In humans, a well-developed serotonin system is localized to the pancreatic islets while being absent in exocrine pancreas. Assessment of pancreatic serotonin biosynthesis could therefore be used to estimate the human endocrine pancreas. Proof of concept was tested in a prospective clinical trial by comparisons of type 1 diabetic (T1D) patients, with extensive reduction of β-cells, with healthy volunteers (HVs). C-peptide–negative (i.e., insulin-deficient) T1D subjects (n = 10) and HVs (n = 9) underwent dynamic positron emission tomography with the radiolabeled serotonin precursor [11C]5-hydroxy-tryptophan ([11C]5-HTP). A significant accumulation of [11C]5-HTP was obtained in the pancreas of the HVs, with large interindividual variation. A substantial and highly significant reduction (66%) in the pancreatic uptake of [11C]5-HTP in T1D subjects was observed, and this was most evident in the corpus and caudal regions of the pancreas where β-cells normally are the major constituent of the islets. [11C]5-HTP retention in the pancreas was reduced in T1D compared with nondiabetic subjects. Accumulation of [11C]5-HTP in the pancreas of both HVs and subjects with T1D was in agreement with previously reported morphological observations on the β-cell volume, implying that [11C]5-HTP retention is a useful noninvasive surrogate marker for the human endocrine pancreas.

The ability to repeatedly quantify the amount of remaining β-cells by noninvasive methods would be of significant importance to obtain further insights into the pathophysiology of diabetes, as well as a tool to evaluate new therapies aiming to prevent β-cell loss. Significant efforts have been made in order to develop such a method (15), but so far no clinically validated technique has been described.

Despite originating from different germinal layers, pancreatic islet cells share many common developmental features with neurons, especially serotonin-producing neurons in the hindbrain (6,7). As shown in mice, a common transcriptional cascade drives the differentiation of β-cells and serotonergic neurons, which imparts the shared ability to produce serotonin (8). Several studies in mice and guinea pigs have also demonstrated the localization of serotonin to the islets of Langerhans, particularly to the granules or microvesicles of the β-cells (811). In humans, similar to mice, guinea pigs, cats, and dogs, the serotonin system is restricted to islets with no expression in the exocrine pancreas (12). In mice, pancreatic serotonin biosynthesis has recently been implicated in insulin secretion (13) and β-cell proliferation during pregnancy (14).

[11C]5-hydroxy-tryptophan ([11C]5-HTP) was originally developed as a positron emission tomography (PET) tracer for assessing the rate of serotonin biosynthesis in health and disease (15) and is currently in clinical use for localization of neuroendocrine tumors (NETs), including insulinomas (16,17). [11C]5-HTP is readily taken up by both endocrine and exocrine tissue by the large amine transporters (LATs) present in most cell types. However, [11C]5-HTP rapidly exits the cell by the same mechanism unless it is further metabolized. This is evident in the pancreas of nonhuman primates, where rapid washout ensues within minutes when the enzyme dopa decarboxylase (DDC, which converts 5-HTP into serotonin) is inhibited (18) as well as in streptozotocin diabetic rats. Hence, retention of [11C]5-HTP demands the presence of the entire molecular machinery involved in serotonin synthesis present in the islet cells, but not in the exocrine cells (having only LATs). [11C]5-HTP will, after decarboxylation, share the fate of endogenous serotonin, including uptake into secretory vesicles by the vesicular monamine transporter 2 (VMAT2) or biodegradation by monoamine oxidase-A (MAO-A) into 5-hydroxyindoleacetic acid (HIAA), which is excreted into urine (15,1821). Radiolabeled 5-HTP therefore accumulates in the form of serotonin in islets of Langerhans as opposed to the exocrine pancreatic parenchyma (9,22). In vitro studies of [11C]5-HTP show high selective retention (8–55 times) in human islets compared with exocrine cells (18). The selectivity in vivo is likely a magnitude higher, due to a small remaining number of islets in density-purified exocrine preparations examined in vitro as well as to a significantly higher blood perfusion in islets (10 times that of acinar pancreas) (23), resulting in proportionally higher tracer delivery. In mice, a similar uptake was found in both α- and β-cells, but the retention in β-cells was markedly higher, resulting in an increased relative accumulation in β-cells 60 min after intravenous administration with 75% of the tracer found within the β-cells (9). These findings in mice have been corroborated by us in nonhuman primates (18) and humans (24) by the observations that the pancreatic [11C]5-HTP uptake is almost exclusively through serotonergic biosynthesis mediated by DDC. Also, in studies of human islets mixed with exocrine tissue in different ratios, a linear correlation between [11C]5-HTP uptake and the percentage of islets present was obtained (18). Importantly, the pancreatic uptake of [11C]5-HTP in streptozotocin diabetic rats, with only 10–20% of β-cells remaining as assessed by morphometric studies, was reduced by 66% when compared with nondiabetic animals (18).

We therefore hypothesized that pancreatic [11C]5-HTP uptake could serve as a noninvasive in vivo surrogate marker for the amount of remaining pancreatic endocrine cells in humans. Initially, we performed a retrospective study in nondiabetic subjects and subjects with type 2 diabetes (T2D) who were examined by [11C]5-HTP due to NET.

Based on the encouraging results obtained, a prospective PET/computed tomography (CT) study measuring pancreatic uptake of [11C]5-HTP in healthy volunteers (HVs) and C-peptide–negative individuals with type 1 diabetes (T1D) was initiated.

