Introduction

Angiogenesis, the formation of new blood vessels, is a common process that may occur in physiologic as well as pathologic conditions. Examples for physiologic processes involving angiogenesis are reproduction, development and wound repair. Benign pathologic conditions associated with angiogenesis are for example arthritis, psoriasis and retinopathy [1].

Regarding malignant pathologic processes, angiogenesis is an essential process in the growth of solid tumours [2]. Once tumours have reached a size of >1 mm3, diffusion alone is no longer sufficient to supply the tumour cells with adequate amounts of oxygen and nutrients, and further tumour growth is only possible when new blood vessels are formed. This “angiogenic switch” is a common event in most malignant neoplasias [3, 4] and therefore represents an interesting molecular target, not only for imaging, but also for targeted forms of therapy. Examples for targeted antiangiogenic therapies are cilengitide [5], which inhibits the integrin receptors αvβ3 and αvβ5, and the antibody bevacizumab (Avastin) [6], targeting the vascular endothelial growth factor (VEGF).

A large variety of methods are in use for the assessment of angiogenic activity in tumours, many of them measuring physical parameters of the tissue as surrogates of angiogenesis, e.g. blood flow, blood volume or the permeability of the blood vessels. These methods include dynamic contrast-enhanced computed tomography (DCE-CT) and magnetic resonance imaging (DCE-MRI), ultrasound (US) and positron emission tomography (PET) with perfusion tracers (e.g. [15O]H2O) [7]. Of these methods, DCE-CT and DCE-MRI are most widely used in the clinical setting [8]; however, especially DCE-MRI is technically demanding and standardization of this method is complex [9]. Furthermore, these methods assess physical parameters of the tissue that are related to angiogenesis, but they do not visualize the process of angiogenesis itself. Often the interpretation of the parameters assessed by these methods with regard to their physiologic meaning in terms of angiogenic activity remains challenging.

Imaging of molecular markers related to the angiogenic process might be a more specific method for the assessment of angiogenic activity in tumours. Interesting molecular markers in this regard are the integrins. Integrins are heterodimeric membrane receptors comprised of an α and a β subunit that mediate interactions between cells, cells and the extracellular matrix (e.g. collagen, laminin, fibronectin and vitronectin) and soluble molecules (e.g. growth factors). So far, 18 different α and 8 different β subunits have been identified, forming 24 different integrin receptors [10]. Of these, the integrin αvβ3 has been studied most extensively regarding angiogenesis. Integrin αvβ3 is highly expressed on the cell surface of activated endothelial cells in newly formed blood vessels, thus representing an interesting molecular marker for angiogenesis imaging [11]. However, as integrins may also be expressed on the surface of some tumour cells, and as integrin αvβ3 is currently assumed to play promotional as well as inhibiting roles in the process of angiogenesis, it has to be kept in mind that the interpretation of the signal obtained by integrin imaging is rather complex [12].

Possible future clinical applications for imaging of integrin expression might include patient risk stratification and patient selection for antiangiogenic forms of therapy. As antiangiogenic therapies usually lead to a reduction of tumour growth rather than to tumour shrinkage, morphologic methods solely assessing tumour size (e.g. CT or MRI) do not seem ideal for monitoring these therapies, as therapy effects would only be detected in the later stages of the therapy course. In this regard, imaging of integrins could pose a more sensitive method to early assess the efficacy of antiangiogenic forms of therapy.

In the preclinical setting, a large variety of imaging strategies have been successfully employed for imaging of integrin expression, e.g. radionuclide techniques [PET and single photon emission computed tomography (SPECT)], MRI, targeted US and optical imaging [13]. Radionuclide techniques (SPECT and especially PET) are advantageous owing to their high sensitivity, and up to now, these remain the only techniques that have successfully made the translation from the laboratory into the clinical setting for imaging of integrin expression. PET is significantly more sensitive than SPECT. Using the PET technique, tracer amounts in the picomolar range can be detected in vivo. This allows the use of minimal amounts of tracer, thus minimizing effects of potential pharmacodynamic effects of the tracer substance, as well as minimizing the possibility of receptor saturation, which is especially important regarding imaging of integrin expression, as the concentration of the target molecules (integrins) is usually low. Furthermore, PET also offers the possibility of absolute tracer quantification and it delivers images with a reasonably good spatial resolution, which is about 3–4 mm for clinical scanners [14], and even in the sub-millimetre range for preclinical scanners [15].

All tracers that are used for imaging of integrin expression are based on the tripeptide sequence arginine-glycine-aspartic acid, or RGD in the single letter code. RGD binds to integrins containing the αv subunit [16], which represents an abundant physiologic integrin binding ligand in proteins of the extracellular matrix. Integrin-specific radiotracers are obtained by radiolabelling RGD-based compounds, which can be used for imaging with scintigraphy and SPECT (by using gamma emitters like e.g. 99mTc) or PET (by using positron-emitting radionuclides, e.g. 18F or 68Ga). In this review, we will focus on RGD-based radiotracers which have already been used in first clinical trials and on the most promising candidates for future clinical use. We will present a variety of potential clinical indications for use of such tracers and a short outlook on the potential use for radionuclide therapy.

