New Pansomatostatin Ligands and Their Chelated Versions: Affinity Profile, Agonist Activity, Internalization, and Tumor Targeting

Somatostatin [somatotropin release-inhibiting factor (SRIF)] is a tetradecapeptide with potent inhibitory actions on several tissues, such as the pituitary, the endocrine pancreas, and the gastrointestinal tract. SRIF also acts as neurotransmitter and neuromodulator (1, 2). The inhibition of the many secretory processes of hormones, such as growth factors, insulin, and glucagon, is an important property of SRIF and is mediated by at least five SRIF receptor subtypes (sst1-sst5), which belong to the G protein–coupled receptor family. Because of the short plasma half-life of <3 min, SRIF-14 (and SRIF-28) does not have therapeutic potential in the treatment of conditions related to pathologic secretory processes. Therefore, metabolically stabilized somatostatin-based peptides were developed (3–7). Two of them, octreotide (Sandostatin) and lanreotide (Somatuline), are being used in the clinic. They display high affinity only to sst2 and moderate affinity to sst3 and sst5. It is therefore conceivable to search for somatostatin analogues with a higher affinity binding profile to all five receptor subtypes (pansomatostatin character) with the aim to improve the established therapeutic potential or discover new indications. Recently, several such molecules have been discovered and developed, such as the cyclohexapeptide SOM 230 (8), which shows high affinity to sst1, sst2, sst3, and sst5 and has potent, long-lasting inhibitory effects on growth hormone and insulin-like growth factor-I release. As part of a small structure-activity relationship (SAR) study, we have synthesized a cyclooctapeptide with pansomatostatin characteristics (KE108; Table 1), which shows equivalent (sst1) or even higher affinity to sst2, sst3, sst4, and sst5 compared with SRIF-28 (9). KE108 was found to be an agonist to all five sst subtypes. This molecule is currently studied for its clinical potential.

In addition to the use of these peptides in the pathologies described above, somatostatin-based peptides were modified for radiolabeling and were successfully introduced into the clinic, both for diagnostic and targeted radiotherapeutic application (10, 11).

The aim of this study was 2-fold, namely, to do a SAR study to better understand the structural variables responsible for a pan character and, particularly, we wanted to design chelator-coupled peptides for radiometal labeling that will allow high flexibility with regard to potential clinical applications in single-photon emission computed tomography, positron emission tomography, and targeted radionuclide therapy.

The design of pansomatostatin radioligands may be of potential interest as the majority of human sst-positive tumors simultaneously express multiple sst subtypes, although there is considerable variation and sst2 is the most frequently expressed subtype (12).

Radiolabeled pansomatostatin ligands have not been developed yet. Our original work was based on cyclo(GABA-Asn-Phe-Phe-D-Trp-Lys-Thr-Phe) (HR2215; ref. 13). This peptide showed high to intermediate affinity to all sst subtypes and therefore constituted already a very promising lead compound. Based on it, we first synthesized a small series of peptides without a handle for the coupling of chelators and selected the most interesting ones for chelator coupling. The parent peptide of these is cyclo(D-Dab-Arg-Phe-Phe-D-Trp-Lys-Thr-Phe) (KE121). Earlier, we had coupled Tyr NH₂ terminally as a prosthetic group for iodination (9) and in this work 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA; KE88) or 1,4,7,10-tetraazacyclododecane-1,4,7,-triacetic acid,10-glutaric acid (DOTAGA; KE87) for radiometal labeling.

Reagents.⁶⁷GaCl₃ and ¹¹¹InCl₃ were purchased from Mallinckrodt. The prochelator DOTA(tBu)₃ was synthesized according to Heppeler et al. (14) and DOTAGA(tBu)₄ [1-(1-carboxy-3-carbotertbutoxypropyl)-4,7,10-(carbotertbutoxymethyl)-1,4,7,10-tetraazacyclododecane] was synthesized according to Eisenwiener et al. (15).