Retrospective Study of Subjects With T2D and Nondiabetic Subjects With NETs

[11C]5-HTP is regularly used for localization or staging of NETs. The routine procedure is to examine the patient by static whole-body PET/CT (Discovery ST; GE Healthcare, Milwaukee, WI) 25 min after an intravenous administration of 2–5 MBq/kg [11C]5-HTP. Patients are given 100–200 mg carbidopa orally 1 h prior to tracer administration to decrease renal and urinary radioactivity concentration of tracer metabolites.

In a retrospective study, we analyzed the pancreatic uptake in 10 patients who were examined by [11C]5-HTP-PET/CT because of known or suspected NETs. The inclusion criterion was more than two examinations by [11C]5-HTP-PET/CT between January 2010 and July 2011. The pancreatic uptake of [11C]5-HTP was measured by delineating the pancreas on sequential coregistered CT images by using the computer software PMOD (PMOD Technologies Ltd., Zurich, Switzerland). The pancreatic uptake was expressed as the standardized uptake value, which is the concentration of [11C]5-HTP in tissue, normalized for subject weight and administered radioactivity.

Eight of the subjects were considered as having a normal pancreas and were diagnosed with carcinoids (n = 5), carcinoma (n = 1), or NET of unknown origin (n = 2) (Table 1). The remaining two subjects (gastrinoma, n = 1; carcinoid, n = 1) had been previously diagnosed by T2D prior to being examined for suspected NET.

Table 1

Data on 10 patients referred for two or more HTP-PET/CT examinations

PatientSexAge (years)Weight (kg)Primary tumorLocationMetastasisTreatment
C1 37 64 Carcinoid, atypical Lung — Resection of lower left lobe 
C2 77 49 NET Neck Skin Surgery of lymph nodes 
C3 48 83 Carcinoid, atypical Lung Bone and lymph nodes Resection of upper right lobe; chemotherapy with temozolomide 
C4 23 46 Medullary thyroid carcinoma Thyroid — Thyroidectomy and bilateral lymph node dissection 
C5 22 93 Carcinoid, atypical Lung — Surgery of upper and middle lobe 
C6 17 54 NET Appendix Lymph nodes Right hemicolectomy 
C7 71 57 NET Rectum — Preoperative radiation therapy, then surgery (amputation of rectum) 
C8 41 64 NET Unknown Lymph nodes Surgery of lymph nodes (neck and abdomen); chemotherapy with temozolomide 
T2D:a 54 102 Gastrinoma Duodenum Lymph nodes Gastrinoma removed by duodenotomy and resection of lymph node conglomerate; chemotherapy with STZ plus 5-FU; Sandostatin-LAR 
T2D:b 64 96 Carcinoid, atypical Thymus Lymph nodes (thorax) Surgery of thymic tumor plus postoperative radiation therapy and chemotherapy 
PatientSexAge (years)Weight (kg)Primary tumorLocationMetastasisTreatment
C1 37 64 Carcinoid, atypical Lung — Resection of lower left lobe 
C2 77 49 NET Neck Skin Surgery of lymph nodes 
C3 48 83 Carcinoid, atypical Lung Bone and lymph nodes Resection of upper right lobe; chemotherapy with temozolomide 
C4 23 46 Medullary thyroid carcinoma Thyroid — Thyroidectomy and bilateral lymph node dissection 
C5 22 93 Carcinoid, atypical Lung — Surgery of upper and middle lobe 
C6 17 54 NET Appendix Lymph nodes Right hemicolectomy 
C7 71 57 NET Rectum — Preoperative radiation therapy, then surgery (amputation of rectum) 
C8 41 64 NET Unknown Lymph nodes Surgery of lymph nodes (neck and abdomen); chemotherapy with temozolomide 
T2D:a 54 102 Gastrinoma Duodenum Lymph nodes Gastrinoma removed by duodenotomy and resection of lymph node conglomerate; chemotherapy with STZ plus 5-FU; Sandostatin-LAR 
T2D:b 64 96 Carcinoid, atypical Thymus Lymph nodes (thorax) Surgery of thymic tumor plus postoperative radiation therapy and chemotherapy 

Prospective Study of T1D and Healthy Subjects

Research Subjects

This study was approved by the regional ethical board in Uppsala (EPN 2011/439). All study participants provided written informed consent before participation in any experiments. The studies were conducted according to the principles expressed in the Declaration of Helsinki.

T1D subjects (n = 10) were recruited from the diabetes center at Uppsala University Hospital. Inclusion criteria were >10 years history of T1D and fasting C-peptide concentration <0.003 nmol/L (lower range for detection with hospital standard assay). Two of the T1D subjects have received a whole-pancreas transplant; both have lost the graft function and require the same amounts of exogenous insulin as prior to transplantation.

HVs (n = 9) without any medications and first-degree relatives with T1D were recruited by advertising.

β-Cell Function

All HVs undertook a mixed-meal tolerance test in order to exclude subclinical impairment of glucose metabolism. All examinations took place after overnight fasting and at bed rest. Fasting plasma samples for HbA1c, plasma glucose (P-glucose), and C-peptide concentrations were taken prior to the test. After oral administration of 360 mL Resource protein (Novartis), P-glucose and C-peptide concentrations were measured after 15, 30, 60, 90, and 120 min.

PET Examination

Prior to the PET examination, all participants were fasting for >4 h, and insulin doses to T1D patients were modified accordingly. During the PET examination, P-glucose levels were measured three times and targeted between 3 and 12 mmol/L (mean 8.1 ± 0.62 mmol/L). In one subject, P-glucose dropped to 3.6 mmol/L during the examination, and a slow infusion of glucose was started in order to prevent hypoglycemia.

