Glucagon-like Peptide-1 Receptor as Emerging Target: Will It Make It to the Clinic?

Radiotracers Based on Exendin-4

Exendin-4 was initially modified at the C terminus by the introduction of Lys⁴⁰, at the side chain of which the linker Ahx (aminohexanoic acid), followed by the chelator DTPA (diethylenetriaminepentaacetic acid), were conjugated. [Lys⁴⁰(Ahx-DTPA-¹¹¹In)NH₂]-exendin-4 showed remarkably high tumor uptake and tumor-to-background contrast in transgenic Rip1Tag2 mice that develop GLP-1R–expressing tumors in the pancreatic islets (7).

^99mTc-labeled exendin-4 using HYNIC (hydrazinonicotinamide) with EDDA (ethylenediaminetetraacetic acid) as coligand and ⁶⁸Ga-labeled exendin-4 using DOTA (1,4,7,10-tetraazacyclododecan-1,4,7,10-tetraacetic acid) followed. [Lys⁴⁰(Ahx-DOTA-⁶⁸Ga)NH₂]-exendin-4 had a biodistribution and tumor uptake similar to [Lys⁴⁰(Ahx-DOTA-¹¹¹In)NH₂]-exendin-4, whereas [Lys⁴⁰(Ahx-HYNIC-^99mTc/EDDA)NH₂]-exendin-4 had significantly lower tumor uptake, but as well lower uptake in the nontargeted organs (8). Consequently, image quality was high for all 3 radiotracers, which were then rapidly translated into clinical applications.

Newer developments addressed potential drawbacks, including radiochemical purity, specific activity (linked to the administered peptide mass), and the high and persistent kidney uptake.

Further Optimization and Specific Activity

Replacement of Met¹⁴, being susceptible to oxidation, by the nonoxidizing isosteric analog norleucine (Nle) resulted in [Nle¹⁴,Lys⁴⁰(Ahx-DOTA)NH₂]-exendin‐4 for ⁶⁸Ga and ¹¹¹In labeling with higher radiochemical purity (9,10).

Focusing on PET imaging, NODAGA (1,4,7-triazacyclononane, 1-glutaric acid-4,7 acetic acid) was used for improving radiochemical purity and specific activity, as it forms stable complexes with ⁶⁸Ga in milder conditions than DOTA. [Lys⁴⁰(NODAGA-⁶⁸Ga)NH₂]-exendin-4 was developed within a EU FP7 project (BetaCure), but no evidence of improvement versus the DOTA conjugate exists.

A variant of exendin-4 by introducing Cys⁴⁰ and conjugating NOTA (1,4,7-triazacyclononane-1,4,7-triacetic acid) functionalized as mono N-ethylmaleimide (MAL) was initially developed for ¹⁸F labeling via ¹⁸F-AlF (11). Nevertheless, the ⁶⁸Ga counterpart [Cys⁴⁰(MAL-NOTA-⁶⁸Ga)NH₂]-exendin-4 was the one translated into clinical practice. The other variant, ⁶⁸Ga‐DO3A-VS-Cys⁴⁰-exendin-4 (VS = vinyl sulfone and DO3A = 1,4,7-tris(carboxymethyl)-1,4,7,10-tetraazacyclododecane), showed improved specific activity versus the DOTA conjugate and the potential of GLP-1R quantification in the pancreas of rodents and nonhuman primates (12).

Antagonists and Reduction of Pancreatic Background Activity

The use of antagonists is attractive for avoiding activation of the GLP-1R signaling that induces glucose-dependent insulin secretion. Unfortunately, all radiolabeled variants of the antagonist exendin(9–39) were proven to be unsuitable, due to their low tumor uptake (13,14). Surprisingly, [¹²⁵I-BH-Lys²⁷]-exendin(9‐39)NH₂ (¹²⁵I-Bolton-Hunter in the residue Lys²⁷), was the only antagonist with a tumor uptake similar to the agonists, and in addition, significantly lower kidney uptake (14).

