-to-background contrast (Fig. 3A). Although VEGF specificity in vivo was shown in this report, the poor immunoreactivity (<35%) of the radiolabeled antibody limits the potential use of this tracer. HuMV833, a humanized version of a mouse monoclonal anti-VEGF antibody MV833, was also labeled with ¹²⁴I, and the distribution and biological effects of HuMV833 in phase I trial cancer patients were investigated (25). Patients with progressive solid tumors were treated with various doses of HuMV833, and PET imaging using [¹²⁴I]HuMV833 was carried out to measure the antibody distribution. It was found that antibody distribution and clearance were quite heterogeneous not only between patients but also between individual tumors of the same patient, suggesting that intra-patient dose escalation approaches or more precisely defined patient cohorts would be preferred in the design of phase I studies with antiangiogenic antibodies such as HuMV833.

We have recently labeled VEGF₁₂₁ with ⁶⁴Cu (t_1/2 = 12.7 hours) for PET imaging of tumor angiogenesis and VEGFR expression (26). DOTA-VEGF₁₂₁ (where DOTA denotes 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid) exhibits nanomolar receptor binding affinity in vitro. MicroPET imaging revealed rapid, specific, and prominent uptake of [⁶⁴Cu]DOTA-VEGF₁₂₁ in highly vascularized small U87MG tumor (high VEGFR-2 expression) but significantly lower and sporadic uptake in large U87MG tumor (low VEGFR-2 expression; Fig. 3B). Western blot of tumor tissue lysate, immunofluorescence staining, and blocking studies with unlabeled VEGF₁₂₁ confirmed that the in vivo tumor uptake is VEGFR specific. Substantial tracer uptake in the kidneys was also observed, most likely due to the high VEGFR-1 expression in this organ. Successful demonstration of the ability of [⁶⁴Cu]DOTA-VEGF₁₂₁ to visualize VEGFR expression in vivo should allow for clinical translation of this tracer to image tumor angiogenesis and to guide antiangiogenic treatment, especially VEGFR-targeted cancer therapy. Based on the in vivo pharmacokinetics of [⁶⁴Cu]DOTA-VEGF₁₂₁, VEGF₁₂₁ may also be labeled with ¹⁸F for PET imaging applications as good tumor-to-background ratio was achieved as early as 1 to 2 hours after injection.

The abovementioned examples showed that molecular imaging of tumor angiogenesis can play a role in target validation, lead optimization, monitoring therapeutic response, and in clinical trials during the drug development process. Despite the critical role of VEGF and VEGFR in tumor angiogenesis, molecular imaging of VEGF or VEGFR has not been well studied. In the clinical setting, the right timing can be critical for VEGFR-targeted cancer therapy, and PET imaging of VEGF/VEGFR can play a very important role in determining whether to start and when to start the VEGFR-targeted treatment. Clinical translation will be critical for the maximum benefit of VEGF-based cancer imaging agents. Much research remains to be done in the near future to optimize VEGF- or VEGFR-targeted molecular imaging. Site-specific labeling via a Cys-tag may offer advantage over direct labeling on the amino group of the lysine side chain for new tracer development or tracer optimization. A VEGFR-2-specific ligand can also be developed, which may be superior to VEGF-A-based tracer, as experimental evidence has shown that VEGFR-2 is more involved in tumor angiogenesis, whereas VEGF-A binds to both VEGFR-1 and VEGFR-2. Peptidic VEGFR antagonists, which can be labeled with ¹¹C or ¹⁸F (more readily available than ⁶⁴Cu or ¹²⁴I), may also be tested (27). Peptide-based tracers may allow for higher throughput than antibody- or protein-based radiotracers, as 1-hour after injection is usually sufficient for a peptide-based tracer to clear from the non-targeted organs and give high-contrast PET images. In contrast, it may take several hours and even days before high-contrast PET images can be obtained for protein- or antibody-based tracer. Transgenic mouse models where either the VEGF or VEGFR-2 promoter drives reporter gene expression will also likely play a useful role in understanding VEGF biology and drug optimization (28).

As VEGF/VEGFR signaling is one of the most important pathways during tumor angiogenesis, the ability to image and quantify VEGF and/or VEGFR expression level during tumor growth and upon antiangiogenic treatment will be of critical importance. It has been shown that the therapeutic window of VEGF/VEGFR–targeted delivery does not depend on the total dose given but rather on the microenvironmental levels of VEGF/VEGFR expression. Visualizing and quantifying VEGF/VEGFR expression in vivo will allow for personalized treatment by choosing the right timing during which the treatment is most effective. In vivo imaging of VEGF/VEGFR expression will also be able to determine the effective dose of VEGF/VEGFR–based treatment, so that therapeutically efficacious dose levels can be given safely.