Chemical syntheses. Two groups of peptides were synthesized: (a) cyclic octapeptides incorporating an aminobutyric acid bridging unit and (b) octapeptides incorporating diaminobutyric acid (Dab) whose α-amino group allows the coupling of a chelator for radiometal labeling.

Synthesis of the pansomatostatin derivatives. The solid-phase peptide synthesis was carried out on a semiautomatic peptide synthesizer using the fluorenylmethoxycarbonyl strategy (16). Details on the synthesis will be published elsewhere.

The synthesis of chelated versions of pansomatostatin ligands with our lead KE88 as an example is depicted in Scheme 1.

Formation of metal complexes and radiolabeling. The DOTA-SRIF analogues were complexed either with Y(NO₃)₃ · 5 H₂O or with Ga(NO₃)₃ · 9 H₂O, as described by Ginj et al. (16). The radiolabeling of these peptide conjugates was done according to Wild et al. (17) and the radiopeptides were obtained in >97% radiochemical purity at specific activities of >37 GBq/μmol peptide.

Binding affinity determination to hsst1-hsst5. The hsst1-hsst5 binding affinity of the various compounds was measured as described previously using in vitro receptor autoradiography with 20-μm-thick sections from membrane pellets of the respective transfected cells (18).

Adenylate cyclase activity. The effect of KE88/Y on forskolin-stimulated cAMP formation was done on hsst1-hsst5–transfected cells, as described previously (9, 19, 20).

In vitro receptor internalization studies. Immunofluorescence microscopy–based internalization assay for sst2 and sst3 was done as previously described by Cescato et al. (21) using HEK-293 cells stably expressing the human sst2 and sst3. Cells were treated with KE88, Y^III-DOTA-TOC, or Y^III-DOTA-NOC (17) at a concentration of 10 nmol/L to 10 μmol/L for 30 min at 37°C in growth medium and then rinsed twice with 100 mmol/L phosphate buffer containing 0.15 mol/L sucrose (PS). After the cells were fixed and made permeable for 7 min with cold methanol (−20°C), they were rinsed twice with PS, and nonspecific binding sites were blocked by incubating the cells in PS containing 0.1% bovine serum albumin for 60 min at room temperature. The cells were subsequently incubated for 60 min at room temperature either with the sst2A-specific antibody (R2-88) or with the sst3-specific primary antibody (SS-850) diluted 1:1,000 in PS (21). The sst3-specific antibody was purchased from Gramsch Laboratories. After the antibody incubation, the cells were washed thrice for 5 min with PS containing 0.1% bovine serum albumin and then incubated for 60 min at room temperature in the dark with the secondary antibody Alexa Fluor 488 goat anti-rabbit IgG (H+L; Molecular Probes, Inc.; Invitrogen) diluted 1:600 in PS. Subsequently, the cells were washed thrice for 5 min each with PS containing 0.1% bovine serum albumin, embedded with 1:1 PS/glycerol, and covered with a glass coverslip. The cells were imaged using a Leica DM RB immunofluorescence microscope and an Olympus DP10 camera.

Radioligand internalization. The AR4-2J cell line and the HEK-293 cells stably expressing sst2, sst3, or sst5 receptors were maintained by serial passage in monolayers in DMEM supplemented with 10% fetal bovine serum, l-glutamine, penicillin-streptomycin, and 500 μg/mL G418 (for HEK cells) or amino acids and vitamins (for AR4-2J cells) in a humidified 5% CO₂ atmosphere at 37°C (22). For all cell experiments, cells were seeded at a density of 0.9 to 1.1 million per well in six-well plates and incubated overnight in growth medium to obtain good cell adherence. The loss of cells during the internalization experiments was <10%.

Furthermore, the internalization rate was linearly corrected to 1 million cells per well in all cell experiments. The procedure for internalization was done according to Ginj et al. (16). To determine nonspecific membrane binding and internalization, cells were incubated with radioligand in the presence of 1 μmol/L HR2252.