Low kidney uptake and subsequent low pancreatic background activity is an important prerequisite for the therapeutic use of radiolabeled GLP-1 analogs as well as the quantification of pancreatic β-cells. In fact, the high kidney uptake are characteristics of radiometal-labeled exendin-4 tracers, which are potentially addressed by ¹⁸F- or ^125/124I-labeled tracers (5). However, fluorinated or iodinated radiotracers are not easy in routine clinical production. Alternatively, other approaches were tested for kidney uptake reduction. Among them, the use of polyglutamic acid, gelofusine or albumin fragments and the introduction of cleavable linkers show significant reduction of the kidney uptake (5,6). Another option to circumvent the high kidney uptake, and consequently reduce the pancreatic background activity, is the use of positron emitters that allow late-time-point imaging, such as ⁶⁴Cu or ⁸⁹Zr.

Physiologic and Pathologic GLP-1R Expression in Humans

Only a single GLP-1R is known so far, which is structurally identical in all tissues (1). Tissue samples from 32 different tumors as well as normal tissue were screened for GLP-1R expression using in vitro receptor autoradiography with a GLP-1R–specific radiotracer, ¹²⁵I-GLP-1(7-36)NH₂ (15–18). The most striking GLP-1R expression with an almost 100% incidence and extremely high density was found in benign insulinomas (15), later also in vivo in insulinomas as part of the multiple endocrine neoplasia type 1 (MEN-1) syndrome, an autosomal dominant inherited tumor syndrome (19). No other peptide receptor, including somatostatin receptor (SSTR) subtype 2, exhibited such high expression levels in benign insulinomas (15). Table 2 summarizes the relevant GLP-1R expression in different tumors and normal tissue notably in the neurohypophysis, Brunner’s glands of the duodenum, pancreatic β-cells (2), and acini (Fig. 1). Importantly, the GLP-1R density in pancreatic β-cells is about 6 times lower than in insulinomas, resulting in good insulinoma-to-background contrast favorable for imaging (Fig. 1). Furthermore, there is an increased GLP-1R expression in pancreatic islet cell hypertrophy/hyperplasia also known as nesidioblastosis, which was first described in children and neonates and is characterized by endogenous hyperinsulinemic hypoglycemia (EHH; congenital hyperinsulinism). So far, the GLP-1R incidence and density in islet cell hypertrophy/hyperplasia was primarily quantified in adult nesidioblastosis after gastric bypass surgery (17), which is another clinical syndrome with mainly postprandial hypoglycemia, typically occurring after bariatric surgery (20).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

⁶⁸Ga-DOTA-exendin-4 PET and PET/CT of a patient with a histologically confirmed benign insulinoma. Whole-body maximum-intensity-projection PET (A) shows intense focal ⁶⁸Ga-DOTA-exendin-4 accumulation in a benign insulinoma (i) located in head of pancreas with good contrast compared with normal pancreas (p). Transaxial PET/CT indicates low ⁶⁸Ga-DOTA-exendin-4 uptake (B) in neurohypophysis (nh), moderate ⁶⁸Ga-DOTA-exendin-4 uptake (C) in normal pancreas (p), and high ⁶⁸Ga-DOTA-exendin-4 uptake (D) in Brunner’s glands of duodenum (b) and (E) in insulinoma (i).

Detection of Insulinomas with GLP-1R Imaging

In 2008, occult insulinomas were localized for the first time in humans using the metabolically stable ¹¹¹In-labeled GLP-1 agonist ¹¹¹In-DTPA-exendin-4 (28). ¹¹¹In-DTPA-exendin-4, ¹¹¹In-DOTA-exendin-4, and ^99mTc-HYNIC-exendin-4 scintigraphy and SPECT/CT were highly sensitive in the detection of benign insulinomas (29–31) but inferior to GLP-1R PET/CT with ⁶⁸Ga-DOTA-exendin-4, ⁶⁸Ga-NOTA-exendin-4, or ⁶⁸Ga-NODAGA-exendin-4, which showed sensitivities between 85% and 98% (22,24,32) (Tables 1 and 3). Importantly, ¹¹¹In- and ^99mTc-labeled exendin-4 can also be used for intraoperative localization of hidden insulinomas using a γ-probe (30). Prospective comparison of ⁶⁸Ga-DOTA-exendin-4/⁶⁸Ga-NOTA-exendin-4 PET/CT with ceMRI or ceCT in the same patients showed significant higher sensitivities for PET/CT in the detection of benign insulinomas (95%/98% vs. 69% or 74%) (22,24). Consequently, European Neuroendocrine Tumor Society guidelines recommend GLP-1R imaging or intraarterial calcium stimulation and venous sampling (ASVS) in patients with EHH with negative ceMRI/ceCT or negative endoscopic ultrasound (EUS) results (21). Methods such as ASVS and EUS show also high sensitivities (between 65% and 100%, Table 3) but are invasive and in the case of ASVS do not indicate the localization but the arterial supply of the insulinoma. A prospective comparison of GLP-1R PET/CT with different imaging modalities including EUS showed sensitivities of 98% for GLP-1R PET/CT and 85% for EUS, which was not significantly different (24).