The subjects were positioned to include pancreas in the center of the 15-cm axial field of view of a Discovery ST PET/CT scanner by assistance of a low-dose CT scout view (140 kV, 10 mAs). Attenuation correction was acquired by a 140-kV Auto mA 10–80 mA CT examination. Prior to administration of [11C]5-HTP, a 6-min dynamic PET examination with oxygene-15–labeled water ([15O]WAT) was performed in order to assess pancreatic perfusion. Thereafter, 2–5 MBq/kg [11C]5-HTP was administered intravenously and subjects underwent a dynamic PET protocol for 60 min (33 frames; 12 × 10 s, 6 × 30 s, 5 × 120 s, 5 × 300 s, and 5 × 600 s).

Image acquisition was performed in three dimensions and reconstructed using an iterative OSEM VUE Point algorithm (two iterations/21 subsets) in a 128 × 128 matrix (zoom 50-cm diameter). Reconstructed data were analyzed using the VOIager 4.0.7 computer software (GE Healthcare, Uppsala, Sweden). Regions of interest were delineated on transaxial CT slices. Entire organs were delineated on sequential slices and combined into volumes of interest (VOIs). Pancreatic volumes were assessed by applying a VOI with VOIager on the CT images.

CT VOIs were transferred to [11C]5-HTP-PET images and modified (due to breathing motion) to match the pancreatic signal within the existing CT VOI. Perfusion was assessed by a one-tissue compartment model from dynamic pancreatic [15O]WAT-PET data, using an aortic VOI as the input function (four pixels in 10 consecutive slices to avoid partial volume effects) (25).

Pancreas was further subdivided into its three main components: caput, corpus, and cauda. The border for cauda was applied at the medial border of the left kidney, and caput was defined as left lateral of the superior mesenteric artery. Mean VOI values for each time frame were used to calculate percentage of injected dose (% ID) in each pancreatic region and for the entire pancreas. To assess background uptake of [11C]5-HTP, VOIs were applied to muscle tissue (erector spinae). In each individual, the muscle VOI volume was identical to that of the pancreas.

Statistical Analysis

All data were expressed as means ± SD. Differences between groups were assessed by a two-tailed Student t test (GraphPad Prism 5, San Diego, CA), where P < 0.05 was considered significant. Normal distribution was confirmed by the D'Agostino-Pearson omnibus normality test.

Retrospective Study of Subjects With T2D and Nondiabetic Subjects With NETs

Ten patients underwent repeated (n = 2–8) [11C]5-HTP-PET/CT during the period 1999–2008 for staging or detection of recurrent NET (Table 1). None of the nondiabetic patients reported hyperglycemias or diabetes. A whole-body (proximal thighs to skull base) PET/CT was performed 25 min after tracer administration. Pancreas was readily visible on PET/CT in all patients (Fig. 1A and B). The uptake in pancreas was high compared with that of other abdominal tissues but varied between patients (% ID/g = 0.0048 ± 0.001) (Fig. 1E). The within-subject variation between examinations was low (coefficient of variation [CV] in this group was 6.7%) (Fig. 1F), demonstrating a high test-retest reproducibility.

Figure 1

Retrospective analysis of uptake of [11C]5-HTP in nondiabetic subjects and subjects with T2D. A–D: [11C]5-HTP-PET maximum intensity projection images (frontal views) showing that the pancreas is clearly visualized (encircled) in representative nondiabetic patients (C3 [A] and C8 [B]), whereas the pancreatic uptake is lower in two patients with T2D (T2D:a [C] and T2D:b [D]). B, bladder; K, kidney. Subjects were examined 25 min post tracer administration. E: Semiquantitative analysis of [11C]5-HTP-PET/CT shows substantial pancreatic tracer accumulation in nondiabetic subjects when compared with the two subjects with T2D. F: The test-retest reproducibility within nondiabetic subjects measured as CV was 6.4 ± 1.7% (range 1.5–14.9). G: Repeated [11C]5-HTP-PET/CT scans in patient T2D:a, developing T2D over a period of 2.5 years, show a decrease in [11C]5-HTP uptake in the pancreas (expressed as standardized uptake value) in parallel with progression of the disease. The third examination, 24 months postdiagnosis of T2D, is the same one as shown in panels C and E.

Figure 1

Retrospective analysis of uptake of [11C]5-HTP in nondiabetic subjects and subjects with T2D. A–D: [11C]5-HTP-PET maximum intensity projection images (frontal views) showing that the pancreas is clearly visualized (encircled) in representative nondiabetic patients (C3 [A] and C8 [B]), whereas the pancreatic uptake is lower in two patients with T2D (T2D:a [C] and T2D:b [D]). B, bladder; K, kidney. Subjects were examined 25 min post tracer administration. E: Semiquantitative analysis of [11C]5-HTP-PET/CT shows substantial pancreatic tracer accumulation in nondiabetic subjects when compared with the two subjects with T2D. F: The test-retest reproducibility within nondiabetic subjects measured as CV was 6.4 ± 1.7% (range 1.5–14.9). G: Repeated [11C]5-HTP-PET/CT scans in patient T2D:a, developing T2D over a period of 2.5 years, show a decrease in [11C]5-HTP uptake in the pancreas (expressed as standardized uptake value) in parallel with progression of the disease. The third examination, 24 months postdiagnosis of T2D, is the same one as shown in panels C and E.