Integrins are a family of cell adhesion molecules consisting of two noncovalently bound transmembrane subunits (α and β), both type I membrane proteins with large extracellular segments that pair to create heterodimers with distinct adhesive capabilities (29). In mammals, 18 α and 8 β subunits assemble into 24 different receptors. Integrin signaling plays a key role in tumor angiogenesis and metastasis (30). Integrins expressed on endothelial cells modulate cell migration and survival during angiogenesis, whereas integrins expressed on carcinoma cells potentiate metastasis by facilitating invasion and movement across blood vessels. The α_vβ₃ integrin, which binds to arginine-glycine-aspartic acid (RGD)–containing components of the interstitial matrix, is significantly up-regulated on endothelium during angiogenesis but not on quiescent endothelium (30, 31). Inhibition of α_vβ₃ integrin activity by monoclonal antibodies, cyclic RGD peptide antagonists, and peptidomimetics has been shown to induce endothelial cell apoptosis, to inhibit angiogenesis, and to increase endothelial monolayer permeability (32).

Crystal structure of the extracellular portion of integrin α_vβ₃ in complex with c(RGDf(NMe)V) has been reported (31, 33). The cyclic RGD peptide binds at the major interface between the α_v and β₃ subunits and makes extensive contacts with both in a transition metal (e.g., Mn²⁺)–dependent mode. Preclinical studies indicated that many integrins other than α_vβ₃ also play important roles in regulating angiogenesis, such as α₁β₁, α₂β₁, α₄β₁, α₅β₁, α₆β₄, α_IIbβ₃, and α_vβ₅ (32). Among all 24 integrins discovered to date, integrin α_vβ₃ is the most intensively studied, and an extensive review on multimodality molecular imaging of integrin α_vβ₃ has recently been published (34).

Antibody-coated paramagnetic liposomes (35), Gd-perfluorocarbon nanoparticles conjugated to anti-integrin α_vβ₃ monoclonal antibody (36), and integrin α_vβ₃-targeted paramagnetic nanoparticles (37) have been reported for magnetic resonance imaging of integrin α_vβ₃ expression (Fig. 4A). In these studies, the targeted paramagnetic nanoparticles are coated with either antibodies or small peptidic/peptidomimetic integrin α_vβ₃ antagonists. Because of the relatively large size of the probes (200–700 nm), these agents target the integrin α_vβ₃ expressed on the tumor vasculature rather than the tumor cells. Ultrasound imaging using integrin α_vβ₃–targeted microbubbles has also been reported (Fig. 4B; ref. 38). Because acoustic destruction of “payload-bearing” microbubbles can be used to deliver drugs or to augment gene transfection (39), integrin α_vβ₃–targeted microbubbles may have applications in site-specific cancer therapy. Further studies are needed to validate the potential therapeutic applications.

Although optical imaging may not be widely used in clinical settings, near-IR (700–900 nm) approaches provide opportunities for rapid and cost-effective preclinical evaluation in small animal models before the more costly radionuclide-based imaging studies. These approaches may also be translated into the clinic with fluorescence-mediated tomography (e.g., for breast cancer imaging). In the near-IR region, the absorbance of all biomolecules reaches minima, providing a clear window for in vivo optical imaging (40). We have shown that near-IR fluorescent dye or quantum dot–conjugated cyclic RGD peptide could be used to visualize s.c. inoculated integrin α_vβ₃–positive tumors (41, 42). The dye-RGD peptide conjugate is small in size; therefore, it targets integrin α_vβ₃ on both tumor cells and tumor vasculature (Fig. 4C). For the quantum dot-RGD peptide conjugate, it mainly targets integrin α_vβ₃ in the tumor vasculature because it does not extravasate well due to the relatively large size (≥20 nm; Fig. 4D). Based on these results, an approach that takes advantage of the high integrin α_vβ₃–targeting efficacy of the cyclic RGD peptides, high stability and brightness of quantum dots, and emission wavelength in the near-IR window will have great potential in cancer diagnosis and imaging as well as imaging-guided surgery and therapy.