18F-Labelled monomeric compounds for imaging integrin expression

Gluco-RGD and Galacto-RGD

A very high activity and specificity (against the platelet integrin αIIbβ3) of the RGD peptide for the integrin subtype αvβ3 has been achieved by cyclization of a linear RGD pentapeptide including one of the amino acids (Phe) in the unnatural d-configuration (d-Phe), resulting in the cyclic pentapeptide cyclo(-Arg-Gly-Asp-D-Phe-Val-) or c(RGDfV) developed by Kessler and co-workers [17, 18]. The cyclic structure and the incorporation of the d-amino acid in addition to the very good binding properties guarantee high stability in vivo against enzymatic cleavage. Hence, almost all of the imaging reagents developed were based on this peptide, which was slightly modified to introduce the radiolabel or to construct oligomers. Studies with radioiodinated 125I-c(RGDyV) have shown a high affinity for integrin αvβ3 in vitro and good tumour targeting properties of the peptide in preclinical models. However, in vivo, a predominantly hepatobiliary excretion of the tracer was observed, leading to a decreased image quality due to high unspecific uptake in the liver, bile and intestinal tract. Chemical modifications of a tracer that increase its hydrophilicity, e.g. glycosylation, introduction of hydrophilic amino acids or coupling with polyethylene glycol (PEGylation), can redirect the excretion route more to the renal pathway. Wester and co-workers have increased the hydrophilicity of c(RGDyK) by coupling sugar moieties to the compound, leading to Gluco-RGD, which can be used as a radiotracer after radiohalogenation with iodine isotopes [19]. The next step was the development of Galacto-RGD based on the cyclic peptide c(RGDfK), which can be labelled with 18F. [18F]Galacto-RGD is highly specific for integrin αvβ3, with a high affinity (IC50 ∼5 nM). Biodistribution studies in mice showed a predominantly renal excretion route of the compound with a fast clearance of the tracer from the blood and muscle tissue, and tracer uptake in integrin αvβ3-expressing xenograft tumours was higher than in most reference tissues, except for the kidneys, liver and colon [20]. The tracer showed high stability in vivo; in mice 86% of the tracer in the blood and 87% of the tracer in the tumour was intact 120 min after injection [21]. Tumour uptake of [18F]Galacto-RGD correlated well with the amount of integrin αvβ3 expression in xenograft tumours in a preclinical mouse model which was comprised of compound tumours with different ratios of integrin αvβ3-overexpressing M21 cells and αvβ3-deficient M21-L cells [22].

However, for imaging of angiogenesis, integrin αvβ3 expressed on the surface of activated endothelial cells, not of the tumour cells themselves, should be imaged. In this regard, it has been shown that [18F]Galacto-RGD PET can detect integrin αvβ3 expression on the tumour vasculature in xenograft tumours generated with A431 cells, which are human squamous cell carcinoma cells not expressing integrin αvβ3 on their surface, however extensively inducing angiogenesis [22, 23].

[18F]Galacto-RGD was the first integrin-specific PET tracer to be evaluated in human subjects in 2005. In this study, the tracer showed high variability of tumour uptake [standardized uptake value (SUV) 1.2–10.0] in nine patients with melanoma, sarcoma, renal cell carcinoma and villonodular synovitis. Tracer application was tolerated well, and no side effects of [18F]Galacto-RGD were reported. Rapid blood clearance (predominantly via the kidneys) was observed, and high background uptake was seen in the kidneys, liver, spleen and intestine. [18F]Galacto-RGD showed a high stability with 96% intact tracer in the blood 120 min after tracer injection. Six patients underwent tumour resection, and angiogenesis assessed by immunohistochemistry (αvβ3-positive vessels per field of view) correlated highly with tumour uptake of [18F]Galacto-RGD [21].

In a later study, tracer uptake of [18F]Galacto-RGD was compared with histologic markers of angiogenesis after tumour resection in 19 patients [24]. All lesions which were positive in the [18F]Galacto-RGD PET also were positive for immunohistochemical αvβ3 staining, whereas normal tissues were negative. Histologic markers of angiogenesis (staining intensity of αvβ3 and microvessel density) correlated well with uptake of [18F]Galacto-RGD (SUV and tumour to blood ratio). There was no correlation between tracer uptake and tumour size, showing that tracer uptake was not caused unspecifically by a large tumour volume.

The detection rate of [18F]Galacto-RGD for malignant lesions was 79% in a clinical study including 19 patients with sarcoma, melanoma, malignant fibrous histiocytoma, renal cell cancer and cholangiocellular carcinoma [25]. In another study including 12 patients with squamous cell carcinoma of the head and neck (SCCHN), [18F]Galacto-RGD correctly identified 10 of 12 primary tumours, whereas the detection rate of lymph node metastases was lower (2 of 6) [26] (Fig. 1). In 16 patients with breast cancer, all primary lesions were identified by [18F]Galacto-RGD PET, with no false-positive lesions [27]. Again, sensitivity for lymph node metastases was lower (3 of 8 patients with lymph node metastases were identified), and only 11 of 24 distant metastases were detected by [18F]Galacto-RGD. Regarding primary tumours of the brain, in patients with glioblastoma multiforme, 11 of 12 primary tumours could be visualized by [18F]Galacto-RGD [28]. In these tumours, tracer accumulation was primarily observed in a small rim in the peripheral region of the tumour, with no accumulation in the necrotic centre.

