Non-invasive quantification of the beta cell mass by SPECT with 111In-labelled exendin
A reliable method for in vivo quantification of pancreatic beta cell mass (BCM) could lead to further insight into the pathophysiology of diabetes. The glucagon-like peptide 1 receptor, abundantly expressed on beta cells, may be a suitable target for imaging. We investigated the potential of radiotracer imaging with the GLP-1 analogue exendin labelled with indium-111 for determination of BCM in vivo in a rodent model of beta cell loss and in patients with type 1 diabetes and healthy individuals.
The targeting of 111In-labelled exendin was examined in a rat model of alloxan-induced beta cell loss. Rats were injected with 15 MBq 111In-labelled exendin and single photon emission computed tomography (SPECT) acquisition was performed 1 h post injection, followed by dissection, biodistribution and ex vivo autoradiography studies of pancreatic sections. BCM was determined by morphometric analysis after staining with an anti-insulin antibody. For clinical evaluation SPECT was acquired 4, 24 and 48 h after injection of 150 MBq 111In-labelled exendin in five patients with type 1 diabetes and five healthy individuals. The tracer uptake was determined by quantitative analysis of the SPECT images.
In rats, 111In-labelled exendin specifically targets the beta cells and pancreatic uptake is highly correlated with BCM. In humans, the pancreas was visible in SPECT images and the pancreatic uptake showed high interindividual variation with a substantially lower uptake in patients with type 1 diabetes.
These studies indicate that 111In-labelled exendin may be suitable for non-invasive quantification of BCM.
Trial registration ClinicalTrials.gov NCT01825148, EudraCT: 2012-000619-10
KeywordsBeta cell mass Exendin Glucagon-like peptide 1 receptor Single photon emission computed tomography
Percentage of injected dose per gram of tissue
Beta cell mass
Diethylene triamine pentaacetic acid
Glucagon-like peptide 1 receptor
Positron emission computed tomography
Single photon emission computed tomography
Vesicular monoamine transporter 2
Volume of interest
Diabetes mellitus is characterised by chronically elevated blood glucose levels. In type 1 diabetes, autoimmune destruction of pancreatic beta cells leads to absolute deficiency of insulin secretion with subsequent hyperglycaemia . It is commonly assumed that type 1 diabetes becomes clinically apparent when the remaining beta cell mass (BCM) has decreased to 5–20% [2, 3]. Recent studies, however, suggest that patients with type 1 diabetes may still have a large amount of remaining beta cells ; in some patients loss of BCM was less than 50% . Moreover, some patients newly diagnosed with type 1 diabetes who underwent autologous non-myeloablative haematopoietic stem cell transplantation became insulin-independent for several months, suggesting restored function of existing beta cells . In type 2 diabetes, failure of beta cells to adapt to the increased insulin demand, posed by insulin resistance in target tissues, leads to impaired glucose homeostasis [1, 7, 8, 9, 10]. BCM changes during the development and course of type 2 diabetes , but this process is neither well understood nor well characterised . Moreover, it also appears that BCM and beta cell function do not always correlate over the course of disease [9, 12]. To date, no reliable method exists to measure BCM in humans in vivo and most information has been obtained from autopsy studies of diabetic patients  and healthy individuals. Autopsies, however, cannot show the dynamics of BCM at onset and progression of the disease. Therefore, a non-invasive imaging method to quantify BCM in vivo would allow us to gain better insight into the pathophysiology of diabetes during the progression of the disease.
MRI and computed tomography (CT) offer the highest resolution of currently available clinical imaging technology, but neither resolution nor contrast suffices to visualise single pancreatic islets in humans in vivo. Alternative nuclear medicine imaging modalities, such as single photon emission computed tomography (SPECT) or positron emission tomography (PET), offer a very high sensitivity, which may be sufficient for the detection of the relatively low number of pancreatic islet cells by using a highly beta cell-specific radiolabelled tracer molecule. A potential target for measurement of the BCM is the glucagon-like peptide 1 receptor (GLP-1R) as it is abundantly expressed in rat, mouse and human pancreatic beta cells but not in α, δ and pp cells [13, 14]. Exendin is a stable agonist of the GLP-1R and it has been used for in vivo targeting of the GLP-1R after labelling with various radionuclides [15, 16, 17, 18, 19]. Exendin labelled with indium-111 could be used to determine the BCM in vivo by SPECT.
