Validation of ¹¹¹In-Exendin SPECT for the Determination of the β-Cell Mass in BioBreeding Diabetes-Prone Rats

Type 1 diabetes (T1D) is the result of autoimmune destruction of the insulin-producing β-cells in the islets of Langerhans in the pancreas. It was long believed that a homogeneous and near-complete destruction of the β-cells occurs, but it has recently been shown that the autoimmune destruction of β-cells is a heterogeneous and chronically ongoing process during the course of T1D (1–3). As a result, β-cells and insulitic regions were found in pancreatic specimens many years after diagnosis of T1D (2). In the pancreata of T1D patients, various degrees of β-cell destruction in different areas of the pancreas could be found, including simultaneous detection of apparently healthy islets, early abnormalities, insulitic regions, and islets completely depleted from β-cells and absence of immune cells (3). More importantly, these studies showed that the degree and pattern of β-cell destruction and immune cell infiltration are unique in every individual. The simultaneous appearance of different states of disease within one pancreas and the highly individual course of T1D would require a method to noninvasively measure the β-cell mass (BCM) to provide additional information about the number and intrapancreatric distribution of β-cells during T1D and for the selection of patients eligible for novel (immune regulatory based) treatments.

We previously showed that radiolabeled exendin accumulates specifically in the β-cells and could potentially be used to measure BCM by single-photon emission computed tomography (SPECT) in rodents and humans (4). Despite these promising data, questions remain about the effectiveness of radiolabeled exendin for the determination of the BCM in the complex pathogenesis of diabetes. A concern is the potential variation in the expression of the glucagon-like peptide 1 receptor (GLP-1R), the receptor targeted by radiolabeled exendin, during the course of diabetes including the inflammation-induced physiological changes in the pancreas. For example, it was shown that cytokine treatment of cell sorter–purified rat β-cells reduced the GLP-1R mRNA expression (5). Furthermore, hyperglycemia results in lower GLP-1R expression in isolated rat islets in vitro and in a rat model of partial duct ligation (6). These results suggest that the GLP-1R expression might change during the progression of T1D, which might result in changes of radiolabeled exendin accumulation impairing the correlation between tracer uptake and BCM. Besides the effects of inflammation and metabolic disease on the GLP-1R expression, changes in blood flow and enhanced vascular permeability due to inflammation could also affect accumulation of the tracer in the pancreas. However, the effect of potential changes in receptor expression on the accumulation of radiolabeled exendin in the β-cells is difficult to predict. We have previously demonstrated the linear correlation between radiolabeled exendin uptake and BCM in an alloxan-induced diabetes model of β-cell reduction (4,7), but this model does not show the typical patterns of inflammation/insulitis found in T1D. Therefore, for determination of the feasibility of accurately measuring the BCM during T1D, the use of the tracer to determine the BCM noninvasively should be tested in a rodent model mimicking all aspects of the human disease progression.

In the current study, we examine the correlation between the pancreatic uptake of ¹¹¹In-diethylenetriaminepentaacetic acid–exendin-3 (¹¹¹In-exendin) and the BCM and the influence of insulitis on the tracer uptake in a rat model for spontaneous T1D: the BioBreeding diabetes-prone (BBDP) rat. These rats are characterized by severe insulitis, followed by rapid onset of T1D, making this an ideal model to study the effects of insulitis and inflammation and short-term fluctuations in blood glucose levels on the uptake of ¹¹¹In-exendin in the pancreas.

Forty-five female BBDP rats 3–5 weeks old were obtained from Biomedical Research Models (Worcester, MA). The rats were housed in individually ventilated cages and were fed with sterilized chow and drinking water for maintenance of their microbiological status to ensure diabetes development. After 1 week of acclimatization in our facility, unilateral nephrectomy (left kidney) was performed, to allow accurate estimation of tracer uptake in the pancreas, under inhalation anesthesia with isoflurane (Abbott Laboratories, Toronto, Canada) in O₂:air (1:4). Carprofen (Rimadyl; Pfizer Animal Health B.V., Capelle aan de IJssel, the Netherlands) was used as analgesia for 2 days after surgery twice daily (5 mg/kg). As of 1 week after surgery, the rats were used for SPECT studies. Every week three rats were subjected to SPECT studies.

Blood glucose levels and body weight were monitored at least twice weekly, and the frequency of monitoring was intensified when blood glucose levels rose or body weight dropped. For blood glucose level measurements, a drop of blood was drawn by puncturing the tail vein with a 25-gauge (25G) needle (BD Biosciences, Breda, the Netherlands). Blood glucose was measured with a glucose meter (Accu-Chek Sensor; Roche Diagnostics, Almere, the Netherlands). When rats were hyperglycemic, insulin was administered via a sustained-release insulin implant (Linshin Canada Inc., Toronto, Canada) with a 12G trocar needle to control blood glucose levels.