Fig. 1
figure 1

[18F]Galacto-RGD scan of a patient with a lymph node metastasis of squamous cell carcinoma of the uvula and corresponding MRI images; note the tracer accumulation in the periphery of the metastasis, whereas the necrotic centre shows no uptake

Full size image

It is notable that also benign chronic inflammatory lesions like villonodular synovitis can show significant uptake of [18F]Galacto-RGD [22]. This underlines the fact that benign inflammatory lesions (which also contain newly formed blood vessels with activated endothelial cells expressing integrin αvβ3) can exhibit a high uptake of [18F]Galacto-RGD, comparably high to malignant lesions. This is in line with results obtained from a preclinical chronic inflammatory mouse model, showing tracer uptake of [18F]Galacto-RGD in inflammatory lesions [29]. Thus, similar to [18F]fluorodeoxyglucose (FDG), [18F]Galacto-RGD PET cannot be used to clearly differentiate benign from malignant lesions. For example, [18F]Galacto-RGD uptake was also observed in haemangiomas [30].

Furthermore, it has to be kept in mind that specific tracer uptake of [18F]Galacto-RGD can be caused by integrin αvβ3 expression on the tumour neovasculature as well as on the tumour cells themselves. Static PET imaging cannot differentiate these two components. Therefore, to solely assess angiogenesis, static [18F]Galacto-RGD PET can only be regarded as a valid approach in lesions without integrin αvβ3 expression of the tumour cells themselves.

Regarding radiation exposure of the patient, the effective dose resulting from an i.v. application of [18F]Galacto-RGD was reported as 0.017 mSv/MBq for male and 0.020 mSv/MBq for female patients [31], which is comparable to other 18F-labelled PET tracers, e.g. [18F]FDG [32, 33].

Overall, preclinical and clinical results have shown that PET imaging of integrin αvβ3 expression with [18F]Galacto-RGD is feasible, except for organs with high physiologic background uptake of the tracer, which are the liver, spleen and the urogenital tract. As of now, [18F]Galacto-RGD is the most extensively studied integrin αvβ3-specific PET tracer in the clinical setting. However, the radiochemical synthesis of this tracer is complex and automation of this process is challenging, thus limiting the widespread use of [18F]Galacto-RGD in the clinical routine.

[18F]Fluciclatide (AH111585)

Another integrin-specific PET tracer which has already been successfully translated into the clinic is [18F]fluciclatide, which has been developed by GE Healthcare. Integrin-specific binding of this compound is also mediated by an RGD motif. However, in contrast to Galacto-RGD, fluciclatide shows highest binding affinity for integrin αvβ5 (IC50 0.1 nM), followed by integrin αvβ3 (IC50 11.1 nM). By cyclization, introduction of disulphide bridges and coupling to a polyethylene glycol (PEG) spacer, pharmacokinetic properties could be enhanced and degradation in vivo was minimized, making it feasible as a PET tracer in human subjects [34]. In seven patients with breast cancer, [18F]fluciclatide PET was able to visualize all lesions which were also visible on CT. However, in the liver, which shows a higher background uptake, metastases were seen as photopenic lesions. Similar to Galacto-RGD, fluciclatide showed inhomogeneous uptake in tumour lesions (SUV 2.0–40.0).

The compound was stable in vivo, with 74% intact tracer in the blood 60 min after injection. In a dosimetry study performed in eight patients, [18F]fluciclatide showed an effective dose to the patient of 0.026 mSv/MBq [35], which is comparable to [18F]Galacto-RGD and other 18F-labelled PET tracers.

Compared to [18F]Galacto-RGD, an advantage of [18F]fluciclatide is the feasibility of its radiochemical synthesis. Therefore, further clinical data regarding this tracer are eagerly awaited, as a phase II study currently is recruiting patients, which will compare [18F]fluciclatide PET in patients with histologic parameters of angiogenesis and perfusion parameters assessed by DCE-CT [36].

[18F]RGD-K5

[18F]RGD-K5 as a click chemistry-derived peptidomimetic tracer based on the RGD sequence for integrin-specific binding was developed by Siemens Molecular Imaging Inc. Its selectivity for integrin αvβ3 is 2.3 times higher than for other related integrins, the affinity for integrin αvβ3 was reported as Kd = 7.9 nM [37] and preclinical evaluation showed a predominantly renal excretion and a high stability in vivo (98% intact tracer in the blood 60 min after tracer injection in mice), which is favourable for imaging studies.

