Molecular Images of Neovascularization
Art for Art’s Sake or Form With a Function?
Therapeutic modulation of vascular growth with growth factors or genes encoding for them is an attractive strategy for treating individuals afflicted with obstructive atherosclerotic coronary artery and peripheral vascular disease. Despite the initial promise of “therapeutic angiogenesis” in experimental animal models and small open-label clinical trials, recent larger blinded placebo-controlled trials in such patients have failed to demonstrate a clear-cut treatment benefit.1–3 The reasons for this are complex, incompletely understood, and the subject of controversy, the scope of which is beyond this editorial and is discussed in other excellent reviews.1,2 It is worth, however, considering some questions raised by the discordance between preclinical and clinical results because the articles on molecular imaging of neovascularization by Hua et al4 and Leong-Poi et al5 in the present issue of Circulation could ultimately shed useful light on this debate.
A question that has been raised in deliberations over why angiogenesis trials in humans have not demonstrated the robust therapeutic effects seen in earlier animal studies is whether the end points chosen for these trials were “correct”; ie, what is the definition of “successful” therapeutic neovascularization? Furthermore, were the measurement tools that were used to capture such end points adequately sensitive to detect them, even if they were present? Most scientists and clinicians would agree that successful neovascularization requires that new blood vessels be functional, supply the ischemic region with blood, and be stable over time. Our patients would emphasize that irrespective of whether the treatment makes them live longer, they would like to feel better, and hence quality of life is an important end point. Here, we consider whether the identification of molecular markers of neovascularization adds useful information and/or should be incorporated into the definition of “successful” neovascularization.
An intuitively appealing approach to the question of whether new vessels have formed is to conduct a functional perfusion assessment before and after treatment. The most clinically accessible method when it comes to the heart is to perform single-photon emission computed tomography (SPECT) stress imaging; however, this has not turned out to be a straightforward path. For one thing, SPECT cannot spatially resolve endocardial/epicardial differences in perfusion.6 Hence, it would be difficult to detect regionally limited angiogenic responses that, in addition to mitigating ischemia, may favorably affect other related physiological phenomena such as left ventricular remodeling or arrhythmia potential. In addition, because collaterals typically have impaired flow reserve,7 stress SPECT images could still show flow heterogeneity despite the anatomic presence of neovascularization. Alternatively, rest imaging after treatment may show improved perfusion of previously hypoperfused areas, but this approach would miss angiogenesis occurring in regions that were not initially ischemic at rest. Moreover, it has been shown that there is considerable variability in consecutive SPECT images even in the same patient,8 indicating that shortcomings of the imaging technique, and/or true temporal variability in perfusion that may exceed changes resulting from angiogenesis, would confound such assessments. Other less clinically routine approaches that may be useful for measuring the efficacy of neovascularization therapies, such as positron emission tomography and MRI, exercise testing, and quality-of-life measures, also have significant limitations.6
One suggested approach to wrestling with the problem of detecting neovascularization is to identify molecular biomarkers of the events that are known to be associated with angiogenesis or arteriogenesis.9 The premise behind this is that if the molecular marker is present, the process of neovascularization, or at least a component of it, must therefore be occurring. By extension, if the molecular marker can be visualized with an imaging technique, neovascularization can be inferred. In this issue of Circulation, 2 studies present such an approach, one via nuclear scintigraphy of angiogenesis and the other via ultrasound arteriogenesis in rodent models of hindlimb ischemia.4,5 Although the 2 imaging strategies have intrinsic differences as described below, they have in common the choice of the endothelial cell integrin αvβ3 as the molecular target to which an entity that can be imaged (ie, 99mTc or a microbubble) is designed to bind. αvβ3 has relatively limited distribution in most normal cells, including quiescent endothelial cells, but is upregulated in activated endothelial cells such as in the neovasculature within ischemic tissue and tumors.10
Using planar scintigraphic imaging and postmortem gamma well counting in a murine model of distal hindlimb ischemia, Hua and colleagues determined the 14-day time course and regional localization of an intravenously injected 99mTc-labeled peptide containing an RGD (Arg-Gly-Asp) sequence (NC100692) with putative binding specificity to αvβ3.4 Compared with nonischemic control hindlimb, the ratio of activity in ischemic to sham-operated limb was higher at 3 and 7 days. Capillary density peaked at 14 days, whereas NC100692 uptake peaked earlier and declined to baseline levels thereafter. Although statistically higher than normal controls, image count ratios were only ≈1.5 times greater in ischemic than in sham-operated limb. The ratios were somewhat better with direct gamma well counting, possibly because the tissue samples excluded overlying postsurgical scar tissue that likely contributed to image counts in the sham-operated limb. Hindlimb-to-blood ratios would have been useful in determining how much of the signal derived from background blood activity. A fluorescent analog of NC100692 confirmed endothelial localization, although specific binding to the αvβ3 integrin was not unequivocally demonstrated. Given this lack of more definitive demonstration of binding specificity by NC100692, it would have been helpful to use a radiolabeled control peptide to parcel out any contribution of nonspecific uptake of the isotope to the signal, including that related to vascular leakage at sites of ischemia or angiogenesis. Methodological issues notwithstanding, this study suggests that this radiolabeled peptide indeed homes to an endothelial epitope, for which the time course of expression tracks the appearance of new capillaries in ischemic muscle.