Recently, Achilefu et al. (43) discovered that conjugating a presumably inactive linear hexapeptide GRDSPK with an near-IR carbocyanine molecular probe yielded Cyp-GRD that targets integrin α_vβ₃–positive tumors. More experiments need to be carried out to fully understand this surprising phenomenon, and docking study may reveal whether Cyp-RGD actually binds to the RGD binding domain in integrin α_vβ₃. Later, they synthesized and evaluated a series of multimeric RGD compounds constructed on a dicarboxylic acid–containing near-IR fluorescent dye cypate for tumor targeting (44). Optimization of the spatial alignment of the RGD moieties through careful molecular design and library construction may induce multivalent ligand-receptor interactions useful for in vivo tumor imaging and tumor-targeted therapy.

RGD peptides have been labeled with ¹¹¹In and ^99mTc for SPECT imaging of integrin α_vβ₃ expression. The in vivo behavior of radiolabeled dimeric RGD peptide E[c(RGDfK)]₂ was studied in an ovarian carcinoma xenograft model (45). ¹¹¹In/⁹⁰Y and ^99mTc were incorporated through DOTA and hydrazinonicotinamide (HYNIC) chelators, respectively (Fig. 4E). RP748, an ¹¹¹In-labeled quinolone that binds to integrin α_vβ₃ with high affinity, was recently studied both in vitro and in vivo to track injury-induced vascular proliferation in rodents (46). Water-soluble N-(2-hydroxypropyl)methacrylamide copolymers have been synthesized with pending doubly cyclized RGD peptides (47). The bioactivity of the polymer conjugates and free peptides was characterized both in vitro and in vivo. It was shown that specific targeting of α_vβ₃ integrin and nonspecific vascular permeability both contributed significantly to the tumor uptake, with specific targeting being more important. It was concluded that peptide oligomers may be more suitable for imaging purposes because of the rapid clearance, whereas peptide-polymer conjugates may be used for high-level targeting and radiotherapeutic approaches.

PET has been the mainstay of integrin α_vβ₃ expression imaging, and most reports focus on the radiolabeling of RGD peptide antagonists. Monomeric RGD peptide c(RGDyV) was first labeled with ¹²⁵I by Haubner et al. (48). A glycopeptide based on c(RGDfK) was later labeled with ¹⁸F via a 2-[¹⁸F]fluoropropionate prosthetic group, and the resulting [¹⁸F]galacto-RGD exhibited integrin α_vβ₃–specific tumor uptake in integrin-positive M21 melanoma xenograft model (49). Initial clinical trials in healthy volunteers and a limited number of cancer patients revealed that this tracer can be safely given to patients and is able to delineate certain lesions that are integrin positive (Fig. 4F; ref. 50). Despite the successful translation of [¹⁸F]galacto-RGD into clinical trials, several key issues remain to be resolved, such as tumor-targeting efficacy, pharmacokinetics, and the ability to quantify integrin α_vβ₃ density in vivo.

We have labeled a series of RGD peptides with ¹⁸F for PET imaging, using PEGylation and polyvalency to improve the tumor-targeting efficacy and pharmacokinetics (51–56). [¹⁸F]FB-E[c(RGDyK)]₂ (abbreviated as [¹⁸F]FRGD2) had predominant renal excretion and almost twice as much tumor uptake in the same animal model compared with the monomeric tracer [¹⁸F]FB-c(RGDyK) (Fig. 4G; refs. 54, 55). The synergistic effect of polyvalency and improved pharmacokinetics may be responsible for the excellent imaging characteristics of [¹⁸F]FRGD2. Graphical analyses of the dynamic microPET scans in six tumor xenograft models were carried out to correlate the tumor uptake with integrin α_vβ₃ expression level measured by SDS-PAGE autoradiography, and excellent linear correlation was observed. More importantly, it was found that at late time points when most of the nonspecific binding had been cleared, the tumor/background ratio had a linear relationship with tumor integrin α_vβ₃ expression level, thus making it possible to quantify integrin α_vβ₃ density in vivo. We are currently in the process of translating [¹⁸F]FRGD2 into the clinic for cancer patient imaging.