¹¹¹In-exendin was prepared as previously described (8). Rats were injected intravenously with 15 MBq ¹¹¹In-exendin, and SPECT scans were acquired 1 h after injection on a dedicated small-animal SPECT/computed tomography scanner (U-SPECT-II; MILabs, Utrecht, the Netherlands). SPECT images were acquired with a general-purpose 1.0-mm pinhole rat and mouse collimator with an acquisition time of 50 min. The images were reconstructed with an ordered subset expectation maximization algorithm (three iterations, eight subsets, voxel size 0.75) using the U-SPECT-Rec software (MILabs). Mean uptake of ¹¹¹In-exendin in the pancreas was quantified using Inveon Research Workplace (Preclinical Solutions, Siemens Medical Solutions, Knoxville, TN). A volume of interest was drawn over the pancreas, and mean uptake (per voxel) was corrected for the injected dose.

After SPECT/computed tomography acquisition, the pancreas was dissected, fixed in formalin, and embedded in paraffin for autoradiography and histological evaluation.

Sections (4 µm) were prepared of the paraffin-embedded pancreata at three levels with an interlevel distance of 100 µm. The sections were mounted and dried, and one section per level was exposed to a phosphor imager screen (Fuji Film BAS-SR 2025; raytest, Straubenhardt, Germany) for 1 week. Images were acquired with a radioluminography laser imager (Fuji Film BAS 1800 II system; raytest) and analyzed with Aida Image Analyzer software (raytest).

A consecutive section of the section used for autoradiography was stained for insulin as follows: Deparaffination and antigen retrieval were performed as previously described (4). Blocking was performed by incubation with PBS containing 1% BSA for 30 min at room temperature. After removal of PBS-BSA, 50–150 µL rabbit anti-insulin antibody (4 µg/mL diluted in PBS containing 1% BSA, category no. sc9168; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) was added and incubated for 1 h at room temperature in the dark. The sections were washed three times with PBS, and the sections were incubated with Alexa Fluor 594 donkey anti-rabbit IgG (10 µg/mL; Invitrogen, Bleiswijk, the Netherlands) for 30–60 min at room temperature in the dark. The sections were counterstained with DAPI. After washing three times with PBS, the sections were mounted with Fluoromount (Sigma-Aldrich, St. Louis, MO).

Insulin-stained sections were scanned with an automated microscope (Zeiss Axioskop 2 Plus) using a 100× magnification and a green light filter set. Thresholding was performed on pancreatic tissue to determine total pancreatic area semi-automatically using ImageJ (version 1.47; National Institutes of Health). Also, a threshold was set for the fluorescently stained β-cells to discriminate between the endocrine and exocrine tissue. Next, insulin-positive β-cells were delineated in iVision (iVision-Mac [4.0.16]; BioVision Technologies) to select only islets and delete autofluorescent regions. Finally, ImageJ was used to calculate the percentage of insulin-positive β-cells within the pancreas automatically.

One section per level was stained with hematoxylin-eosin (three sections per rat). All islets in the sections were counted and scored for the presence of insulitis as follows: 0, no inflammation; 1, intraislet inflammation occupying <50% within the islet; 2, intraislet inflammation occupying >50% within the islet; and 3, complete or near-complete obliteration of islets by inflammation. The sections were read blindly by two independent observers. The index of insulitis of each rat was calculated according the following formula: I = (0 × n₀) + (1 × n₁) + (2 × n₂) + (3 × n₃)/3 × (n₀ + n₁ + n₂ + n₃) (9).

The correlation coefficient between the BCM and pancreatic uptake of ¹¹¹In-exendin was calculated with a Pearson correlation coefficient with SPSS (IBM SPSS Statistics 22). The influence of insulitis on this correlation was calculated by using insulitis as a covariant. The statistical difference between these correlations was examined by a two-tailed Fisher r-to-z transformation. The level of significance was set to 0.05.

The results of the blood glucose measurements of BBDP rats during the complete experiments are summarized in Fig. 1. All rats were normoglycemic until 8 weeks of age, and hyperglycemia occurred from 9 weeks of age. Hyperglycemic rats were treated with insulin to control the blood glucose levels. Despite insulin treatment, short-term fluctuations in blood glucose levels were observed.

In animals with a high β-cell fraction, ¹¹¹In-exendin accumulated in the pancreas, as measured by ex vivo counting. The pancreatic ¹¹¹In-exendin decreased in rats with declining BCM leading to a linear correlation between β-cell fraction and ¹¹¹In-exendin accumulation in the pancreas (Fig. 2) (Pearson r = 0.89, P < 0.0001). The glycemia levels at the time point of imaging are presented in Fig. 2B. The correlation between the BCM mass and the pancreatic uptake, corrected for the blood glucose levels, was Pearson r = 0.89 and was not significantly different than the uncorrected correlation (P < 0.05).