Biodistribution and imaging studies in tumor-bearing nude mice. Animals were kept, treated, and cared for in compliance with the guidelines of the Swiss regulations (approval 789). Five-week-old athymic female Swiss nude mice were implanted s.c. with 10 to 12 million AR4-2J or HEK-sst2 cells on one flank and HEK-sst3 cells on the other, freshly suspended in sterile PBS. Fourteen days after inoculation, the mice showed solid palpable tumor masses (tumor weight, 70-150 mg) and were used for biodistribution studies and imaging. Three to six mice were injected under ether anesthesia with 0.15 to 0.2 MBq of 10 pmol [¹¹¹In]KE88 and [⁶⁷Ga]KE88, respectively, in 0.05 mL NaCl solution (0.9%, 0.1% bovine serum albumin) into the tail vein. At 15 min, 30 min, 1 h, 4 h, 24 h, and 48 h after injection, mice were sacrificed under ether anesthesia. Organs and blood were collected and the radioactivity in these samples was determined using a gamma counter.

To determine the nonspecific uptake of the radiopeptide, mice were injected with 50 μg KE88 and 50 μg In^III-DTPA-TATE or 50 μg DOTA-TATE in 0.1 mL NaCl solution (0.9%, 0.1% bovine serum albumin) as a coinjection with the radioligand. The radioactivity uptake in the tumor and normal tissues of interest was expressed as percentage of the injected (radio)activity per gram tissue (% IA/g).

To study the effect of lysine on the kidney uptake of [¹¹¹In]KE88, four mice were coinjected with a 0.1 mL solution containing 10 mg lysine in PBS with the radiopeptide; they were sacrificed at 1 h after injection. Two mice were used for imaging studies. They were injected with 0.5 MBq of 50 pmol [¹¹¹In]KE88. Fifteen min, 30 min, 1 h, 2 h, and 4 h after injection, the mice were anesthetized and images were acquired in the prone position using a γ-camera equipped with a medium energy parallel hole collimator (Basicam, Siemens).

Statistical methods. To compare differences between the radiopeptides, the Student's t test was used. P < 0.05 was considered to be significant.

Syntheses and radiolabeling. The synthetic steps toward the new octapeptides and the structure of our lead KE88 are depicted in Scheme 1. The compounds were characterized by electrospray ionization-mass spectrometry and analytic high-performance liquid chromatography. The yields of the chelated peptides were ∼40%. Complexation yields with the metal salts Y(NO₃)₃ · 5 H₂O and Ga(NO₃)₃ · 9 H₂O were >95%. Labeling yields using ¹¹¹InCl₃ and ⁶⁷GaCl₃, respectively, were ≥97% at specific activities of 37 GBq/μmol peptide.

sst affinity profiles.Table 1 shows the normalized “IC₅₀ values” of the peptides studied in this work for the five sst subtypes with SRIF-28 as the control peptide set at 1. The table also contains KE108, a pansomatostatin peptide, which was published recently (9).

All peptides, except KE105, showed a broad SRIF receptor affinity profile with pansomatostatin character. The lead HR2215 exhibited 27 times lower affinity to sst1, 4.1 times lower affinity to sst2, and 2.3 times lower affinity to sst4 than SRIF-28. On sst3 and sst5, the two peptides were almost equipotent. The replacement of Asn in position 5 (numbering used as for SRIF-14) by the positively charged Arg (HR2252) resulted in an improvement of affinity on all receptor subtypes, whereas Lys in position 5 instead of Arg resulted in a distinct decrease of sst2 affinity (data not shown). Phe⁷ replaced by Tyr (KE109 versus HR2252) led to a reduction of sst1 affinity by a factor of 2.4 and of sst2 affinity by ∼50%, whereas this modification did not change the sst3 and sst5 affinity but improved the sst4 affinity 4-fold. Introducing D-Dab (KE121) as a cyclizing unit instead of γ-aminobutyric acid improved the overall potency, giving values that were close to the sst1 affinity of SRIF-28 and showing a strikingly improved affinity to all other sst subtypes. Replacement of Phe⁷ in this compound for Tyr⁷ (KE106B versus KE121) resulted in a loss of affinity of all receptor subtypes except sst4. KE105 was studied for its potential sst1 selectivity. Indeed, the peptide showed good sst1 affinity and very low affinity to the other receptors. The affinities dropped by a factor of >10 on DOTA conjugation (data not shown).