In addition to ¹⁸F, ⁶⁴Cu-labeled RGD peptides are also of considerable interest. Copper-64 is an attractive radionuclide for both PET imaging and targeted radiotherapy of cancer. PET imaging of tumors with low doses of ⁶⁴Cu-labeled RGD peptides may be used to determine radiation dosimetry before therapy with high dose of ⁶⁴Cu- or ⁶⁷Cu-labeled RGD peptides. We have labeled RGD peptides with ⁶⁴Cu for PET imaging, again using PEGylation and polyvalency to optimize the tumor-targeting efficacy and pharmacokinetics (53, 57–59). Recently, we reported a tetrameric RGD peptide-based tracer, [⁶⁴Cu]DOTA-E[E[c(RGDfK)]₂]₂, which showed significantly higher receptor binding affinity than the corresponding monomeric and dimeric RGD analogues (60). This tracer exhibited rapid blood clearance, high metabolic stability, predominant renal excretion, significant receptor-mediated tumor uptake, and good contrast in xenograft-bearing mice (Fig. 4H). The high integrin avidity and favorable biokinetics makes [⁶⁴Cu]DOTA-E[E[c(RGDfK)]₂]₂ a promising agent for peptide receptor radionuclide imaging as well as targeted internal radiotherapy of integrin α_vβ₃–positive tumors.

Abegrin (MEDI-522, also called Vitaxin; MedImmune, Inc., Gaithersburg, MD), a humanized monoclonal antibody against human integrin α_vβ₃ (picomolar binding affinity), is in clinical trials for cancer therapy. We have conjugated Abegrin with macrocyclic chelating agent DOTA and labeled it with ⁶⁴Cu for PET imaging of tumor xenografts (61). MicroPET studies revealed that [⁶⁴Cu]DOTA-Abegrin had very high tumor uptake in integrin α_vβ₃–positive U87MG tumor (Fig. 4I). The receptor specificity of [⁶⁴Cu]DOTA-Abegrin was confirmed by effective blocking of tumor uptake with coadministration of nonradioactive Abegrin. The success of integrin α_vβ₃–specific tumor imaging using [⁶⁴Cu]DOTA-Abegrin may be translated into the clinic to characterize the pharmacokinetics, tumor-targeting efficacy, dose optimization, and dose interval of Abegrin and/or Abegrin conjugates. Chemotherapeutics or radiotherapeutics using Abegrin as the delivering vehicle may also be effective in treating integrin α_vβ₃–positive tumors.

Integrin α_vβ₃ is one of the most extensively studied molecular targets involved in tumor angiogenesis (30–34). The numerous reports on multimodality molecular imaging of integrin α_vβ₃ again showed that tumor angiogenesis imaging can participate in multiple stages of the drug development process, such as target validation, lead optimization, and clinical trials. However, to date, only [¹⁸F]galacto-RGD has advanced into clinical settings for further testing, and the tracer itself is probably suboptimal (62). Polyvalent integrin α_vβ₃ antagonists, such as multimeric RGD peptides, are also promising ligands for the molecular targeting of integrins involved in tumor angiogenesis. Translation of new multimeric RGD peptide-based tracers into the clinic will dramatically benefit antiangiogenic cancer therapy based on integrin α_vβ₃ antagonism. The ability to quantify integrin α_vβ₃ expression level in vivo will be very important in monitoring antiangiogenic treatment efficacy. Some of the abovementioned tracers with high integrin α_vβ₃-positive tumor uptake, as shown by molecular imaging studies, may also have the potential to evolve into radioimmunotherapeutic agents for cancer therapy.

MMPs are a family of Zn²⁺-dependent endopeptidases, which play important roles in tumor angiogenesis, in particular the 72-kDa (MMP-2) and 92-kDa (MMP-9) gelatinases (63). MMP-2, capable of degrading type IV collagen (major component of the basement membranes), can be localized in a proteolytically active form on the surface of invasive cancer cells based on its ability to bind to integrin α_vβ₃ (9). A number of MMP inhibitors (MMPI) have been developed as cytostatic and antiangiogenic agents and are currently in clinical testing (64). Until recently, clinical trials with MMPIs have yielded disappointing results, highlighting the need for better insight into the mechanisms by which MMPs contribute to tumor growth. Molecular imaging to monitor MMP expression noninvasively in vivo will be critical for future drug development targeting MMPs.