Importantly, we were also able to identify two subjects (T2D:a and T2D:b) with T2D among the patients with NET. The uptake of [11C]5-HTP in the pancreas of these subjects was markedly lower (% ID/g = 0.0016 ± 0.0005) than that observed in nondiabetic subjects (Fig. 1C–E). No overlapping values between the two groups were found. One of the individuals with diabetes (T2D:a) was followed over time with repeated [11C]5-HTP-PET/CT examinations (Fig. 1G). T2D was diagnosed in May 2008 (month 0), and subsequent [11C]5-HTP-PET/CT examinations (months 1, 16, 24, and 30) show a substantial and progressive decrease of tracer uptake in the pancreas over time. In December 2008 (month 7), patient T2D:a started medication with glimepiride. However, her diabetes progressed, and in September 2009 (month 15), exogenous insulin treatment was initiated.

Prospective Study of T1D and Healthy Subjects

Research Subjects

The mean HbA1c in T1D subjects was 7.2 ± 0.2% (55 ± 2 mmol/mol) as compared with 5.3 ± 0.1% (34 ± 0.7 mmol/mol) in HVs (P < 0.05). There were no significant differences in age (26 ± 2 vs. 22 ± 0.6 years), BMI (23 ± 0.7 vs. 25 ± 1 kg/m2), or sex distribution (5/10 vs. 4/9 female). The HVs had a normal fasting and 2-h P-glucose (5.5 ± 0.1 and 5.4 ± 0.2 mmol/L, respectively).

[11C]5-HTP Uptake and Retention in the Pancreas

During the initial minute(s) after [11C]5-HTP administration, we measured high uptake in kidney cortex, pancreas, and spleen, reflecting the rich perfusion of these tissues, which was paralleled by intense tracer concentration in the aorta (Fig. 2, left panels). After 1–10 min, we observed retention of the tracer in the pancreas of HVs, but not in the nonendocrine tissues kidney cortex and spleen, or in the pancreas of T1D patients (Fig. 2, middle panels). Finally, up to 60 min after [11C]5-HTP administration, we recorded that the uptake had faded in most abdominal tissues besides the pancreas of HVs and the kidney medulla (Fig. 2, right panels). The latter likely represented the rapid excretion of the metabolite [11C]HIAA.

Figure 2

Prospective study on [11C]5-HTP uptake in the human endocrine pancreas. Transaxial images of [11C]5-HTP uptake in pancreas (white arrows) in HVs and T1D subjects. Initial high uptake in both HV and T1D pancreas during the perfusion/uptake by LAT phase (0–1 min, left panels). The difference in pancreatic retention of [11C]5-HTP between the two individuals is seen already between 1 and 10 min (middle panels) but is more pronounced 60 min after tracer administration (right panels). A, aorta; K, kidney.

Figure 2

Prospective study on [11C]5-HTP uptake in the human endocrine pancreas. Transaxial images of [11C]5-HTP uptake in pancreas (white arrows) in HVs and T1D subjects. Initial high uptake in both HV and T1D pancreas during the perfusion/uptake by LAT phase (0–1 min, left panels). The difference in pancreatic retention of [11C]5-HTP between the two individuals is seen already between 1 and 10 min (middle panels) but is more pronounced 60 min after tracer administration (right panels). A, aorta; K, kidney.

When quantified, the total pancreatic uptake of [11C]5-HTP, measured as % ID, was markedly reduced throughout the dynamic scan in T1D subjects (Fig. 3A). The reduction was most prominent at the end of the examination, with a mean decrease of 66% (P < 0.001) in [11C]5-HTP uptake in T1D subjects when compared with HVs (Fig. 3B). There was no difference in the amount of unmetabolized [11C]5-HTP in blood plasma between groups, except for a tendency for higher initial plasma concentration in subjects with T1D (Fig. 4A).

Figure 3

Prospective study on [11C]5-HTP uptake in the different anatomical regions of human endocrine pancreas. A: The caput, corpus, and cauda were delineated separately in CT images to assess regional differences in pancreatic uptake. The red lines denote the approximate boundary between caput (left arrow), corpus (middle arrow), and cauda (right arrow). B: Uptake and retention of [11C]5-HTP in pancreas over time as measured by % ID. There is a marked decrease in retention in subjects with T1D, which approaches background levels after 60 min. The background uptake in muscle is presented as a dotted line in each panel. C: Pancreatic uptake at individual level after 60 min. D: The caput is decreased to a lesser extent (60%) and exhibits significant overlap between the T1D and HV groups. E and F: Percent ID is reduced in corpus (68% [E]) and cauda (70% [F]). The two subjects with T1D who received a pancreas graft are shown as circles. **P < 0.01; ***P < 0.001.

Figure 3

Prospective study on [11C]5-HTP uptake in the different anatomical regions of human endocrine pancreas. A: The caput, corpus, and cauda were delineated separately in CT images to assess regional differences in pancreatic uptake. The red lines denote the approximate boundary between caput (left arrow), corpus (middle arrow), and cauda (right arrow). B: Uptake and retention of [11C]5-HTP in pancreas over time as measured by % ID. There is a marked decrease in retention in subjects with T1D, which approaches background levels after 60 min. The background uptake in muscle is presented as a dotted line in each panel. C: Pancreatic uptake at individual level after 60 min. D: The caput is decreased to a lesser extent (60%) and exhibits significant overlap between the T1D and HV groups. E and F: Percent ID is reduced in corpus (68% [E]) and cauda (70% [F]). The two subjects with T1D who received a pancreas graft are shown as circles. **P < 0.01; ***P < 0.001.