We therefore selected KE121 as new lead for modification and chelator coupling. The DOTA coupling and Y^III complexation has a minor influence on the sst1 affinity, but the sst2 affinity dropped 8-fold and sst3 and sst5 affinities dropped 2- and 3-fold, respectively. No change was observed on the sst4 affinity.

When replacing Y^III with Ga^III (KE88/Ga versus KE88/Y), some loss was observed on sst1 affinity but a 2.5-fold improvement on sst2, and no change on sst3 and sst5 but a 2.5-fold decrease on sst4 was measured.

Replacing Arg⁵ for Lys⁵ (KE131/Y versus KE88/Y) gave lower affinity values on sst1 and sst2 and no or little change on sst3, sst4, and sst5. A negative charge on the NH₂ terminus due to the metal complex (KE87/Y versus KE88/Y, DOTAGA has four carboxylic acid groups) led to an affinity decrease except on sst2 and sst3.

β-Ala acts as a spacer like the propionic acid arm of DOTAGA but maintains the neutrality of the NH₂-terminal metal complex (KE124/Y versus KE88/Y). This modification resulted in a 3-fold loss on sst1, an improvement on sst2, and little change on sst3, sst4, and sst5.

Agonistic properties. To test its agonistic properties, KE88/Y was studied for its effect on forskolin-stimulated cAMP accumulation in expressing cells using SRIF-28 as agonist control. Table 2 shows the similar EC₅₀ values of KE88/Y in comparison with SRIF-28 for sst1, sst2, and sst5 and a very similar potency for sst3 and sst4.

Internalization studies.Figure 1A and B illustrates the peptide-induced receptor internalization, analyzed by immunofluorescence microscopy. The clinically studied ⁹⁰Y^III-DOTA-TOC was used as a control along with KE88. Y^III-DOTA-TOC (IC₅₀ = 11.2 nmol/L) has good affinity to sst2 (18) and triggers sst2 internalization at 10 nmol/L, whereas KE88 induces sst2 internalization only at a concentration ≥1,000 nmol/L despite a 2.5-fold higher binding affinity. Y^III-DOTA-NOC has high affinity to sst2, sst3, and sst5 and was used as control peptide for sst3 (17). It triggers sst3 receptor internalization at ∼10 nmol/L, at the same concentration as KE88 (Fig. 1B) despite a 15-fold lower binding affinity.

Radioligand internalization was studied at 0.167 nmol/L concentration in AR4-2J, HEK-sst2, HEK-sst3, and HEK-sst5 cell lines with [¹¹¹In]KE88, [⁶⁷Ga]KE88, [¹¹¹In]KE87, and [¹¹¹In]KE131. Table 3 shows the 4-h internalization values of the radiopeptides in AR4-2J (sst2 positive), HEK-sst2, HEK-sst3, and HEK-sst5 cells. The uptake in the sst2-expressing cell lines is negligible under concentration conditions applied in molecular imaging in vivo. We did not study the binding on sst2-expressing cell lines as IC₅₀ values were determined on membranes, but as a “byproduct” of the internalization studies, we found practically 100% of the radioligand in the bound fraction. Conversely, a high internalization rate into HEK-sst3 is evident. [⁶⁷Ga]KE88 internalizes more efficiently than [¹¹¹In]KE88 (P < 0.05 at 2 and 4 h); there is no difference between [¹¹¹In]KE88 and [¹¹¹In]KE87, whereas [¹¹¹In]KE131 shows a somewhat lower rate of internalization. No internalization was observed into HEK-sst5.