Leong-Poi and colleagues used a similar targeting principle but a different imaging technology to track the expression of αv- and α5β1.5 Unlike nuclear tracers, the microbubbles used in this study were intravascular gas-filled microspheres conjugated to a disintegrin containing an RGD motif conferring binding to α5β1 and αv- integrins, especially αvβ3. Targeted or control bubbles were intravenously injected into rats with up to 28 days of proximal hindlimb ischemia, half of which received local controlled-release fibroblast growth factor-2 (FGF-2). Several minutes after bubble injection, ultrasound was used to destroy the microbubbles in the imaging sector (presumably only adhered microbubbles remained at that point), generating nonlinear acoustic signals that were displayed as increased videointensity. Targeted bubble signals from all rats increased and then returned to baseline by 28 days, with the FGF-2–treated rats having an earlier and higher peak than the nontreated rats. Unlike the study of Hua et al, this study used a control microbubble against which to compare the targeted agent, but a sham surgery control was not used to help tease out signal deriving from postoperative wound healing versus ischemia-induced angiogenesis.
Interestingly, peak targeted bubble binding preceded blood flow restoration measured via perfusion contrast ultrasound. The concurrent perfusion imaging allowed adjustment of the targeted images for the magnitude of microbubble delivery. Also of note is that in both experimental groups, flow tended to decrease over time, and because rats were not observed past 28 days, we do not know whether neovascularization was sustained beyond this period.
So where do these studies lead us? Will these data help move the field of therapeutic neovascularization forward? Let us examine some issues raised by these reports. If we assume for a moment that molecular imaging of angiogenesis is a legitimate and helpful enterprise (which as discussed below, I believe it is), then these 2 studies highlight some issues that should be overcome to move this effort closer toward clinical application. Some of these challenges are unique to each imaging strategy and confer relative advantages of one over another, whereas others are common to all of molecular imaging. A shared challenge relates to the relatively low signal-to-noise ratio of the images. In general, this is partly the result of the choice of the target itself, which, in addition to being specific for the physiological process under study, should be expressed in sufficient density to bind detectable amounts of the targeted agent (although much is “enough” is unknown). Furthermore, in the case of microbubbles, which circulate exclusively within the intravascular space, the target must be endothelial and endoluminal in location. In addition, the target should be expressed at the level of the neovasculature that is accessible by the imaging modality (eg, capillaries versus arterioles versus larger arterial conduits). The micrographs in both studies suggest that expression of αvβ3 is not abundant at the time points of measurement and that there is considerable extravascular expression. A disadvantage of contrast ultrasound is that unlike nuclear tracers, microbubbles cannot access such extravascular targets. Conversely, however, nuclear agents are more susceptible to nonspecific extravasation in settings associated with increased vascular permeability such as angiogenesis.
In addition to issues of target specificity and availability, a strong image signal will result only if the imaging agent itself is bioavailable once administered. Many known and unknown factors affect this, including the molecular size of the agent, hydrophilicity, charge, stability, rate of elimination, and nonagent parameters, such as vascular permeability, flow, and in the case of extravascular targets, interstitial pressures.9 The imaging ligand should have high affinity and specificity for the target. Peptide sequences or the naturally occurring ligands for receptors are attractive targeting moieties.4,5,11,12 Particularly in the case of a particle such as a microbubble, the targeting ligand should be capable of sustaining binding in the face of shear forces imposed by flow. For a nuclear tracer, the dose activity achievable by the agent would also affect the signal-to-noise ratio. Data on the specific activity of the NC100692 (ie, what is the ratio of 99mTc to peptide?) were not provided, but improvements in this parameter would enhance the signal-to-noise ratio, particularly if the target is not abundantly overexpressed.