Figure 4

Input functions for the prospective study and pancreatic uptake corrected for volume. A: Input functions based on unmetabolized [11C]5-HTP in arterial blood plasma normalized for tracer concentration in blood, subject weight, and administered radioactivity. There are no differences between nondiabetic subjects and subjects with T1D. B: Pancreatic uptake and retention of [11C]5-HTP corrected for pancreatic volume. There is lower retention in subjects with T1D 35–60 min postadministration of [11C]5-HTP. *P < 0.05.

Figure 4

Input functions for the prospective study and pancreatic uptake corrected for volume. A: Input functions based on unmetabolized [11C]5-HTP in arterial blood plasma normalized for tracer concentration in blood, subject weight, and administered radioactivity. There are no differences between nondiabetic subjects and subjects with T1D. B: Pancreatic uptake and retention of [11C]5-HTP corrected for pancreatic volume. There is lower retention in subjects with T1D 35–60 min postadministration of [11C]5-HTP. *P < 0.05.

When subdividing the pancreas into its three anatomical regions, we observed that the reduced [11C]5-HTP uptake in the pancreas of T1D patients was mainly due to lower uptake in the corpus (68% decrease, P < 0.001) (Fig. 3E) and cauda (70% decrease, P < 0.01) (Fig. 3F) and to a lesser extent to decreased uptake in the caput region (60% decrease, P < 0.01) (Fig. 3D).

When correcting for the pancreatic volume in each individual (% ID/g), i.e., tracer accumulation per gram pancreas, we found that the decreased pancreatic tracer accumulation in T1D patients persisted (0.0028 ± 0.001 % ID/g in T1D compared with 0.0046 ± 0.001 % ID/g in HVs corresponding to a 39% decrease 60 min after tracer administration, P < 0.05). Again, tracer accumulation was reduced in the corpus (41%, P < 0.01) and caudal parts of the pancreas (47%, P < 0.01) and to a lesser extent in the caput part (34%, P < 0.05). The differences between groups in % ID/g were seen only between 40 and 60 min postadministration (Fig. 4B).

Size and Perfusion of the Healthy and Diabetic Pancreas

Pancreatic volumes assessed by CT scans were on average 89.4 ± 23.5 cc in HVs and 54.9 ± 14.6 cc in T1D subjects, corresponding to 61% of HV volume (P < 0.01) (Fig. 5A). There was a tendency toward reduced pancreatic blood flow (mL blood/min/mL tissue) (P = 0.068) in the T1D group (28% reduction compared with HVs) (Fig. 5B).

Figure 5

Impact of the diabetic condition on pancreatic size and perfusion. A: Pancreas volume is decreased in subjects with T1D (61% of HVs, P < 0.01). However, pancreatic size alone cannot discriminate between T1D and HV on an individual level due to significant overlap between groups. B: Pancreatic blood flow as assessed by [15O]WAT has a tendency to decrease in the T1D group (28% reduction compared with HVs, P = 0.068). **P < 0.01.

Figure 5

Impact of the diabetic condition on pancreatic size and perfusion. A: Pancreas volume is decreased in subjects with T1D (61% of HVs, P < 0.01). However, pancreatic size alone cannot discriminate between T1D and HV on an individual level due to significant overlap between groups. B: Pancreatic blood flow as assessed by [15O]WAT has a tendency to decrease in the T1D group (28% reduction compared with HVs, P = 0.068). **P < 0.01.

In this clinical study, we present evidence that [11C]5-HTP can be used as an imaging biomarker for the native endocrine pancreas. In the prospective clinical study, accumulation of the tracer was assayed 60 min after injection in order to preferentially target the β-cells, utilizing their prolonged retention time of the tracer when compared with non–β-cells within the islets and lack of retention in the exocrine pancreas.

Pancreatic [11C]5-HTP accumulation in subjects with T1D was substantially reduced (mean reduction of 66%) when compared with that recorded in HVs. We stress that all remaining pancreatic neuroendocrine tissue should retain [11C]5-HTP, not only β-cells. Thus, complete loss of all β-cells would only partly reduce the endocrine signal. The contribution of the β-cell mass to the mass of individual islets is still a matter of disagreement, due to differences in study design, sample size, and whether number or volume contribution was measured. The volumetric contribution is likely the suitable parameter when assessing β-cell mass and can be approximated to 62% (range 46–75%) by averaging previously reported values from nine studies using human pancreatic sections (26). Thus, a complete loss of all β-cells should reduce the [11C]5-HTP signal by ∼62%.

There are ∼1.5 million islets within the human endocrine pancreas. Of these, the absolute majority has a diameter <50 µm and only 200–300 islets have a diameter of 250 µm or more (27). With the current resolution of PET technology, individual islets cannot be detected. Instead, the integrated signal from the pancreas (% ID) represents a composite value from the total endocrine pancreas. However, subanalysis of the signal from different regions of the pancreas is feasible, which enabled us to compare changes in [11C]5-HTP accumulation in regions differing in their fraction of β-cells within the islets. We recorded a more substantial decrease in [11C]5-HTP accumulation in the β-cell–rich corpus and cauda of the pancreas of T1D subjects than in the caput region where pancreatic polypeptide and, to a lesser extent, α-cells dominate. Interestingly, in autopsy studies, the total endocrine mass in C-peptide–negative subjects with long-standing T1D has been reported to be ∼30% of that in nondiabetic controls (28). Noteworthy, similar to our observations by [11C]5-HTP-PET, no changes in endocrine density were found in the ventral pancreatic polypeptide cell–rich segment of the pancreas (28).