Biodistribution studies.Table 4 summarizes the pharmacokinetics of [¹¹¹In]KE88 and [⁶⁷Ga]KE88 in a mouse xenograft tumor model carrying s.c. tumors in each flank, one exclusively expressing sst2 (HEK-sst2 or AR4-2J) and the other expressing sst3 (HEK-sst3). The most important finding of the biodistribution of [¹¹¹In]KE88 and [⁶⁷Ga]KE88 is the marked difference in the uptake kinetics of the two tumors. The HEK-sst3 tumor shows a high uptake of [¹¹¹In]KE88 peaking between 1 and 4 h after injection, the highest value amounting to ∼23% IA/g tumor at 1 h, whereas in HEK-sst2 the uptake was highest at 15 min (18.5 ± 0.65% IA/g); the washout from the HEK-sst2 tumor is fast, and at 24 h, only 1.14 ± 0.05% IA/g is left. Similarly, the uptake in the AR4-2J tumor is 8.5 ± 2.5% IA/g at 15 min followed by a fast washout amounting to only 1.2 ± 0.7% IA/g at 4 h after injection. In HEK-sst3, the uptake is long lasting being 23.2 ± 4.2% IA/g at 4 h and 14.9 ± 2.5% IA/g tumor at 24 h. The tumor uptake is specific and receptor mediated as was shown by coinjection of 50 μg KE88 and 50 μg In^III-DTPA-TATE (sst2-selective ligand) or by coinjection of DOTA-TATE. The sst3 tumor uptake was reduced at 1 h after injection by 84% and the sst2 tumor uptake by 89.5%. [⁶⁷Ga]KE88 was also specifically taken up by the sst3 tumor and to a significantly higher percentage at 4 h (30.1 ± 4.8% IA/g versus 23.2 ± 4.2% IA/g for [¹¹¹In]KE88; P < 0.05) but a somewhat faster washout with 15.3 ± 0.8% IA/g remaining at 24 h. In addition to the tumor, the pituitary shows specific uptake with a blocking efficiency of 60% at 1 h. The highest uptake of [¹¹¹In]KE88 was found in the kidney leading to low tumor to kidney ratios at all time points. For [⁶⁷Ga]KE88, this ratio is higher by a factor of 3 due to the higher tumor uptake of [⁶⁷Ga]KE88 but also to the significantly lower kidney uptake. The kidney uptake of [¹¹¹In]KE88 was efficiently lowered by >60% on coinjection of 10 mg lysine PBS solution. None of the other organs was influenced significantly by the lysine injection. The blood clearance of both radiopeptides was very fast with only 0.1% IA/g remaining in blood at 4 h. This results in excellent tumor (sst3 mainly) to blood ratios already at very early time points, such as 16.9 at 30 min, 32.7 at 1 h, 232 at 4 h, and 745 at 24 h for [¹¹¹In]KE88. Even for the HEK-sst2–expressing tumor, the tumor to blood ratio was ∼20 at 1 h, 37 at 4 h, and 57 at 24 h. A similar high ratio was found for [⁶⁷Ga]KE88.

Interestingly, the washout from all normal organs except from the kidneys is very fast as well. These pharmacokinetic data are reflected in the planar images obtained with a γ-camera (Fig. 2).

In the past, radiopeptides have been established as an important class of targeting vectors for the imaging and targeted radionuclide therapy of human tumors. There was consensus that agonists need to be selected as radioligands because they usually internalize into tumor cells. This represents an important mechanism for the active transport and accumulation of radiolabeled peptide analogues into the tumor cell. This is of importance for imaging and localization but more so for therapeutic application when a long residence time is mandatory for the success of this therapy modality.

The aim of this study was (a) to develop new pansomatostatin ligands, (b) to understand some SARs, and (c) possibly to select a pharmacologically active peptide as a substitute for octreotide or lanreotide in the treatment of hypersecretory tumors. Of particular importance was the selection of the most promising peptides for radiometal labeling after peptide modification with a chelator.