No adverse effects of [18F]RGD-K5 were observed in a first evaluation in 12 patients with breast cancer [38]. In this study, all patients also underwent [18F]FDG PET. Of a total of 157 FDG-positive lesions, 122 were also seen with [18F]RGD-K5. In most lesions, FDG uptake was higher compared to [18F]RGD-K5 and there was no correlation between the uptake of these two tracers, which is in line with the results obtained by studies with other RGD-based integrin-specific tracers.

However, in a histologic analysis, no correlation was observed between [18F]RGD-K5 uptake and histologic parameters of angiogenesis (e.g. microvessel density). This might be explained by the binding properties of [18F]RGD-K5, which is not specific for integrin αvβ3, but also binds to other integrins, as well as the fact that integrin αvβ3 is not exclusively expressed on activated endothelial cells, but may also be expressed on the tumour cell surface.

The radiation exposure to the patient was evaluated in a dosimetric study in four patients, which resulted in an estimated effective dose of 0.015 mSv/MBq of [18F]RGD-K5 [39].

64Cu- and 68Ga-labelled monomeric compounds for imaging integrin expression

64Cu-DOTA-RGD

DOTA was coupled by Chen et al. to c(RGDyK) and complexed with 64Cu for PET imaging. The biodistribution of the resulting compound 64Cu-DOTA-RGD was compared to the radiohalogenated tracers 125I-RGD and [18F]FB-RGD in xenograft models of breast cancer (MDA-MB-435 cells). Of these, 125I-RGD showed the highest tumour uptake and retention, which is most likely due to the decreased affinity of 64Cu-DOTA-RGD and [18F]FB-RGD to integrin αvβ3, due to the bulky fluorobenzoyl (FB) and DOTA groups close to the RGD motif in these compounds. Furthermore, 64Cu-DOTA-RGD showed a prolonged activity in the blood pool and an unfavourably high background activity in the liver, most likely due to the dissociation of 64Cu from the chelate, indicating the need for development of more stable chelators for Cu isotopes. MicroPET imaging with 64Cu-DOTA-RGD in mice was reported to be feasible [40].

Pharmacokinetic properties of 64Cu-DOTA-RGD could be improved by the introduction of a PEG spacer into the compound. In vivo, 64Cu-DOTA-PEG-RGD showed a significantly lower background uptake in the liver, as well as a fast blood and kidney clearance. High tumour to organ ratios of 64Cu-DOTA-PEG-RGD could be observed by microPET imaging of glioblastoma xenograft tumours (U87MG cells) in mice [41].

68Ga-NOTA-RGD

68Ga is a radiometal that can be coupled to radiotracers via chelators, e.g. DOTA (1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid) or NOTA (1,4,7-triazacyclononane-1,4,7-triacetic acid), forming stable radiometal chelates. The positron emitter 68Ga is attractive for PET imaging for several reasons. Compared to 18F-labelled compounds, 68Ga has the advantage that it can be eluted from commercially available 68Ge/68Ga generator systems, which makes its use feasible also in centres that do not have access to an on-site cyclotron. The short half-life of 68Ga (68 min) is ideal for labelling of small peptidic tracers with fast pharmacokinetics in vivo [42].

68Ga-NOTA-RGD was produced by coupling SCN-Bz-NOTA [2-(p-isothiocyanatobenzyl)-NOTA] to the RGD-containing cyclic peptide c(RGDyK) over the lysine side chain. Labelling with 68Ga was possible by incubating at room temperature over only 10 min. An in vitro binding assay showed a high affinity (Ki = 3.6 nM) of 68Ga-NOTA-RGD to integrin αvβ3, indicating that the conjugation with NOTA does not substantially affect the binding affinity of c(RGDyK). In mice, 68Ga-NOTA-RGD was excreted predominantly via the renal pathway, and therefore the highest uptake was observed in the kidneys, whereas uptake in the liver and intestine was substantially lower. Colorectal xenograft tumours (SNU-C4 cells) showed a very high tumour to blood and tumour to muscle ratio (10.3 and 9.3, respectively), and 68Ga-NOTA-RGD was reported to be feasible for microPET imaging [43]. A similar strategy has been followed by Haubner et al. with the compound 68Ga-NODAGA-RGD [44].

Overall, 68Ga-labelled RGD peptides have a high potential for translation to the clinic. In a first clinical study, [18F]FDG and 68Ga-NOTA-RGD PET was performed in six patients with hepatic metastases of colorectal carcinoma. Mildly elevated uptake of 68Ga-NOTA-RGD in liver metastases was seen in three of six patients, and the other half was negative. After antiangiogenic combination therapy with FOLFOX and bevacizumab, a VEGF antibody, patients who had an initially elevated uptake of 68Ga-NOTA-RGD showed a partial response, whereas the other half had stable or progressive disease [45].

Multimeric compounds

Despite the chemical modifications mentioned above to improve the pharmacokinetic properties of RGD peptides, like coupling to sugar moieties or PEGylation, it is still desirable to further improve tumour uptake of these compounds, especially for detection of lesions in areas with high physiologic background uptake of the tracer, e.g. in the liver, where metastases can often only be seen as photopenic lesions. As the interaction between integrins and their physiologic binding partners in the extracellular matrix involves multivalent binding sites with clustering of integrins, the concept of developing multimeric tracers to improve avidity to integrin αvβ3, thus enhancing specific tumour uptake, seems promising.