The total α- and β-cell mass in nondiabetic subjects has been described in carefully conducted morphometric studies and shows a surprisingly high interindividual variation from 0.35 to 2.4 g, with a strong correlation between the α-cell (0.1–1 g) and β-cell (0.25–1.5 g) masses (29). The pancreatic [11C]5-HTP accumulations of the HVs in the current study are in good agreement with previously reported morphological findings, with a large interindividual variation in total pancreatic uptake of [11C]5-HTP (0.2–0.7 % ID).

The retrospective study, based on an analysis of static [11C]5-HTP images of NET patients, should be interpreted with caution due to the lack of tracer pharmacokinetics and information on pancreatic perfusion as well as the nonstandardization of the investigated subjects. The high uptake of [11C]5-HTP in NETs, which is the basis of using serotonergic biosynthesis as a diagnostic biomarker, may also increase the endogenous levels of 5-HTP or serotonin in plasma. This could potentially induce changes in uptake of [11C]5-HTP in pancreatic endocrine tissue due to competition for DDC.

Therefore, a prospective study in subjects with T2D, repeatedly examined over a period of several years, will be required to confirm these initial results. However, NET patients diagnosed with T2D in the herein reported retrospective study had markedly low pancreatic accumulation of [11C]5-HTP. A progressive loss of [11C]5-HTP uptake in the pancreas was observed in one patient (T2D:a) who was repeatedly examined during a 2-year period. It is not known how the chemotherapy, including STZ, in combination with glimepiride and concurrent β-cell exhaustion may affect the β-cells and their uptake and retention of [11C]5-HTP in subjects with T2D. At the time of the PET examination, none of the other retrospectively investigated subjects were diabetic. These subjects could potentially have been prediabetic and later developed T2D. Unfortunately, this information was unavailable.

It would be of interest to compare the pancreatic uptake in T1D and T2D subjects, but several factors make it difficult to draw conclusions from a direct comparison between the retrospective and prospective studies. First, the % ID/g values are reported at different time points (T2D, 25 min post tracer administration; T1D, 60 min post administration). Second, the T2D and nondiabetic subjects in the retrospective study received carbidopa 1 h prior to [11C]5-HTP, which also affects the pancreatic uptake. It is therefore possible to compare between groups within each study as they were treated similarly but not between the studies.

Di Gialleonardo et al. (22) recently showed that absolute uptake of [11C]5-HTP in nonendocrine PANC1 cells was comparable to that of INS-1 cells, but not mediated by normal serotonergic metabolism. Based on these results, they posited that the specific endocrine signal in vivo would not be higher than the nonspecific signal from exocrine pancreas. Following this line of reasoning, the pancreatic signal would therefore be nonspecific in nature, thereby reducing the possibilities of visualizing the specific uptake, mediated by serotonin biosynthesis, in endocrine pancreas.

Thus, the conclusion by Di Gialleonardo et al. (22) should, as pointed out by the authors, be interpreted with caution since it relies on in vitro observations in cell lines. By selective inhibition of DDC in nonhuman primate, as well as in STZ diabetic rats, we have shown that the entire [11C]5-HTP uptake in pancreas is specific for serotonergic biosynthesis and thus consists of [11C]serotonin or its metabolites (18). Based on these preclinical studies, as well as on the herein presented clinical studies showing a correlation between plasma C-peptide and pancreatic uptake of [11C]5-HTP, we conclude that in humans, the nonspecific uptake in vivo is negligible and washed out within minutes after injection of the tracer.

Patients with T1D are known to have a marked reduction in the total exocrine volume, concomitant with a decrease in the exocrine functional capacity (30,31). Also in this study, we observed a mean substantial reduction of the total pancreatic volume in T1D patients. However, this measurement could not discriminate between nondiabetic and diabetic subjects due to a large overlap between the two groups (Fig. 5A and Fig. 6A). Moreover, the uptake of [11C]5-HTP per gram pancreas was significantly decreased in subjects with T1D when compared with that in HVs, which excludes mere nonspecific uptake of the tracer in exocrine pancreas to explain the observed differences between HVs and T1D patients.

Figure 6

Uptake of [11C]5-HTP in the human pancreas is not dependent on perfusion. A: Correlation between pancreatic size and % ID in all individuals. Pancreatic uptake of [11C]5-HTP as % ID completely separates between the TID and HV groups, whereas there is a substantial overlap in pancreatic volume despite the difference at group level. B: There is no correlation between the perfusion of pancreas and the uptake of [11C]5-HTP as measured by % ID.

Figure 6

Uptake of [11C]5-HTP in the human pancreas is not dependent on perfusion. A: Correlation between pancreatic size and % ID in all individuals. Pancreatic uptake of [11C]5-HTP as % ID completely separates between the TID and HV groups, whereas there is a substantial overlap in pancreatic volume despite the difference at group level. B: There is no correlation between the perfusion of pancreas and the uptake of [11C]5-HTP as measured by % ID.

Pancreatic blood flow comprises a sum of the blood perfusion of the exocrine and endocrine parts. Rodent studies show that the islet blood flow normally constitutes ∼10% of the total pancreatic blood flow even though the islet tissue comprises only 1–2% of the total pancreatic volume (23). In humans, no techniques are available for specific studies of islet blood perfusion, but the perfusion of the whole pancreas in HVs and T1D subjects, as measured by magnetic resonance perfusion imaging, did not differ (32). In the current study, a tendency to a decrease in mean pancreatic blood flow by 20% was recorded. There was no clear correlation (P = 0.28) between the uptake of [11C]5-HTP and the blood perfusion as measured by [15O]WAT (Fig. 6B). There was an apparent increase in initial tracer delivery in nondiabetic subjects during the first minutes. This could be considered as a flow-dependent effect and would govern the subsequent difference in the uptake over the course of the examination. However, in this case, this was likely largely dependent on pancreatic size rather than perfusion, since no difference in initial delivery remained when correcting for pancreatic volume (Fig. 4B).