In this study, according to the best of our knowledge, the first radiopeptides having a pansomatostatin binding profile are described. In addition, we show for the first time that SRIF receptor radiometalloligands can have high binding affinities and similar effects on receptor-mediated signaling but different effects on the triggering of receptor internalization.

Several important findings are emerging from these studies. (a) The new family of (metallo)cyclooctapeptides shows high potency, comparable with the endogenous SRIF-28, with regard to the binding affinity to sst1 to sst5. The peptides also seem to be full agonists on sst1 to sst5. (b) [¹¹¹In]KE88 (and KE108) shows high potency with regard to internalization into sst3-expressing cells at concentrations relevant to in vivo tumor targeting. The internalization is receptor mediated. (c) Conversely, [¹¹¹In]KE88 and KE108 (21) show inefficient or no internalization into sst2 and sst5. They trigger receptor internalization at ∼100-fold higher concentration than do currently established second-generation chelated and iodinated cyclohexapeptides (16, 23, 24). None of the (radio)peptides triggers sst5 internalization, a property that was also found with other potent agonists, whereas the endogenous ligands SRIF-14 and SRIF-28 induced sst5 receptor internalization (21). (d) The inefficient sst2 internalization of these radiopeptides is contrary to what we (16, 17, 25) and others have found (26–30) using octreotide-derived radiopeptides promoting receptor and radioligand internalization at “physiologic” receptor-subsaturating concentrations. It therefore seems that some SRIF analogues may be able to dissociate between their sst2 signaling properties and their internalizing ability. (e) As a consequence of the internalization properties, biodistribution and imaging with [¹¹¹In]KE88, using a double tumor model, show the importance of efficient agonist internalization as an active uptake mechanism for prolonged tumor uptake.

With regard to the pansomatostatin ligand development, we encountered the fortunate situation that our lead peptide HR2215 (13) already had moderate to high affinity to all SRIF receptor subtypes. HR2215 was developed more than 20 years ago as a potential drug for the treatment of diabetes type 2. The replacement of Asn⁵ in HR2215 (SRIF-14 numbering) by Arg (HR2252) resulted in a distinct improvement on all receptor subtypes. We conclude that the positive charge on aa⁵ is important and responsible for the increased affinities. When replacing Arg⁵ for Lys⁵, there is a significant loss in sst2 and sst5 affinity but little difference in affinity to sst1, sst3, and sst4, again underlining the importance of a positive charge in this position. The substitution of Phe⁷ by Tyr⁷ (KE109) resulted in a drop of a factor of 7.7 in sst1 affinity (versus SRIF-28) but less than a factor of 2 in sst2 affinity, whereas the affinity for sst3, sst4, and sst5 is equal or even better for KE109. This substitution will allow iodination within the structure of the octapeptide and increases the hydrophilicity of the peptide.

The internalization data correspond to the in vivo pharmacokinetic data in the animal models. Whereas [¹¹¹In]KE88 has a high and persistent uptake in the sst3 tumors, the uptake in the sst2 tumors was lower and a fast washout was observed, resulting in a very low tumor targeting as early as 4 h after injection. These biodistribution data consequently translate into the imaging results where a distinct and long-lasting localization of the sst3 tumor can be seen but the sst2 tumors can only be faintly visualized up to 60 min. Remarkably, at 4 h, only the sst3 tumor and the kidneys are clearly visible, a consequence of the fast washout of the tracer from all other organs.

The higher uptake of [⁶⁷Ga]KE88 in the sst3 tumor at 4 h is most likely due to the faster rate of internalization of [⁶⁷Ga]KE88 as measured in vitro, a feature we have also found on sst2-expressing cells using DOTA-octreotide–based radiopeptides (31).