A restricted range correlation coefficient, excluding the animals with a BCM <0.4%, was calculated, resulting in a Pearson r = 0.56.

Insulitis scored by two independent observers showed a high level of agreement as determined by a Bland-Altman plot (Supplementary Fig. 1). There was no statistical correlation between the difference between the observers and the mean of the observers (one-sample t test, P = 0.292). Therefore, there is no level of proportional bias between both observers and the mean insulitis score of both observers was used for further analysis. In young BBDP rats, no infiltration of the islets was observed and first signs of insulitis were observed after 7 weeks of age (Fig. 3), preceding the onset of hyperglycemia. The correlation between pancreatic ¹¹¹In-exendin accumulation and the β-cell fraction, corrected for the insulitis score, was Pearson r = 0.81 (partial correlation coefficient) and was not statistically different than the correlation coefficient when not corrected for insulitis (P = 0.18).

High accumulation of ¹¹¹In-exendin was observed in the islets of Langerhans in the autoradiographic images of rats with high BCM (Fig. 4A). Very low accumulation was observed in the exocrine pancreas, and the accumulation of ¹¹¹In-exendin colocalized with insulin expression (Fig. 4D). The ¹¹¹In-exendin uptake decreased with declining amounts of β-cells (Fig. 4B) and was not detectable in slides without any β-cells (Fig. 4C), which was in line with the decline in insulin-positive cells visualized by histology (Fig. 4E and F). Furthermore, β-cell identity was verified by Nkx6.1 staining (Fig. 5).

SPECT images of rats with a high BCM showed a clear ¹¹¹In-exendin accumulation in the pancreas (Fig. 6A) (β-cell fraction 0.6%). Besides ¹¹¹In-exendin accumulation in the pancreas, a high signal was observed in the kidney. The signal in the pancreas decreased in rats with declining BCM (Fig. 6B and C). The pancreatic uptake of ¹¹¹In-exendin, determined by quantitative analysis of the SPECT images, correlated linearly with the ¹¹¹In-exendin accumulation as determined by ex vivo counting (Fig. 7) (Pearson r = 0.64, P < 0.0001). The pancreatic uptake of ¹¹¹In-exendin determined by SPECT analysis showed a linear correlation with the β-cell fraction (Pearson r = 0.50, P = 0.0006).

In this study, we validated ¹¹¹In-exendin as a method to determine the BCM noninvasively in BBDP rats, a model for spontaneous T1D. We demonstrated a strong correlation between pancreatic ¹¹¹In-exendin accumulation, as determined by ex vivo counting, and the BCM in this T1D rat model. Moreover, autoradiography revealed that the accumulation is restricted to the islets with a very low signal in the exocrine pancreas. Importantly, the signal from the islets completely disappeared when no β-cells were present, while α-cells could still be observed histologically. These observations indicate high specificity of ¹¹¹In-exendin for β-cells and are in line with our previous studies in a rat model for chemically induced β-cell loss, where a strong correlation between pancreatic exendin accumulation and BCM, but not α-cell mass, was found (4,7).

The expression of the GLP-1R in the presence of an immune response remains a matter of debate. Previous studies showed a decrease in GLP-1R expression in islets treated with cytokines in vitro (5) or after partial duct ligation in vivo (6). Our data clearly demonstrate that in the presence of insulitis also there is a strong linear correlation between the pancreatic uptake of the tracer and the BCM, comparable with the correlation found in chemically induced diabetes without insulitis (Pearson r = 0.89 uncorrected vs. Pearson r = 0.81 corrected for insulitis, P = 0.18). This nonsignificant difference indicates that insulitis has no relevant effect on the correlation between these parameters. The use of the sophisticated rat model for T1D, representing a complete biological system, should reflect the complex pathophysiological processes better than a rather crude in vitro model of cytokine-treated β-cells. It should be noted that in humans, insulitis is described by the infiltration of three or more islets with >15 leukocytes (10), which is a much milder inflammatory process than the massive infiltration of the islets in rat and mouse models. It is therefore expected that the effect of insulitis on the accumulation of radiolabeled exendin is of even less importance in humans.