To test this hypothesis, we first explored biodistribution and specificity of the 111In-labelled exendin tracer in a rat model of alloxan-induced beta cell loss and subsequently in five patients with type 1 diabetes and five healthy individuals.
Peptides and radiolabelling for studies in rats
[Lys40(DTPA)]exendin-3 and [Lys40]exendin-3 were purchased from Peptides Specialty Laboratories (Heidelberg, Germany). Diethylene triamine pentaacetic acid (DTPA) was conjugated to the ε-amino group of the lysine (K40) and the C-terminal carboxyl group was amidated . Indium-111 labelling and quality control was performed as previously described . 111In-labelled exendin was injected into rats without further purification.
SPECT and biodistribution in diabetic rats
To determine the correlation of the uptake of 111In-labelled exendin-3 and the BCM, 6- to 8-week-old female Brown Norway rats (Harlan, Horst, the Netherlands) were injected with various concentrations of alloxan monohydrate (Sigma, St Louis, MO, USA): 15, 30, 45 or 60 mg/kg body weight (n = 4 per group; see electronic supplementary material [ESM] Methods for further details). Blood glucose concentrations were determined using a glucose meter (Accu-Chek Sensor; Roche Diagnostics, Almere, the Netherlands) 1 week after injection of alloxan. One week after the injection of alloxan, when hyperglycaemia was confirmed, rats received 15 MBq 111In-labelled exendin-3 intravenously; five control rats were co-injected with an excess of unlabelled exendin-3 (25 μg). One hour post injection (p.i.), SPECT/CT scans were acquired using a dedicated small-animal SPECT/CT scanner (NanoSPECT/CT; Bioscan, Paris, France). After SPECT imaging, the pancreas and other relevant tissues were dissected, weighed and counted in a well-type γ-counter (Wallac 1480-Wizard; Perkin-Elmer, Boston, MA, USA). Pancreases of one rat per group were harvested and snap-frozen in isopentane on dry ice for autoradiography. Cryosections (10 μm) were exposed to an imaging plate (Fuji Film BAS-SR 2025; Raytest, Straubenhardt, Germany) for 72 h. Images were acquired with a radioluminography laser imager (Fuji Film BAS 1800 II system; Raytest) and analysed with Aida Image Analyzer software (Raytest). The pancreases of the remaining rats were fixed in formalin and embedded in paraffin for the determination of the BCM as described below.
Pancreatic BCM was plotted against the pancreatic uptake determined by ex vivo counting.
111In-labelled exendin-3 uptake was quantified using Inveon Research Workplace (Preclinical Solutions, Siemens Medical Solutions USA, Knoxville, TN, USA). See ESM Methods for further details.
SPECT of dissected tissues
Brown Norway rats were injected intravenously with 60 mg/kg alloxan monohydrate as described above. When hyperglycaemia was confirmed, diabetic (n = 4) and healthy (n = 4) rats were injected with 15 MBq 111In-labelled exendin-3. Two control rats were co-injected with an excess of unlabelled exendin-3 (25 μg/rat). Rats were killed 1 h after injection of the radiolabelled exendin and the pancreas, gastrointestinal tract, liver and spleen were dissected and fixed in 4% formalin. SPECT images of the dissected pancreases, along with the gastrointestinal tract, liver and spleen, were acquired on a dedicated small-animal SPECT scanner (U-SPECT-II; MILabs, Utrecht, the Netherlands).
Determination of the BCM
Five sections (4 μm) of paraffin-embedded pancreases, 100 μm apart from each other, were stained with an anti-insulin antibody (4 μg/ml in PBS containing 1% BSA w/v) (sc 9168; Santa Cruz Biotechnology, Santa Cruz, CA, USA) as previously described .
The surface of the beta cells in the sections was estimated by drawing regions of interest around insulin-stained tissue and total pancreatic tissue on images of the pancreatic section captured using a digital image processing system for bright field microscopy  (400 × magnification). The total pancreatic pixel area was determined by automated image thresholding  of the hematoxylin-stained pancreatic tissue. The total insulin-positive pixel area was determined by drawing regions of interest over the insulin-positive tissue and the insulin-positive fraction was calculated by dividing the insulin-positive pixel area by the total pancreatic pixel area . The BCM was calculated by multiplying the insulin-positive fraction by the total pancreatic weight.