Recently it was shown that ⁶⁸Ga-DOTA-exendin-4 PET/CT can specifically identify insulin-producing tumors (insulinomas) in the context of MEN-1 (19). This is clinically relevant because 80% of patients with MEN-1 develop repetitive, multiple pancreatic neuroendocrine tumors (NET), either functioning or nonfunctioning, which can become life-threatening due to hypoglycemia or malignant progression (19). Therefore, European Neuroendocrine Tumor Society recommends in patients with MEN-1 background a pancreas-preserving resection of clinically relevant NET that includes insulinomas independent of size and nonfunctioning NET with a diameter of ≥20 mm (21). In such patients, the detection of clinically relevant lesions is significantly higher with a combination of ⁶⁸Ga-DOTA-exendin-4 PET/CT plus MRI than just with MRI (92% vs. 39%), as MRI determines only size but not functionality of the lesion (19).

GLP-1R imaging is much less sensitive in the detection of malignant insulinomas than benign insulinomas because malignant insulinomas express the GLP-1R with an incidence of <40% (Table 2) (18). Furthermore, radiolabeled GLP-1 agonists show high accumulation in the kidneys (Fig. 1), restricting the therapeutic use of such compounds. Conversely, malignant insulinomas often express SSTR2 (incidence of 73%), which can be successfully targeted therapeutically (18) using SSTR peptide receptor radionuclide therapy (33).

Detection of Insulinomas with Alternative Nuclear Medicine Methods

Alternative nuclear medicine methods for localizing insulinomas include SSTR PET/CT, SSTR SPECT/CT, and ¹⁸F-fluorodopa (¹⁸F-DOPA) PET/CT (Table 3). Prasad et al. reported the sensitivity of SSTR PET to be 87% in 13 patients (33). However, preliminary data from a prospective head-to-head comparison in 29 patients (NCT03189953) indicate that the sensitivity of GLP-1R PET/CT is 85% compared with 55% for SSTR PET/CT (32). This is in concordance with in vitro autoradiography results of Reubi et al. who found that only 69% of benign insulinoma tissue samples expressed the SSTR2 (15).

SSTR SPECT/CT shows an insufficient sensitivity in the localization of benign insulinomas. It ranges between 20% and 29%, most probably due to the lower spatial resolution of this imaging modality (24,27).

¹⁸F-DOPA PET/CT can also be performed for localizing insulinoma, but data are scarce, retrospective, and controversial. Nakuz et al. detected the insulinoma in 5 of 10 patients (50%) with ¹⁸F-DOPA PET, whereas Imperiale et al. showed in 11 patients that detection rate of insulinomas with ¹⁸F-DOPA PET can be improved (73%) through serial scans (early plus late scan) and carbidopa pretreatment known to reduce physiologic uptake in the pancreas (34,35). A prospective direct comparison of GLP-1R PET/CT and ¹⁸F-DOPA PET/CT has not yet been performed.