Blood flow is another variable that affects bioavailability of the imaging agent. Sufficient flow should exist to the ischemic area to allow delivery of the targeted agent, which is particularly important if there is constitutive expression of the target in nonischemic tissue. Normalization of the images to flow, as was done by Leong-Poi et al, can help to adjust for low flow, particularly when the agent is short lived, and again reminds us that even were we to enlist the aid of molecular imaging, it is difficult to escape the requirement to measure perfusion.
An important limitation to date of both nuclear and ultrasound molecular imaging is the technology for signal detection. For targeted ultrasound imaging, for example, systems with adequate spatial resolution and sensitivity for detecting adhered microbubbles are currently suboptimal, and the 2 studies discussed here have attempted to retrofit existing clinical instrumentation to the scale of rodent organs. The current detection strategy for targeted microbubbles employs bubble destruction to retrieve the strongest backscatter signal. This “destroy it to see it” approach precludes repetitive imaging. In clinical application, if ultrasound imaging commenced too early after contrast injection, then not only would the signal be contaminated by background noise from unbound bubbles persisting in the blood pool but also any adhered bubbles would be destroyed in the process and would be unavailable for subsequent imaging. Alternatively, ultrasound provides better spatial resolution than does planar nuclear imaging or SPECT and can likely resolve transmural gradients in target expression.13 Furthermore, the simultaneous 2-dimensional imaging permitted by ultrasound enables target localization without signal spillover from neighboring tissue, whereas nuclear scintigraphy only provides “hot spot” images with less precise anatomic localization of the target.
Comparative methodological issues aside, let us return to the broader question raised in my initial comments: Will integrin imaging advance the field of therapeutic angiogenesis? What is the clinical significance of identifying αvβ3 integrin, or for that matter, other epitopes associated with neovascularization? Does the presence of the molecular marker guarantee an improvement in perfusion? What is the place of molecular imaging in defining end points in therapeutic neovascularization trials?
It should be emphasized that neovascularization is a highly complex process that is spatially and temporally dynamic and incompletely understood.14 The harnessing of molecular markers to interrogate angiogenesis requires a complete understanding of the molecular events in the process, which currently does not exist. Our ability to interpret the molecular images we generate and to use them clinically is only as good as our understanding of the neovascularization process itself, and in particular, where the molecular target falls within that continuum. As both studies reviewed here demonstrate, αvβ3 expression is a time-dependent and transient phenomenon in the spectrum of events leading to neovascularization. Both studies indicate that its maximum expression predates the peaks in perfusion or histological neovascularization and that with passing days, αvβ3 expression decreases. Thus, at best, the imaging of αvβ3 provides an isolated snapshot in time.
It is increasingly recognized that no one event or growth factor leads to neovascularization and that effective therapeutic angiogenesis may require supplementation of multiple growth factors at different times.2 Just as no single event leads to a fully mature blood vessel, no one image in time can summarize the process of neovascularization. This applies not just to molecular nuclear scintigraphy or ultrasound but to other molecular imaging technologies under development9 and not discussed here.
How, then, might molecular imaging of an appropriately selected target be useful in the context of therapeutic neovascularization? In vivo molecular imaging of neovascularization can be a useful research tool in basic studies of the mechanisms of neovascularization. In the clinical arena, the ability to image molecular markers of neovascular milestones (to the extent that we know what they are) may increase the sophistication of therapeutic angiogenic strategies by permitting an assessment of the status of specific, discrete aspects of the neovascularization response. A mature technology of angiogenic molecular imaging could be used to identify defects in receptor and postreceptor intracellular signaling in response to growth factors, which may partly explain individual differences in endogenous neovascularization in patients with ischemia.2 For example, it has been shown that vascular responsiveness to growth factor stimulation decreases with age, atherosclerosis, hyperglycemia, and diabetes.15 Molecular imaging could be helpful in identifying individuals who are susceptible to growth factor stimulation (eg, do they upregulate αvβ3 in response to exogenous FGF?), and hence are most likely to respond to angiogenic interventions. Its application to the identification of biomarkers predictive of neovascularization could help to optimally select clinical trial populations rather than limiting enrollment, as has been done in previous trials, to those at the end of the road, who arguably are least likely to respond to growth factor therapy. To the extent that quality-of-life measures can be more difficult to accept by many as an end point, some evidence to indicate that the target tissue is “transducing” the treatment (eg, by expressing αvβ3) could be helpful in proving that treatment is having some sort of biological effect, even when improvements in gross “soft” measures such as exercise tolerance or symptoms are not immediately obvious, and long-term outcome measures such as mortality are still months or years away.