The groups of Wester and Kessler synthesized monomeric, dimeric and tetrameric RGD tracers by connecting multiple cyclic c(RGDfE) peptides via PEG linkers, using lysine moieties as branching elements [46]. In vitro, the binding affinity of these compounds to integrin αvβ3 increased according to the number of RGD peptides in the compound. This was confirmed by in vivo biodistribution studies in mice bearing integrin αvβ3-expressing M21 melanoma xenograft tumours, which showed improved excretion behaviour and tumour uptake in the series monomer < dimer < tetramer. Furthermore, a preclinical study comparing the RGD tetramer containing four c(RGDfE) peptides with a pseudo-monomeric tetramer comprised of only one c(RGDfE) and three non-integrin binding c(RaDFE) structures showed a three times lower tumour uptake of the pseudo-monomeric tetramer, which indicates that the higher tumour uptake was indeed caused by multimerization and was not based on other structural effects.

The group of Chen synthesized a number of RGD peptides while optimizing pharmacokinetic properties by multimerization. In a murine breast cancer model using orthotopic MDA-MD-435 xenograft tumours, biodistribution studies, microPET imaging and autoradiography showed a better tumour retention of the two 64Cu-labelled RGD dimers DOTA-E[c(RGDyK)]2 and DOTA-E[c(RGDfK)]2, compared to their monomeric analogues [47]. A later study showed that the tetramer 64Cu-DOTA-E[E[c(RGDfK)]2]2 had an about threefold and almost tenfold increased integrin avidity compared to its dimeric and monomeric analogue, respectively, and the in vivo tumour to non-tumour contrast of the tetramer was high [48]. Integrin αvβ3 affinity and specificity as well as in vivo tumour uptake and retention could be even further increased by development of the octamer 64Cu-DOTA-E[E[E[c(RGDyK)]2]2]2 [49].

Further improvements could be achieved by introducing PEG or Gly spacers into dimeric RGD compounds, optimizing the distance between the two RGD motifs, better allowing simultaneous integrin binding. Accordingly, it has been shown that 64Cu- and 68Ga-labelled dimers containing PEG or Gly spacers showed higher affinity to integrin αvβ3 in vitro and higher tumour uptake in vivo, as compared to the unmodified RGD dimer [50, 51].

Recently, the group of Wester et al. developed the promising tracer TRAP(RGD)3, a c(RGDfK)-trimer, which uses tri-azacyclononane-phosphinic acid (abbreviated as TRAP), a compound derived from NOTA, as a chelating agent for 68Ga. The advantages of the chelator TRAP over DOTA and NOTA are a more quantitative incorporation of 68Ga into the complex at low ligand concentrations and at low temperatures, even in strongly acidic media, facilitating tracer labeling with very high specific activity. Compared to [18F]Galacto-RGD, 68Ga-TRAP(RGD)3 revealed a 7.3-fold higher affinity for integrin αvβ3 and a 3.9-fold higher tumor uptake in a preclinical M21 melanoma xenograft model [52].

Regarding 18F-labelled multimeric compounds, especially the RGD dimer [18F]FRGD2 ([18F]fluorobenzoyl-E[c(RGDyK)]2) is a promising candidate for translation to the clinic. Its in vitro affinity to integrin αvβ3 has been reported as 55 nM (IC50) [53]. Preclinical in vivo evaluation showed that the uptake of [18F]FRGD2 in U87MG human glioblastoma xenograft tumours was significantly higher compared to its monomeric analogue [18F]fluorobenzoyl-c(RGDyK) [54], and a linear relationship was observed between tumour to background ratio of [18F]FRGD2 and tissue integrin expression [55], which shows that static [18F]FRGD2 PET scans can be used to quantify integrin expression, which is favourable for the translation of this tracer to the clinic.

Similar to the 64Cu- and 68Ga-labelled RGD dimers described above, the introduction of PEG spacers into 18F-labelled RGD dimers to optimize the distance between the two RGD motifs also resulted in an increased in vitro affinity to integrin αvβ3, as well as increased tumour uptake and tumour to non-tumour ratios in preclinical in vivo models [56].

18F-Labelled RGD tetramers have also been developed. Due to spatial hindrance, only low yields were obtained by direct labelling of the tetramer E[E[c(RGDyK)]2]2 with [18F]fluorobenzoate; however, reasonable yields were obtained by labelling the PEGylated tetramer PEG-E[E[c(RGDyK)]2]2. In a mouse model using U87MG human glioblastoma xenografts, the absolute tumour uptake of the tetramer was higher compared to its dimeric analogue; however, also background activity increased, resulting in similar tumour to background ratios for the tetramer and the dimer [57].