The lack of a correlation between [11C]5-HTP accumulation and blood flow does not rule out that endocrine blood flow may be partly or entirely responsible for the differences seen between nondiabetic and diabetic subjects. PET studies in primates show that the pancreatic uptake of [11C]5-HTP is an active process that can be enhanced or diminished by pharmaceutical intervention, i.e., carbidopa or clorgyline (18), but it is unclear if these results are directly translatable to humans. Full quantification by compartmental models using the metabolite-corrected arterial plasma input functions is required to answer this question, which is important for defining the potential and limitation of the use of [11C]5-HTP in PET studies of the endocrine pancreas.

VMAT2, which mediates serotonin uptake into secretory vesicles, has been identified as a tissue-restricted transcript in human islets and suggested as a suitable target for imaging of β-cells (33). PET studies using radiolabeled dihydrotetrabenazine (DTBZ), a ligand for VMAT2, have been conducted in a range of in vitro and in vivo studies. Although DTBZ has many attributes that suggest that it may provide a suitable probe for PET imaging, recent published communications raise severe questions concerning the specificity of this probe in vitro and in vivo (3437) and conclude that the pancreatic uptake of DTBZ is mainly due to nonspecific binding in the exocrine pancreas. In line with these observations, the signal response in subjects with manifest T1D was reduced with only 14% of the pancreatic signal in HVs (2). Significant overlap between subjects with T1D and HVs was also found in a recent study both when measuring radiotracer concentration and total uptake in the pancreas (1). Importantly, these studies did not correct for the impact of reduced blood flow to the diabetic pancreas, which should reduce pancreatic accumulation of a perfusion-dependent tracer in subjects with T1D. Our results demonstrate that the reduction of blood perfusion in T1D approaches 30%, which mirrors the decrease of DTBZ uptake found in subjects with manifest T1D in the aforementioned studies (1,2). Notably, VMAT2 may still constitute a suitable target for β-cell imaging, but DTBZ in its current form has not been shown to be a suitable PET tracer to quantify the β-cell mass (38).

The ability to measure the amount of remaining β-cells by noninvasive methods such as [11C]5-HTP-PET will most likely have no role in the diagnosis of T1D. However, it would be of immense value for our understanding of the pathophysiology of both T1D and T2D and as a tool for evaluation of new therapies aiming to prevent or reverse β-cell loss or to regenerate or replace β-cells. As indicated in the retrospective study, optimal utilization of the herein presented [11C]5-HTP-PET technology will most likely entail repeated examinations within the same subject, which should enable detection of even small changes in the endocrine pancreas when considering its low CV for within-subject variation. The PET technology can be combined with either CT or MRI to provide an anatomical correlate to the PET signal. We have developed a noise reduction filter based on principal component analysis technology, which allows for reducing the radioactivity dose of the PET tracer by a factor of 4 without compromising spatial resolution or quantification (38). The use of radioisotopes with short half-lives and optimal data acquisition using PET/MRI instead of PET/CT allows further reduction of the total radiation dose to the extent that repeated examinations can be frequently performed (three to four times annually) in the same subject and also allow for examinations of children. The results from the herein presented prospective and retrospective clinical studies show that [11C]5-HTP constitutes a promising PET tracer for quantitative imaging of the human endocrine pancreas that readily can discriminate between nondiabetic and diabetic subjects. The dependency on blood flow for tracer uptake will require further studies to define the potential and limitations of the use of [11C]5-HTP in this setting.

Clinical trial reg. no. NCT01552811, clinicaltrials.gov.

Acknowledgments. The authors thank Lina Carlbom (Uppsala University) for excellent technical assistance.

Funding. This study was supported by grants from JDRF, the Swedish Medical Research Council, VINNOVA, the Novo Nordisk Foundation, Diabetes Wellness Network Sverige, EXODIAB, the Tore Nilsons Foundation, and Barndiabetesfonden.