One Brown Norway rat was injected intravenously three times with 15 MBq 111In-labelled exendin-3 at intervals of 1 h and the pancreas was harvested 1 h after the last injection. The pancreas was fixed in formalin, embedded in paraffin and 4 μm sections were prepared. The sections were deparaffinised and covered with a homogeneous layer of hypercoat emulsion (Amersham LM-1; GE Healthcare, Little Chalfont, UK) by dipping the slides in the emulsion at 42°C in the dark. After incubation in the dark for 1 week the sections were developed for 4 min in a 6% black and white developer (Agfa studional liquid; Agfa, Leverkusen, Germany). The sections were rinsed in water for 20 s and fixed in 24% (wt/vol.) solution of sodium thiosulphate for 4 min. The sections were rinsed with water, stained with hematoxylin for 30 s, rinsed with water for 10 min, dehydrated and the slide was mounted with mounting fluid. Consecutive sections were stained with an anti-insulin monoclonal antibody as described above for co-localisation.
mRNA extraction and real-time PCR
Primers used for quantitative PCR
2 Sequence (5′–3′)
3 Base pairs
Preparation and labelling of 111In-labelled exendin for human use
[Lys40(Ahx-DTPA)]exendin-4 (where Ahx represents aminohexanoic acid) for clinical use was produced (piCHEM, Graz, Austria), dissolved and portioned according to GMP standards to a final concentration of 2 μg/ml in 1 mol/l HEPES (pH 5.5) containing 0.1% Tween 80. The vials were stored at −20°C. Radiolabelling was performed by adding exendin to the vial containing 111InCl3 to a final concentration of 150 MBq 111InCl3 per μg [Lys40(Ahx-DTPA)]exendin-4. After incubation at room temperature for 30 min 0.25 ml EDTA (3.75 mg/ml) was added and quality control was performed as described above. Finally, the volume was adjusted to 5 ml with 0.9% NaCl before administration.
Type 1 diabetic patients and healthy controls
No previous treatment with exenatide or dipeptidyl-peptidase IV inhibitors
Creatinine clearance (MDRD), ml/min
Type 1 diabetic patients
No stimulated insulin production: C-peptide, nmol/l
Not measurable (< 0.03)
Time since diagnosis of type 1 diabetes, years
Normal glucose tolerance: blood glucose 2 h after OGTT, mmol/l
Overview of included participants
Type 1 diabetes duration, years
HbA1c, % (mmol/mol)
CT-based pancreas volume, cm3
Uptake in the whole pancreas, counts
Patients with type 1 diabetes
Administration of radiotracer
After participants had fasted for at least 4 h, 150 MBq 111In-labelled exendin, corresponding to a peptide dose of 1 μg exendin, was injected as a slow bolus over 5 min.
At 4 h, 24 h and 48 h after 111In-labelled exendin administration, SPECT was acquired using an integrated SPECT/CT scanner (Symbia T16; Siemens Healthcare, Molecular Imaging, Hoffman Estates, IL, USA) and CT images were acquired 24 h post injection. See ESM Methods for further details.
SPECT images were reconstructed with Hermes Hybrid Recon software (Version 1.0C, Hermes Medical Solutions, Stockholm, Sweden), in six iterations, 16 subsets and a matrix size of 128 × 128 (corresponding with an isotropic voxel size of 4.7952 mm). CT-based attenuation correction and Monte Carlo-based scatter correction was performed. Post-reconstruction filtering was applied with a 1.0 cm Gauss filter.
Quantification of SPECT images
Image quantification was performed with Inveon Research Workplace software (Version 4.1, Siemens, Munich, Germany). The CT image was used for delineation of the pancreas. Within the pancreas two spherical volumes of interest (VOIs), one in the head and one in the corpus, were used for determination of the pancreatic uptake. See ESM Methods for further details.
All mean values are expressed as mean ± SD. Statistical analysis was performed using the unpaired two-tailed t test using GraphPad Prism (version 4, GraphPad Software, San Diego, CA, USA). The level of significance was set at p < 0.05.