However, it has to be kept in mind that a higher absolute tumour uptake of multimeric tracers does not obligatorily lead to improved imaging characteristics. A preclinical in vivo comparison in a mouse model using U87MG human glioblastoma cells between [18F]Galacto-RGD monomer and 18F-labelled PEGylated or glycosylated RGD dimers showed a higher absolute uptake of the dimeric compounds in the tumour; however, the tumour to background ratio of the dimeric tracers was not higher compared to the [18F]Galacto-RGD monomer [58].

Recently, the PEGylated dimeric RGD peptide PEG3-E[c(RGDyk)]2 or [18F]FPPRGD2, was successfully tested in volunteers with no side effects and favourable pharmacokinetic and biodistribution properties [59]. However, no clinical data are available directly comparing monomeric and multimeric integrin-specific tracers in patients. Further clinical studies are now eagerly awaited.

99mTc-Labelled compounds for planar scintigraphy and SPECT imaging

Though in recent years many groups concentrated their efforts on the development of integrin αvβ3-specific PET tracers, also tracers labelled with 99mTc for SPECT imaging or planar scintigraphy have been developed and have already shown promising results in preclinical and clinical studies. Compared to 18F-based PET tracers, 99mTc-based tracers are more widely available as 99mTc can be conveniently eluted from commercially available generator systems and does not require access to a cyclotron. Furthermore, gamma camera systems for planar scintigraphy and SPECT are more widely spread than PET cameras. However, compared to PET imaging, drawbacks of SPECT are its lower spatial resolution and lower sensitivity. Furthermore, absolute quantification with SPECT imaging is challenging, and long image acquisition times do not allow whole-body SPECT imaging in the routine clinical setting.

99mTc-αP2

99mTc-αP2 is a peptide comprised of ten amino acids that contains two copies of the RGD motif for integrin-specific binding, and coupling to 99mTc was accomplished over cysteine residues introduced into the peptide. 99mTc-αP2 for imaging of integrin expression had been reported in 1988, even before integrin-specific PET tracers became available [60]. A clinical study in patients with malignant melanoma revealed a good detection rate of metastases in the neck, axilla, abdomen and soft tissue; however, sensitivity was reported to be low in the thorax due to high blood pool activity in the heart. Optimal imaging time for best tumour to background ratio was reported as 60–120 min post-injection (p.i.). However, the compound’s specificity regarding integrin binding has not been evaluated. Also no data are available about the correlation between tracer uptake and integrin expression in vivo.

99mTc-NC100692

Recently, GE Healthcare introduced NC100692, a new RGD-containing peptide labelled with 99mTc for SPECT imaging. 99mTc-NC100692 has been shown to bind to integrin αvβ3 with high affinity (Ki ∼ 1 nM) [61]. A clinical study in patients with breast cancer showed a good detection rate of 99mTc-NC100692 for malignant lesions larger than 1 cm; overall 19 of 22 tumours could be detected [62]. Further studies in patients with breast and lung cancer showed a high sensitivity for metastases in the lung and brain, but a poor detection rate regarding metastases to the liver and the bone [63].

Potential clinical use of integrin imaging with radiolabelled peptides

Assessment of malignant potential and patient selection for antiangiogenic forms of therapy

Integrins act as receptors for growth factors, proteases and protease inhibitors, transducing intracellular signals regulating gene transcription, cell growth and cell survival [64]. They are involved in the regulation of cell adhesion, motility and migration. Some integrins, e.g. αvβ3, αvβ5, α5β1, α6β4 and α9β13, have been shown to be associated with tumour progression [65]. For example, it is established that integrin αvβ3 plays an important role in the progression of malignant melanoma, regarding the change from superficial to horizontal growth [66, 67]. It is intriguing that especially in patients with melanoma pronounced heterogeneities in the uptake of [18F]Galacto-RGD were observed [25]. Regarding sarcomas, uptake of [18F]Galacto-RGD was dependent on the grade of differentiation, with high-grade tumours showing a higher tracer uptake (and high αvβ3 expression by histology), whereas low-grade tumours displayed a lower tracer uptake (and also lower αvβ3 expression) [24] (Fig. 2). Preclinical studies have shown that integrin αvβ3 also plays a significant role in breast cancer regarding tumour aggressiveness and metastatic potential [68, 69], and targeted antiangiogenic therapies have shown promising results in breast cancer xenograft models [70, 71]. Integrin αvβ3 has also been reported to be associated with malignant potential in gliomas [28], and antiangiogenic therapy with cilengitide, an αvβ3 antagonist, has shown promising results in clinical studies in patients with malignant glioma [5, 72, 73]. Therefore, RGD PET might be a useful tool for patient risk stratification. Furthermore, this also emphasizes that imaging of integrin αvβ3 might also be useful in the selection of patients for antiangiogenic forms of therapy (Fig. 3).