Other Applications of GLP-1R Imaging

Another application for GLP-1R imaging is the detection of nesidioblastosis (hypertrophy/hyperplasia of pancreatic islets). This entity was first described in children and neonates and is a differential diagnosis of EHH. However, nesidioblastosis can also occur in adults (36). Clinically, the differential diagnosis between insulinoma and adult nesidioblastosis remains difficult, and the distinction between benign insulinoma, focal, or diffuse pancreatic nesidioblastosis is critical to suggest a treatment strategy. Although for childhood-onset nesidioblastosis ¹⁸F-DOPA PET/CT appears to be a valid tool (37), preliminary data in adult nesidioblastosis indicate that GLP-1R imaging might be an alternative imaging procedure differentiating between insulinoma, focal, or diffuse nesidioblastosis (38).

Another cause of EHH is postprandial hypoglycemia, occurring several years after bariatric bypass surgery (20). In a small subset of individuals, this form of hypoglycemia is severe and unresponsive to nutritional and medical management, and partial pancreatectomy has to be performed for symptom control (39). There is an ongoing controversy whether this syndrome is associated with nesidioblastosis (20,40). In vitro studies of surgical samples of patients with severe postbariatric hypoglycemia syndrome suggest that GLP-1R is not overexpressed in these islets compared with normal islets (Table 2).

β-cells represent only a tiny fraction of the volume of the pancreas (1%–3%) and are scattered in the islets throughout the whole organ. Changes in pancreatic β-cell mass (BCM) contribute to the development of type 1 and type 2 diabetes mellitus, and the in vivo assessment of BCM is critical in patients with type 1 and type 2 diabetes, since drug development aims at preserving BCM and preventing BCM death (41). Although the efficacy of such new drug compounds is relatively easy to determine in mice and rats using postmortem histology, it is challenging to translate these results to humans (41). GLP-1R imaging could be a promising method for BCM quantification. However, recent GLP-1R imaging studies revealed marked differences only in patients with established type 1 diabetes (absence or presence of BCM) (42), indicating difficulties in the quantification of small signal changes in early onset type 1 diabetes or type 2 diabetes (41). Such limitations may prevent the detection of submaximal changes of BCM or their use in regenerative β-cell–targeted therapies.

Pancreatic islets transplantation is an established method for patients with brittle type 1 diabetes after kidney transplantation (43). However, the live span of graft survival is limited due to inflammatory processes, which cannot be easily detected by serum markers. To monitor the graft volume, a noninvasive method to visualize viable β-cells is critical. GLP-1R imaging has been shown to reliably quantify engrafted β-cells using an intramuscular islet transplantation model in mice and in humans (44,45). This method has certainly potential, however, we need to keep in mind that skeletal muscle has no GLP-1R and, therefore, the BCM-to-background ratio is favorable. In contrast, background uptake of the liver, where the islets are usually engrafted, is somewhat higher (Fig. 1). Therefore, it remains to be demonstrated whether GLP-1R imaging in the context of transplanted islets holds its promise.

Pitfalls, Adverse Effects, and Limitations of GLP-1R Imaging

There are 2 groups of pitfalls/challenges that have to be considered while reading GLP-1R scans: peripancreatic uptake and missing focal uptake in the pancreas (46).

Peripancreatic uptake of radiolabeled exendin-4 is often seen in Brunner’s glands of the proximal duodenum. It is also the most common pitfall as Brunner’s glands are always physiologically present. Uptake in Brunner’s glands can be confused with an insulinoma or vice versa (46) (Fig. 1). Ectopic insulinomas, although extremely rare, can be another cause for peripancreatic uptake.

Biochemical evidence for an insulinoma but absence of focal radiolabeled exendin-4 uptake in the pancreas might occur in the following situations: size of insulinoma below camera detection limit, insulinoma overlap with kidney uptake, no GLP-1R expression (malignant insulinoma), insulinoma is not localized in the pancreas (ectopic insulinoma), and other cause of EHH (e.g., nesidioblastosis) (46).

Injection of radiolabeled exendin-4 can cause severe hypoglycemia in patients with EHH and nausea (30). Concomitant glucose infusion (e.g., 10%, 1,000 mL for 5 h) efficiently prevents hypoglycemia and is therefore recommended (22). Other adverse effects are nausea and sporadic vomiting. A low amount of radiolabeled exendin-4 (5–7 μg) may prevent nausea and vomiting (32).

The main limitation of GLP-1R imaging is its limited availability as it is not yet authorized for routine clinical use.