Because of the physiological complexity of neovascularization and its spatial and temporal dynamism, it is unlikely that a single imaging strategy for proving any benefit of therapeutic angiogenesis will “do the trick.” Molecular imaging, however, can uniquely answer a number of important questions as described above. Importantly, it can be incremental by reframing the question as to what constitutes “successful” neovascularization along the submicroscopic dimension, by offering an in vivo glimpse of molecular surrogates of neovascularization, where standard clinical measurements may fail. As discussed, however, much work needs to be done to bring this promising technology beyond colorful images to the level of useful purpose. It is likely that such efforts will come to fruition in step with expanded insights into the basic process of neovascularization itself, and in turn molecular imaging of neovascularization will help increase our grasp of this intricate multigene process. Ultimately, a multidimensional approach involving early molecular visualization, perfusion assessment, and tissue hypoxia measurement over the course of time may give us the best chance of truly discovering whether therapeutic neovascularization is helpful to patients with atherosclerotic cardiovascular disease.
The author thanks Gregory Weller, PhD, for his review of this editorial and his predoctoral work in her laboratory in this area. Dr Villanueva is supported in part by grants from the National Institutes of Health, Bethesda, MD (HL-RO158865 and HL-RO1077534). She is an Established Investigator of the American Heart Association.
The opinions expressed in this article are not necessarily those of the editors or of the American Heart Association.
Annex BH, Simons M. Growth factor–induced therapeutic angiogenesis in the heart: protein therapy. Cardiovasc Res. 2005; 65: 649–655.
Simons M. Angiogenesis: where do we stand now? Circulation. 2005; 111: 1556–1566.
Hua J, Dobrucki LW, Sadeghi MM, Zhang J, Bourke BN, Cavaliere P, Song J, Chow C, Jahanshad N, van Royen N, Buschmann I, Madri JA, Mendizabal M, Sinusas AJ. Noninvasive imaging of angiogenesis with a 99mTc-labeled peptide targeted at avβ3 integrin after murine hindlimb ischemia. Circulation. 2005; 111: 3255–3260.
Leong-Poi H, Christiansen J, Heppner P, Lewis CW, Klibanov AL, Kaul S, Lindner JR. Assessment of endogenous and therapeutic arteriogenesis by contrast ultrasound molecular imaging of integrin expression. Circulation. 2005; 111: 3248–3254.
Rupp PA, Little CD. Integrins in vascular development. Circ Res. 2001; 89: 566–572.
Lu E, Wagner WR, Schellenberger U, Abraham JA, Klibanov AL, Woulfe SR, Csikari MM, Fischer D, Schreiner GF, Brandenburger GH, Villanueva FS. Targeted in vivo labeling of receptors for vascular endothelial growth factor: approach to identification of ischemic tissue. Circulation. 2003; 108: 97–103.
Weller GE, Wong MK, Modzelewski RA, Lu E, Klibanov AL, Wagner WR, Villanueva FS. Ultrasonic imaging of tumor angiogenesis using contrast microbubbles targeted via the tumor-biding peptide RRL. Cancer Res. 2005; 65: 533–539.
Linka AZ, Sklenar J, Wei K, Jayaweera AR, Skyba DM, Kaul S. Assessment of transmural distribution of myocardial perfusion with contrast echocardiography. Circulation. 1998; 98: 1912–1920.
Conway EM, Collen D, Carmeliet P. Molecular mechanisms of blood vessel growth. Cardiovasc Res. 2001; 59: 507–521.
Waltenberger J. Impaired collateral vessel development in diabetes: potential cellular mechanisms and therapeutic implications. Cardiovas Res. 2001; 49: 554–560.