Fig. 2
figure 2

Upper row Galacto-RGD PET coronal, 60 min p.i. Lower row MRI post Gd-DTPA coronal. Whereas the well-differentiated G1 chondrosarcoma in the right shoulder and the dedifferentiated G3 chondrosarcoma in the right pelvis both show contrast enhancement in MRI, the G3 tumour shows intense tracer uptake (especially in the upper parts of the tumour), but the G1 tumour shows only faint tracer uptake

Full size image
Fig. 3
figure 3

[18F]Galacto-RGD PET scan of a prostate cancer patient with pelvic lymph node metastases (a–c, arrows). The pelvic enlarged lymph nodes show intense tracer uptake (a); however, the MRI shows more enlarged and suspicious lymph nodes along the right iliac vessels (b). Image fusion of MRI and PET helps to localize the tracer uptake to the lymph nodes (c arrow, closed tip). However, some lymph nodes do not show tracer uptake (arrow, open tip), which suggests very heterogeneous αvβ3 expression in the lesions

Full size image

This is corroborated by the results of a clinical study performing 68Ga-NOTA-RGD PET in six patients with hepatic spread of colorectal carcinoma, before initiation of antiangiogenic combination therapy with FOLFOX and bevacizumab, a VEGF antibody. Three patients who had elevated uptake of 68Ga-NOTA-RGD showed a partial response to the therapy, whereas the other half who did not show uptake of 68Ga-NOTA-RGD in the liver metastases had stable or progressive disease during therapy [45].

Next to the expression of integrin αvβ3, also [18F]FDG uptake is believed to correlate with the malignant potential in several tumour entities [74–76]. Based on these considerations, it could be speculated that [18F]Galacto-RGD and [18F]FDG provide identical or similar clinical information, despite their different molecular mechanisms of accumulation. This topic was evaluated in a study comparing these two tracers in 18 patients with non-small cell lung cancer (NSCLC), renal cell carcinoma, rectal carcinoma, sarcoma, squamous cell carcinoma of the head and neck, breast cancer and carcinoid [77]. In general, tumour uptake of [18F]FDG was significantly higher compared to [18F]Galacto-RGD, and accordingly the detection rate of lesions was higher for [18F]FDG. This can be explained by the different binding mechanisms of the tracers. FDG is taken up into the tumour cells, whereas Galacto-RGD binds to integrin αvβ3 predominantly expressed on the surface of activated endothelial cells, which are much less abundant in tumour lesions compared to the amount of tumour cells. There was no correlation between uptake of [18F]Galacto-RGD and [18F]FDG in the tumour lesions. Interestingly, bone metastases showed a tendency for higher uptake of [18F]Galacto-RGD, which is in line with the observation that integrin αvβ3 is an indicator for metastatic potential, especially regarding metastases to the bone [78]. Furthermore, integrin αvβ3 is also expressed on osteoclasts, which are involved in the bone resorption process in osseous metastases [79].

Overall, from the lack of correlation between FDG and Galacto-RGD uptake in malignant lesions, these two tracers seem to provide different information. Due to the higher tumour uptake, FDG is more feasible for tumour detection and staging, whereas Galacto-RGD might provide additional information, e.g. for patient selection for antiangiogenic forms of therapy or for therapy monitoring. It still remains to be evaluated if multimeric RGD tracers provide a higher sensitivity for tumour staging; however, as of now, no clinical studies with these tracers have been performed in cancer patients.

Therapy monitoring

Next to patient selection for targeted antiangiogenic forms of therapy, imaging of integrin αvβ3 expression might also be an interesting tool for monitoring antiangiogenic therapies. This is of particular relevance as these forms of therapy usually lead to a stop of tumour growth, rather than to tumour shrinkage. Thus, conventional staging methods that solely assess morphologic aspects like tumour size do not seem ideal, as these might take a long time to assess response in terms of tumour growth arrest. On the other hand, functional imaging of angiogenesis with RGD-based peptides might pose a tool for early prediction of clinical response to non-cytotoxic antiangiogenic forms of therapy.

A preclinical study using lung cancer (LLC) xenograft tumours in mice evaluated the ability of the integrin-specific PET tracer [18F]fluciclatide to monitor the effect of targeted antiangiogenic therapy with ZD4190, a VEGF receptor tyrosine kinase inhibitor. After therapy, a decreased uptake of [18F]fluciclatide was observed in treated tumours, whereas tracer uptake of non-treated tumours increased. In the same study, tumours treated with paclitaxel showed a reduction of [18F]fluciclatide uptake, correlating with a reduction of microvessel density of the tumours, whereas tumour uptake of [14C]FDG did not decrease after therapy [80]. It is notable that the tumour model used in this study (LLC) does not express integrin αvβ3 on the cell surface; thus, the signal derived from [18F]fluciclatide PET is mainly dependent on the integrin expression on activated endothelial cells of the tumour neovasculature. These results indicate that molecular imaging of integrin expression might be useful in detecting early vascular changes in response to therapy, and in the case of paclitaxel treatment [18F]fluciclatide PET was more sensitive than [18F]FDG PET.

Similar results were obtained in a glioblastoma xenograft model using targeted therapy with sunitinib, which inhibits the VEGF receptor and platelet-derived growth factor (PDGF) receptor tyrosine kinases. Sunitinib-treated tumours showed a significant reduction of [18F]fluciclatide uptake which correlated with a histologically reduced microvessel density, as opposed to non-treated tumours [81].