Molecular imaging of MMP expression, as well as imaging of other enzymes, can be divided into two approaches. In the “targeted” probes approach, labeled small molecules, peptides, metabolites, aptamers, antibodies, or other molecules are injected i.v., and the living systems can be imaged when a fraction of the agent has bound to its target and the non-bound agent has been cleared. Non-peptidyl broad-spectrum MMPIs, tissue inhibitors of metalloproteinases, MMP-2-specific inhibitors, carboxylic and hydroxamic acid–based MMPIs, selective inhibitors of MMP-2/MMP-9, and other MMPIs have been labeled with ¹²³I, ¹²⁵I, ¹¹¹In, ^99mTc, ¹¹C, ¹⁸F, and/or ⁶⁴Cu for in vivo biological and clinical investigation of MMP expression using SPECT and PET (65–72). However, in most of the reports, selective binding of the labeled compounds to specific MMPs was not shown, and high nonspecific binding was observed possibly due to low in vivo stability of the tracers. Except for one most recent report (Fig. 5A; ref. 71), none of the abovementioned reports showed any convincing in vivo imaging results. The cyclic decapeptide CTTHWGFTLC, a selective MMP-2 and MMP-9 inhibitor, was conjugated with DOTA and labeled with ⁶⁴Cu for PET imaging of MMP expression in xenograft models. Zymography of tumor extracts supported the in vivo PET imaging results. MMP-2 and MMP-9 bands were clearly detectable in the mouse imaged at 7 weeks, which had prominent tumor uptake, whereas the MMP-2 and MMP-9 expression was very week afterwards, which gave low tumor uptake (Fig. 5A). However, MMP expression in the MDA-MB-435 tumor model used in this study had quite large individual variance. The low in vivo stability of the radiotracer also limits further application of this strategy.

The abovementioned approach based on affinity ligands is usually more useful for imaging receptors and cell surface–expressed molecules, but it may not be the best approach for imaging enzyme function in a living organism. Another approach is the use of “activatable probes.” Activatable probes undergo chemical or physicochemical changes on target interaction and result in signal amplification. This approach was first shown by Bremer et al. where MMP activity was imaged in live animals, and that the inhibition of MMP activity can be recorded within hours after treatment by a potent MMPI (Fig. 5B; refs. 73, 74). Later, similar approaches were used for imaging MMP expression in the heart after myocardial infarction where an near-IR fluorescent probe was activated upon proteolytic cleavage by MMP-2 and MMP-9 (75). In another report, cellular association of polyarginine-based cell-penetrating peptides is effectively blocked when they are fused to an inhibitory domain composed of negatively charged residues, which was called “activatable cell-penetrating peptides” (76). Cleavage of the MMP-sensitive linker between the polycationic and polyanionic domains releases the cell-penetrating peptide portion and its attached cargo to bind to and enter cells. In xenograft tumor models expressing MMP-2/MMP-9, a fluorescent dye Cy5-conjugated activatable cell-penetrating peptide showed modest in vivo tumor contrast (2- to 3-fold when compared with contralateral normal tissue). The similar approach has also been applied to modulate the cellular uptake of quantum dot conjugates (77). Another fluorogenic activatable probe, which is MMP-7 selective, has been reported for in vivo detection and imaging of tumor-associated MMP-7 activity (78). Although the activatable probe strategy may give good tumor/background contrast, the major drawback of optical imaging is that it has limited clinical applications.

In vivo imaging of MMP expression is still underdeveloped due to many issues. First, the expression level of MMPs varies during different stages of tumor progression, and imaging may also be complicated by naturally occurring tissue inhibitors of metalloproteinases. Second, specificity for one particular MMP is hard to achieve. There is still plenty of room for improving the selectivity between different MMPs, especially MMP-2 and MMP-9. The experimental outcome would be much better interpreted if the ligand is specific for one particular MMP. Third, quantifying the MMP expression level in vivo has not been demonstrated. The imaging results reported thus far are mostly qualitative. The ability of correlating the MMP imaging result with the MMP expression in vivo will dramatically help the advancement of MMP imaging, anti-MMP drug screening, and monitoring of treatment efficacy.

In vivo imaging of endothelial markers in intact tumor neovasculature can significantly help assessing the efficacy of antiangiogenic agents in clinical trials. Although many endothelial markers have been described, only few of them have been evaluated as imaging markers. Endoglin (CD105) is emerging as a prime vascular target for antiangiogenic cancer therapy (79). It is a cell membrane glycoprotein mainly expressed on endothelial cells and overexpressed on tumor vasculature. It functions as an accessory component of the transforming growth factor-β receptor complex and is involved in vascular development and remodeling. Avidin-coated microbubbles have been linked to biotinylated monoclonal antibodies for endoglin targeting in vitro (80). ¹²⁵I-labeled monoclonal antibody MAEND3 has been reported to target CD105 on tumor vasculature in canine models (81). Another radiolabeled monoclonal anti-endoglin antibody has also been used for in vivo imaging (82). The major advantage of imaging abundantly expressed endothelial targets is that this strategy circumvents delivery barriers normally associated with other tumor-targeting strategies.