Also, changes in GLP-1R expression during hyperglycemia were observed in previous studies. It should be noted that the marked decrease in GLP-1R expression after near-complete pancreatectomy was observed after 4 weeks of hyperglycemia (6). Similar results were observed by glucose infusion experiments, where hyperglycemia was present for 4 days. In our study, the blood glucose levels were controlled by insulin treatment, and a strong correlation between radiotracer accumulation and the BCM was observed. Although the rats were treated with insulin, fluctuations in blood glucose were observed (Fig. 1), closely mimicking the clinical situation. The strong correlation between ¹¹¹In-exendin accumulation and BCM suggests a negligible effect of fluctuating glycemic levels in T1D on radiotracer uptake. However, the effects of long-term hyperglycemia on the GLP-1R expression and radiolabeled exendin accumulation could not be elucidated in this model.

Besides promising results with radiolabeled and fluorescent exendin analogs for the measurement of the BCM, other radiotracers have been evaluated in animal models and clinical trials. The first tracer that was evaluated for targeting of the β-cells was dihydrotetrabenazine (DTBZ) (which targets VMAT2). Studies in BBDP rats, a spontaneous model for T1D, showed a decrease of 65% in pancreatic standardized uptake value of [¹¹C]DTBZ in severely diabetic rats (11,12). Similar results were obtained with the same radiotracer in streptozotocin-induced β-cell loss in Lewis rats: reduced specific binding index (pancreas-to-kidney ratio) in diabetic rats with a variation in degree of reduction between rats (12). Clinical evaluation of [¹⁸F]FP-(+)-DTBZ revealed a reduction of 38% in standardized uptake value and 40% in binding potential between healthy subjects and subjects with long-standing T1D (>9 years) (13). Moreover, a linear correlation between arginine-stimulated C-peptide and DTBZ binding parameters was observed. A more recent publication showed VMAT2 expression on pancreatic polypeptide cells and cells of the nervous system in the exocrine part of the pancreas (14), which should be taken into account in evaluating the imaging studies using DTBZ. Recently, Eriksson et al. (15) showed that accumulation of the serotonin receptor tracer [¹¹C]5-hydroxytryptophan is reduced by 66% in the pancreas of C-peptide–negative subjects with T1D compared with healthy volunteers, implying that this tracer could also be a useful noninvasive marker to determine the total mass of endocrine cells in the pancreas. Since the serotonin receptor is expressed in all endocrine cells, a combination of [¹¹C]5-hydroxytryptophan imaging of serotonin activity and ¹¹¹In-exendin imaging could provide useful complementary information about the β-cell and total endocrine mass as well as changes in endocrine cell conformation during the development of diabetes.

Previous studies suggest GLP-1R expression in other endocrine and exocrine cells of the pancreas (16–19). In the BBDP rat model, the accumulation of ¹¹¹In-exendin appears to be restricted to β-cells, based on the observation that no signal is observed in autoradiograpic images of rats without any β-cells while α-cells were still present. These observations are in line with our previous findings of a lack of correlation between α-cell mass and ¹¹¹In-exendin accumulation in the pancreas (7).

The pancreatic uptake measured by quantitative analysis of the SPECT images showed a linear correlation with the pancreatic uptake as measured by ex vivo counting of the activity in the pancreata. The relatively high background activity in the SPECT images represents a challenge for accurate quantification of ¹¹¹In-exendin in the pancreas of rats with a low BCM, reducing the correlation coefficient. This phenomenon is of less importance in clinical SPECT or PET imaging, since the background in these images is low and the distance between the pancreas and the kidneys is larger.

The results of this study clearly indicate that radiolabeled exendin imaging represents a reliable technology for the noninvasive determination of the BCM in T1D, and the potential of this method should clearly be exploited to study the changes in BCM in clinical studies.

In conclusion, the strong correlation between pancreatic accumulation of ¹¹¹In-exendin and the BCM observed in BBDP rats is independent of insulitis and fluctuations in blood glucose levels, suggesting that this noninvasive imaging method holds great promise for clinical determination of BCM in individuals with T1D.

Acknowledgments. The authors thank Bianca Lemmers, Kitty Lemmens, Iris Lamers, and Henk Arnts from the Central Animal Facility for technical assistance with the animal experiments.

Funding. This work was supported by JDRF International (37-2011-635).

Duality of Interest. M.G. is a consultant for Boehringer Ingelheim and patent holder in the field. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. M.B. wrote the manuscript. M.B., L.J., C.F., H.P., D.B., and M.v.Z. conducted the experiments. M.B., L.J., and M.G. designed the experiments. M.B., L.J., C.F., H.P., D.B., M.v.Z., O.B., and M.G. revised and approved the manuscript. M.B. and M.G. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 51st Annual Meeting of the European Association for the Study of Diabetes, Stockholm, Sweden, 14–18 September 2015; 29th Annual Congress of the European Association of Nuclear Medicine, Barcelona, Spain, 15–19 October 2016; and the International Diabetes Federation 2015 World Congress, Vancouver, British Columbia, Canada, 30 November–4 December 2015.