Correlation between the BCM and pancreatic uptake was determined by the Pearson correlation coefficient (r) using two-tailed analysis of variance with GraphPad Prism (version 4). The level of significance was set at p < 0.05.
A specific activity of 700 GBq/μmol for 111In-labelled exendin-3 was obtained with a radiochemical purity exceeding 95%.
The uptake of 111In-exendin in the pancreas was plotted against the BCM (Fig. 1f) showing a good correlation between these two variables (Pearson r = 0.89).
Ex vivo autoradiography
Autoradiography of pancreatic sections of alloxan-treated and control rats demonstrated that the tracer specifically accumulated in the islets of Langerhans: high concentrations of 111In-exendin were observed in hotspots throughout the pancreatic section in control rats, representing the islets of Langerhans (Fig. 1b, c). The uptake was GLP-1R-mediated since it could be blocked by an excess of unlabelled exendin (Fig. 1b). Very low uptake of the tracer in the exocrine pancreas was observed (Fig. 1b, c). When rats were treated with alloxan, the tracer uptake in the islets was reduced (Fig. 1b).
Immunohistochemical staining of the pancreatic sections with an anti-insulin monoclonal antibody (mAb) showed co-localisation of the stained islets and the hotspots observed in autoradiography (Fig. 1c), demonstrating specific accumulation in the islets. Microautoradiography (Fig. 1d) also showed islet-specific tracer uptake, which was co-localised with immunohistochemical staining of a consecutive pancreas section with an anti-insulin mAb (Fig. 1e).
SPECT of dissected tissues
SPECT and quantitative analysis of SPECT images of diabetic rats
The SPECT images of dissected tissues enabled accurate interpretation of the SPECT images of Brown Norway rats (Fig. 2g). After injection of 111In-labelled exendin into healthy Brown Norway rats, uptake in the pancreas was visible in SPECT scans.
In diabetic rats there was negligible uptake in the pancreatic region and lower uptake was observed in rats with reduced BCM. Besides high uptake in the pancreas, uptake was also observed in the kidneys and lungs. Moreover, SPECT images after injection of 99mTc-labelled demobesin 3, specifically accumulating in the pancreas , correlated with the pancreatic uptake of 111In-labelled exendin (ESM Fig. 1), confirming that the VOI used for quantification of BCM is indeed located in the pancreas. As previously described, the pancreas is the organ with the highest uptake of the bombesin analogue in the peritoneal cavity of rats , due to efficient binding to the gastrin-releasing peptide receptor, which is abundantly expressed in the pancreas [29, 30]. Removal of both kidneys enabled more accurate visualisation of the pancreas, showing exact co-localisation of 99mTc-labelled demobesin and 111In-labelled exendin as an additional proof that the VOI used for quantification contained the pancreas (ESM Fig. 1). The tracer uptake of 111In-labelled exendin, as determined by quantitative analysis of the SPECT images, correlates linearly with the BCM determined by morphometric analysis (Fig. 2h, Pearson r = 0.83).
SPECT/CT in humans
In this study, we systematically characterised 111In-labelled exendin in rats for the non-invasive determination of pancreatic BCM by SPECT. The key findings of these studies are: (1) There is specific GLP-1R-mediated uptake of 111In-labelled exendin in the pancreas; (2) 111In-labelled exendin localises specifically in the islets of Langerhans in rats; (3) the pancreatic uptake of 111In-labelled exendin determined by ex vivo counting correlates linearly with BCM (Pearson r = 0.89); (4) rats treated with high doses of alloxan, representing a controlled model of severe beta cell loss, showed an 80% reduction of tracer uptake in the pancreas indicating that even small differences in BCM can be detected; (5) the pancreatic tracer uptake decreased with decreasing BCM and only minimal uptake was observed in rats treated with high doses of alloxan; (6) most importantly, the uptake of tracer by the pancreas could be visualised by in vivo SPECT and quantitative analysis of the SPECT data correlated with the BCM in a linear manner (Pearson r = 0.83).