The kidney uptake of [¹¹¹In]KE88 is very high at all time points, probably due to the 2-fold positive charge of the peptide. This hypothesis is corroborated by the finding that the uptake can be quite efficiently blocked by coinjection of lysine, a basic amino acid, which reduces kidney uptake by >60%. Also an interesting finding is the good blocking of the kidney by excess of cold peptide, possibly again a consequence of the 2-fold positive peptide charge as there is no indication of sst expression in the mouse kidneys. These nonoptimized kidney blocking effects indicate that more effective kidney blocking is feasible in case of potential human applications.

The kidney uptake of [⁶⁷Ga]KE88 at 4 and 24 h is strikingly lower than the one of [¹¹¹In]KE88, a phenomenon that we have also seen in a variety of cyclic octapeptides coupled with DOTA (14, 31). We have explained the improved kidney uptake of radiogallium-labeled somatostatin-based octapeptides with the different coordination chemistry of the monoamide-conjugated Ga^III-DOTA complex. The hexacoordination of Ga^III affords a free carboxymethyl arm, which may alter the kidney handling of the peptide (14).

The results show that it is worthwhile to search for new radiolabeled pansomatostatin ligands for the targeting of tumors. Obviously, for this family of peptides, there is dissociation between strong binding, induction of signaling, and the ability to induce radioligand receptor endocytosis on sst2.

We do not know at this time how to explain the findings that the same high affinity radioligand internalizes into one sst subtype very efficiently but not into the other, at least not under conditions relevant to nuclear oncology applications. It is not clear which steric and/or electronic features are responsible for these somewhat surprising pharmacologic properties and the differences between the molecules studied in this work and the octreotide-based radiopeptides studied by us and others. One structural difference between the two different groups, the natural peptides SRIF-14 and SRIF-28, the octapeptides currently used in the clinic, and these peptides is the mode of cyclization. The new peptides are carbocyclic and not cyclized via the Cys-Cys disulfide bridge. Future studies may show if this structural feature and concomitant decrease in conformational flexibility is responsible for these pharmacologic properties.

These studies clearly show the importance of agonist-induced ligand internalization for targeted visualization of G protein–coupled receptor-positive tumors using radionuclide-coupled somatostatin analogues. The double tumor models show the fast washout of the pansomatostatin radioligand from tumor xenografts expressing the sst2 receptor, whereas high and persistent uptake was found in the sst3-expressing tumor, which obviously internalizes the radioligand also in vivo. The KE88-based radioligands may be interesting and suitable for imaging sst2-expressing tumors at early time points [e.g., with the metallic positron emitter ⁶⁸Ga (t_1/2 = 68 min; ref. 32)] and sst3-expressing tumors at later imaging time points using a longer-lived radionuclide [e.g., ⁶⁴Cu (t_1/2 = 12.7 h; ref. 23) and ⁸⁶Y (t_1/2 = 14.74 h; ref. 33)].

Obviously, agonistic properties are not sufficient for an adequate tumor targeting and a single thermodynamic constant is not sufficient to describe the efficacy of a ligand binding to a G protein–coupled receptor (34). Efficacy in terms of extended residence in a tumor targeted with an agonistic ligand for imaging and targeted therapy seems to depend on efficient internalization. These results do not contradict our recently reported findings on the use of radiolabeled somatostatin antagonists as new tools in nuclear oncology (20). As shown, antagonists recognize potentially many more receptor binding sites than agonists and rebinding after dissociation of the radioantagonist may be the efficient mechanism explaining a slow tumor washout. At the moment, peptidic antagonists are not available for every peptide receptor of interest and therefore need to be developed and studied with regard to tumor residence time. The present studies irrefutably underline the importance of internalization when using agonists for tumor targeting. In addition to the imaging and targeted radionuclide therapy aspects, some of the peptides compiled in Table 1 may be very good candidates for replacing octreotide or lanreotide in the treatment of symptoms of sst-overexpressing tumors.

We thank Dr. A. Schonbrunn (Houston, TX) for providing R2-88, Novartis Pharma for analytic assistance, B. Waser and S. Tschumi for their expert technical help, and the staff of Nuclear Medicine (University Hospital Basel) for technical assistance.