Dumont et al. evaluated the Src kinase inhibitor dasatinib in glioblastoma xenograft tumours. Src kinases are involved in the activation and clustering of integrins to focal adhesions on the cell surface. Tumours treated with dasatinib showed a reduced uptake of the integrin-specific tracer 64Cu-DOTA-c(RGDfK), whereas FDG uptake did not change significantly after therapy. However, on histologic analysis, neither the tumour vasculature nor integrin expression was decreased in treated tumours. Thus, the decrease of 64Cu-DOTA-c(RGDfK) uptake was most likely related to inactivation of integrin αvβ3 [82]. It has to be noted that this study focused on integrin expression of the tumour cells and not of endothelial cells; thus, the relevance of these results regarding monitoring antiangiogenic therapies needs to be further elucidated.

Next to early assessment of therapy response, imaging of integrin αvβ3 might also be useful for dose finding in anti-integrin αvβ3 targeted therapies, e.g. with cilengitide. Assessment of receptor saturation caused by the therapeutic agent might help to minimize therapy inefficiency by undertreatment as well as cost-ineffective overtreatment.

Regarding monitoring antiangiogenic therapies, no clinical data are available on this subject yet. Currently a trial is recruiting patients undergoing targeted therapy, which will evaluate the uptake of [18F]fluciclatide before and after antiangiogenic therapy [83]. Another clinical trial is also currently recruiting patients undergoing combination treatment with bevacizumab and chemotherapy to evaluate the utility of [18F]RGD-K5 PET to monitor early treatment response to targeted antiangiogenic therapy [84].

Radionuclide therapy

An interesting concept is to use the tumour targeting properties of RGD peptides not only for diagnostic applications and therapy monitoring, but to convert these tracers into therapeutically active compounds by exchanging the diagnostic gamma- or positron-emitting radionuclides with therapeutically active radioisotopes, e.g. beta emitters. This concept, known as peptide receptor radionuclide therapy (PRRT), has already found entry into the clinical routine especially in cases of neuroendocrine tumours, which typically overexpress somatostatin receptors, targetable with radiolabelled somatostatin analogues. In recent decades, somatostatin analogues labelled with the Auger emitter 111In, and more recently with the beta emitters 90Y and 177Lu, have been in clinical use worldwide and have been successfully employed in a number of trials, yielding promising results [85]. Somatostatin receptor-specific tracers, e.g. DOTATOC and DOTATATE, exhibit excellent characteristics for PRRT, as absolute tumour uptake and tumour to non-tumour ratios are very high, leading to high cytotoxic doses in the tumour tissue, while sparing most normal tissues.

Janssen and co-workers explored the possibility of radionuclide therapy in a preclinical model with a radiolabelled RGD dimer, 111In/90Y-DOTA-E-[c(RGDfK)]2 and 99mTc-HYNIC-E-[c(RGDfK)]2. After i.v. injection of the maximum tolerated dose of 90Y-labelled peptide (37 MBq), a significant growth delay of subcutaneous xenograft tumours (human OVCAR-3 cell line) was noted in comparison to mice treated with 90Y-labelled scrambled peptide or untreated controls (tumour doubling times were 5.2, 3.9 and 3.2 days, respectively). However, considerable uptake of tracer in non-target organs was noted in the spleen, liver and kidneys, which limits the dose that can be safely administered [86].

Cu-labelled RGD peptides are interesting compounds regarding radionuclide therapy, as copper offers diagnostic (60Cu, 61Cu, 62Cu and 64Cu) as well as therapeutic (64Cu and 67Cu) isotopes. This would facilitate pretherapeutic dosimetry, as both the diagnostic tracer as well as the therapeutically active radiopharmaceutical exhibit the same distribution and kinetics in the body [41]. This is not necessarily the case for tracer pairs labelled with different elements, e.g. 111In/90Y or 68Ga/177Lu.

However, though marked progress can be noted in the last decade regarding the development of RGD-based integrin-specific tracers for targeting angiogenesis regarding tumour uptake, excretion pathway and clearance from normal tissues, the application of the currently available compounds as carriers for therapeutically active radioisotopes like 90Y or 177Lu in humans does not yet seem practicable. High tumour to non-tumour ratios have to be achieved to deliver tumouricidal radiation doses to malignant lesions, while sparing the organs at risk, which are predominantly the bone marrow, kidneys and liver, which does not yet seem feasible.

Conclusion

The approach of using radiolabelled RGD peptides for imaging of integrin expression has successfully been translated from bench to bedside. However, clinical data on the ultimate clinical value of imaging of integrin expression are still necessary. A potential indication might be patient selection and monitoring in the context of antiangiogenic or combined cytotoxic/antiangiogenic therapy, whereas the use of RGD peptides for radionuclide therapy does not seem realistic in the near future. Thus, in the next step, large-scale trials using radiolabelled RGD peptides within the context of response assessment or evaluation of patient prognosis are warranted to define the ultimate role of imaging of integrin expression in the clinic.