We have previously shown that the peptide dose is a critical factor that determines the uptake of 111In-labelled exendin in GLP-1R-positive tissues . It should be noted that the specific and high uptake of the tracer in the pancreas in this rat model is only observed at peptide doses not exceeding 0.1 μg per rat; at higher doses the GLP-1Rs in the pancreas are saturated as shown in the in vivo blocking experiments. Therefore, a tracer with a very high specific activity is required for SPECT . We have therefore improved the specific activity of 111In-labelled exendin up to 700 MBq/nmol, which allows administration of sufficient radioactivity for SPECT without exceeding the tracer dose of 0.1 μg/rat.
Previous preclinical studies with exendin labelled with 18F or 64Cu have demonstrated their ability to visualise transplanted islets and subcutaneous tumours in rodents [18, 19]. For the GLP-1R antagonist exendin-(9–39) labelled with 18F, however, no correlation between uptake and pancreatic BCM could be demonstrated in a rat model . This is probably due to the fact that the antagonist exendin-(9–39) is not suitable for in vivo targeting of the GLP-1R , indicating that agonistic activity is required for successful visualisation of beta cells in vivo. Importantly, Reiner et al have recently demonstrated the in vivo targeting of beta cells with a fluorescent labelled exendin-4 analogue .
Previous studies suggested that GLP-1R expression is detected in the exocrine pancreas by in vitro autoradiography with 125I-GLP-1 and immunohistochemistry on human pancreatic sections [13, 33]. Our ex vivo autoradiographic analysis showed scattered focal hotspots throughout the pancreas that were co-localised with the islets in immunohistochemical staining with only low background activity in the exocrine pancreas. This focal uptake decreased with alloxan-induced loss of BCM and was almost completely absent in severely diabetic rats. The results of our ex vivo autoradiography are in line with the quantitative PCR analysis on rat tissues: high GLP-1R expression was observed in the endocrine pancreas, while low expression was apparent in the exocrine pancreas. The differences observed between our results and the previously published data [13, 33] may be explained by differences between binding characteristics of 125I-labelled glucagon-like peptide 1 and GLP-1R antibodies in an in vitro assay as compared with the binding characteristics of 111In-labelled exendin in vivo. Of note, in vivo 111In-labelled exendin also shows internalisation and metabolic trapping , which makes this technology highly sensitive and specific for the detection of GLP-1R-positive tissues. In addition, artefacts potentially occurring in vitro leading to non-specific binding are avoided.
It was previously reported that the GLP-1R expression is decreased in pancreatic islets in hyperglycaemic rats after near total pancreatectomy or constant glucose infusion , as determined by immunohistochemical staining and quantitative PCR. Our data show a strong correlation of tracer uptake and BCM under euglycaemic as well as severe hyperglycaemic conditions. This observation argues against changes in GLP-1R expression having a significant influence on 111In-labelled exendin uptake. In addition, quantitative RT-PCR showed that the endocrine–exocrine GLP-1R expression ratio is even higher in humans than in rats (endocrine–exocrine ratio 47 in humans vs 13 in rats). Of note, we cannot exclude some contamination of the exocrine pancreas with insulin-producing cells, so that the actual endocrine–exocrine ratio of GLP1R mRNA expression may be even higher.
After these promising preclinical findings, we explored the feasibility of this technology to determine the BCM in vivo in patients with type 1 diabetes (n = 5) and matching healthy controls (n = 5). The pancreatic uptake of 111In-labelled exendin showed considerable interindividual variation, in line with variations in BCM as reported in the literature [12, 36]. The pancreatic uptake of 111In-labelled exendin was clearly decreased in patients with type 1 diabetes, with marked differences between diabetic patients and healthy individuals. To date, no other radiotracer tested for BCM determination has shown such large maximum differences between healthy individuals and patients with type 1 diabetes, nor has such a high interindividual variability of the uptake been reported previously [37, 38, 39, 40, 41, 42, 43, 44]. A recent study in healthy individuals and patients with type 1 diabetes, involving PET after injection of [18F]fluoropropyl-dihydrotetrabenazine ([18F]FP-(+)-DTBZ), showed a maximum difference in pancreatic uptake of <40% . The difference in binding potential between patients with type 1 diabetes and healthy individuals was not greater than 50% . Moreover, a recent publication showed that expression of vesicular monoamine transporter 2 (VMAT2) in pancreatic polypeptide cells causes overestimation of the BCM in type 1 diabetes, as determined by PET measurements . In this study, the authors demonstrated that PP cells express VMAT2 and this contributes to the total pancreatic uptake of [18F]FP-(+)-DTBZ. These data are in line with the reduction in pancreatic uptake of [18F]FP-(+)-DTBZ by less than 40% in patients with type 1 diabetes. The markedly higher difference in pancreatic uptake of 111In-labelled exendin between healthy individuals and patients with type 1 diabetes points towards a considerably higher specificity of 111In-labelled exendin for beta cells.