E-selectin is a cell adhesion molecule and CD antigen that mediates neutrophil, monocyte, and memory T-cell adhesion to cytokine-activated endothelial cells (83). It is expressed exclusively by activated endothelial cells, and it recognizes sialylated carbohydrate groups related to the Lewis X or Lewis A family. Fluorescence reflectance imaging (a photographic process that captures views of a surface under varying lighting conditions to enhance surface detail that may otherwise be difficult to see) of E-selectin expression in mouse xenograft models of Lewis lung carcinoma has been reported (84). The imaging probe was constructed by conjugating an E-selectin-binding peptide (CDSDSDITWDQLWDLMK) to CLIO(Cy5.5) nanoparticles, where CLIO represents cross-linked iron oxide that can be used for magnetic resonance imaging.

Endostatin, which binds fibulin and nidogen, is a 20-kDa COOH-terminal fragment of collagen XVIII, and it is a potent naturally occurring inhibitor of angiogenesis (7). It has been labeled with a fluorescent dye Cy5.5 for tumor localization after i.p. injection (85). Endostatin-Cy5.5 was quickly absorbed after administration, producing a near-IR fluorescence image of the tumors that persisted through 7 days. [^99mTc]ethylenedicysteine-endostatin has also been synthesized for evaluating the efficacy of antiangiogenic therapy (86). Tissue distribution and planar imaging of radiolabeled endostatin were determined in tumor-bearing rats. It was claimed that [^99mTc]ethylenedicysteine-endostatin could assess treatment response, and there was a correlation between tumor uptake and cellular targets expression level.

Significant advances have been made in developing novel probes for multimodality molecular imaging of tumor angiogenesis. Small molecules, peptides, peptidomimetics, proteins, and antibodies have been labeled with radioisotopes, superparamagnetic nanoparticles, fluorescent dyes, quantum dots, and microbubbles for PET, SPECT, magnetic resonance imaging, near-IR fluorescence, and ultrasound imaging of small animal tumor models, a few of which are now in early clinical trials. The major roles of tumor angiogenesis imaging in the drug development process will be the following: target identification, characterization, and validation; patient stratification (e.g., selecting the right population of cancer patients for new clinical trials); pharmacokinetic/pharmacodynamic studies (e.g., candidate drug screening and optimization, phase I clinical trials); as well as treatment monitoring and dose optimization (phase II/III trials). Through the development of a robust tumor angiogenesis imaging platform, molecular imaging can dramatically facilitate and speed up many steps of antiangiogenic drug development in both the preclinical and clinical stages.

Despite the strong potentials of angiogenesis imaging probes, most of the research efforts have thus far been limited to probe optimization for enhanced tumor-targeting efficacy and improved in vivo kinetics. The translation of the imaging probes from bench to bedside has been slow. The limited margins for marketing the very special probes makes some of de novo imaging approaches considered too risky by investigators. The situation is, however, being improved. Food and Drug Administration recently developed exploratory Investigational New Drug mechanism to allow faster first-in-human studies. Microdosing studies with novel imaging probes can provide an opportunity for early assessment of the safety profile and pharmacokinetics in healthy volunteers. Such rapid initial clinical studies will definitely accelerate the drug discovery process. Furthermore, the molecular imaging field has grown extremely fast over the last decade, and the value of molecular imaging in drug development and screening is more widely accepted by pharmaceutical companies. By repeated imaging in preclinical models using one or more of the discussed imaging strategies, one can have several readouts of angiogenesis before and after drug administration. Even if the drug target is different from the imaging target, one can still use imaging as a potential surrogate for the efficacy of the drug at a given dose. It is expected that in the foreseeable future molecular imaging will be routinely applied in many steps of the drug development process. The combination of molecular and anatomic/functional imaging techniques in assessing tumor angiogenesis and in the response to antiangiogenic cancer therapy will be a powerful tool.

To foster the continued discovery and development of angiogenesis-targeted imaging probes, cooperative efforts are needed from cellular/molecular biologists to identify and validate molecular imaging targets, chemists/radiochemists to synthesize and characterize the imaging probes, and medical physicists/mathematicians to develop high-sensitivity/high-resolution imaging devices/hybrid instruments and to develop better algorithms to further improve signal-to-noise ratio of a given imaging device. Close partnerships among academic researchers, clinicians, pharmaceutical industries, the National Cancer Institute, and the Food and Drug Administration are also needed to promote further development of imaging probes, to apply molecular/functional imaging approaches to predict and evaluate antiangiogenic effect during and after treatment, to move molecular imaging guided intervention strategy quickly into the clinic, and to accelerate antiangiogenic drug development.