Of note, the dynamic scanning protocol previously described, including arterial blood sampling, makes this a laborious procedure. Our SPECT protocol with 111In-labelled exendin is a much simpler method that requires a single injection of the radiotracer followed by single time point static SPECT acquisition and subsequent quantitative analysis of the images. In addition, the method can be used in every hospital equipped with a gamma camera, does not require a cyclotron and the labelled compound can easily be distributed ready to use for clinical studies.
In summary, we here demonstrate in a preclinical imaging model of beta cell loss that the uptake of 111In-exendin, as determined by in vivo imaging, is specific for the beta cells and shows a direct correlation with BCM. In humans, we observed considerable interindividual differences in pancreatic tracer uptake (in line with available autopsy data) and a marked reduction in pancreatic uptake in patients with type 1 diabetes. To the best of our knowledge, so far there is no comparable preclinical evidence for any other radiotracer tested for in vivo imaging of BCM with respect to beta cell specificity and correlation of uptake to actual BCM. Of note, clinical imaging does not require complicated dynamic scanning techniques or calculation of binding capacity, making it feasible for efficient use in large clinical trials . Therefore, our data indicate that this technology could indeed be the first to enable non-invasive determination of BCM in humans, offering great potential to better elucidate the pathophysiology of the different types and stages of diabetes.
We thank T. Maina and B. Nock (National Centre for Scientific Research ‘Demokritos’, Athens, Greece) for kindly providing demobesin-3 and M. Melis, J. de Swart and E. de Blois (Department of Nuclear Medicine, Erasmus Medical Centre, Rotterdam, the Netherlands) for technical support of the SPECT experiments. H. de Wit (Department of Internal Medicine, Radboud university medical center, Nijmegen, the Netherlands) assisted in the recruitment of patients. We acknowledge W. Grootjans (Department of Radiology and Nuclear Medicine, Radboud university medical center, Nijmegen, the Netherlands) for technical assistance in image analysis of the patient data.
Our work was supported by NIH grant 1R01 AG 030328-01 and the European Community’s Seventh Framework Programme (FP7/2007-2013) project BetaImage, under grant agreement No. 222980.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.
MB and WW designed the experiments and acquired, analysed and interpreted the data and drafted and critically revised the manuscript. PR, LJ, CF, KA, TL, MJ, OCB, MG and TB provided a substantial contribution to the conception and design, acquisition of data, or analysis and interpretation of data. BJG, MdJ, DLE, MBe, WO and CJT contributed to analysis and interpretation of the data. LJ, CF, KA, CJT, MBé, MdJ, BJG, DLE, WJG, TB, PR, TL and MJ critically revised the manuscript for important intellectual content. OCB and MG provided a substantial contribution to the conception and design, acquisition of data, or analysis and interpretation of data and critically revised the manuscript for important intellectual content. All authors approved the final version of the manuscript. MG is responsible for the integrity of the work as a whole.
- 1.American Diabetes Association (2006) Diagnosis and classification of diabetes mellitus. Diabetes Care 29(Suppl 1):S43–S48Google Scholar
- 7.DeFronzo RA (1997) Pathogenesis of type 2 diabetes: metabolic and molecular implications for identifying diabetes genes. Diabetes Rev 5:177–269Google Scholar
- 40.de Lonlay P, Simon-Carre A, Ribeiro MJ et al (2006) Congenital hyperinsulinism: pancreatic [18F]fluoro-L-dihydroxyphenylalanine (DOPA) positron emission tomography and immunohistochemistry study of DOPA decarboxylase and insulin secretion. J Clin Endocrinol Metab 91:933–940PubMedCrossRefGoogle Scholar