Duality of Interest. J.W.E. and L.J. are employed by AstraZeneca R&D. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. O.E., D.E., P.-O.C., and O.K. researched data and wrote the manuscript. R.K.S., E.J., M.L., and A.-R.B. researched data and reviewed and edited the manuscript. G.A., J.S., J.W.E., A.S., and L.J. contributed to discussion and reviewed and edited the manuscript. H.A. and B.E. researched data, contributed to discussion, and reviewed and edited the manuscript. O.E. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Normandin
MD
,
Petersen
KF
,
Ding
YS
, et al
.
In vivo imaging of endogenous pancreatic β-cell mass in healthy and type 1 diabetic subjects using 18F-fluoropropyl-dihydrotetrabenazine and PET
.
J Nucl Med
2012
;
53
:
908
916
[PubMed]
2.
Goland
R
,
Freeby
M
,
Parsey
R
, et al
.
11C-dihydrotetrabenazine PET of the pancreas in subjects with long-standing type 1 diabetes and in healthy controls
.
J Nucl Med
2009
;
50
:
382
389
[PubMed]
3.
Andralojc
K
,
Srinivas
M
,
Brom
M
, et al
.
Obstacles on the way to the clinical visualisation of beta cells: looking for the Aeneas of molecular imaging to navigate between Scylla and Charybdis
.
Diabetologia
2012
;
55
:
1247
1257
[PubMed]
4.
Selvaraju
RK
,
Velikyan
I
,
Johansson
L
, et al
.
In vivo imaging of the glucagonlike peptide 1 receptor in the pancreas with 68Ga-labeled DO3A-exendin-4
.
J Nucl Med
2013
;
54
:
1458
1463
[PubMed]
5.
Brom
M
,
Woliner-van der Weg
W
,
Joosten
L
, et al
.
Non-invasive quantification of the beta cell mass by SPECT with ¹¹¹In-labelled exendin
.
Diabetologia
2014
;
57
:
950
959
[PubMed]
6.
Wilson
ME
,
Scheel
D
,
German
MS
.
Gene expression cascades in pancreatic development
.
Mech Dev
2003
;
120
:
65
80
[PubMed]
7.
Cordes
SP
.
Molecular genetics of the early development of hindbrain serotonergic neurons
.
Clin Genet
2005
;
68
:
487
494
[PubMed]
8.
Ohta
Y
,
Kosaka
Y
,
Kishimoto
N
, et al
.
Convergence of the insulin and serotonin programs in the pancreatic β-cell
.
Diabetes
2011
;
60
:
3208
3216
[PubMed]
9.
Ekholm
R
,
Ericson
LE
,
Lundquist
I
.
Monoamines in the pancreatic islets of the mouse. Subcellular localization of 5-hydroxytryptamine by electron microscopic autoradiography
.
Diabetologia
1971
;
7
:
339
348
[PubMed]
10.
Bird
JL
,
Wright
EE
,
Feldman
JM
.
Pancreatic islets: a tissue rich in serotonin
.
Diabetes
1980
;
29
:
304
308
[PubMed]
11.
Schraenen
A
,
Lemaire
K
,
de Faudeur
G
, et al
.
Placental lactogens induce serotonin biosynthesis in a subset of mouse beta cells during pregnancy
.
Diabetologia
2010
;
53
:
2589
2599
[PubMed]
12.
Ding
WG
,
Fujimura
M
,
Tooyama
I
,
Kimura
H
.
Phylogenetic study of serotonin-immunoreactive structures in the pancreas of various vertebrates
.
Cell Tissue Res
1991
;
263
:
237
243
[PubMed]
13.
Paulmann
N
,
Grohmann
M
,
Voigt
JP
, et al
.
Intracellular serotonin modulates insulin secretion from pancreatic beta-cells by protein serotonylation
.
PLoS Biol
2009
;
7
:
e1000229
[PubMed]
14.
Kim
H
,
Toyofuku
Y
,
Lynn
FC
, et al
.
Serotonin regulates pancreatic beta cell mass during pregnancy
.
Nat Med
2010
;
16
:
804
808
[PubMed]
15.
Lundquist
P
,
Blomquist
G
,
Hartvig
P
, et al
.
Validation studies on the 5-hydroxy-L-[beta-11C]-tryptophan/PET method for probing the decarboxylase step in serotonin synthesis
.
Synapse
2006
;
59
:
521
531
[PubMed]
16.
Orlefors
H
,
Sundin
A
,
Ahlström
H
, et al
.
Positron emission tomography with 5-hydroxytryprophan in neuroendocrine tumors
.
J Clin Oncol
1998
;
16
:
2534
2541
[PubMed]
17.
Orlefors
H
,
Sundin
A
,
Garske
U
, et al
.
Whole-body (11)C-5-hydroxytryptophan positron emission tomography as a universal imaging technique for neuroendocrine tumors: comparison with somatostatin receptor scintigraphy and computed tomography
.
J Clin Endocrinol Metab
2005
;
90
:
3392
3400
[PubMed]
18.
Eriksson
O
,
Selvaraju
RK
,
Johansson
L
, et al
.
Quantitative imaging of serotonergic biosynthesis and degradation in the endocrine pancreas
.
J Nucl Med
2014
;
55
:
460
465
[PubMed]
19.
Eriksson
O
,
Selvaraju
R
,
Borg
B
,
Asplund
V
,
Estrada
S
,
Antoni
G
.
5-Fluoro-[β-¹¹C]-L-tryptophan is a functional analogue of 5-hydroxy-[β-¹¹C]-L-tryptophan in vitro but not in vivo
.
Nucl Med Biol
2013
;
40
:
567
575
[PubMed]
20.
Gylfe
E
,
Hellman
B
,
Sehlin
J
,
Taljedal
IB
.
Amino acid conversion into 5-hydroxytryptamine in pancreatic beta-cells
.
Endocrinology
1973
;
93
:
932
937
[PubMed]
21.
Jaim-Etcheverry
G
,
Zieher
LM
.
Electron microscopic cytochemistry of 5-hydroxytryptamine (5-HT) in the beta cells of guinea pig endocrine pancreas
.
Endocrinology
1968
;
83
:
917
923
[PubMed]
22.
Di Gialleonardo
V
,
Signore
A
,
Scheerstra
EA
, et al
.
11C-hydroxytryptophan uptake and metabolism in endocrine and exocrine pancreas
.
J Nucl Med
2012
;
53
:
1755
1763
[PubMed]
23.
Carlsson
PO
,
Olsson
R
,
Källskog
O
,
Bodin
B
,
Andersson
A
,
Jansson
L
.
Glucose-induced islet blood flow increase in rats: interaction between nervous and metabolic mediators
.
Am J Physiol Endocrinol Metab
2002
;
283
:
E457
E464
[PubMed]
24.
Orlefors
H
,
Sundin
A
,
Lu
L
, et al
.
Carbidopa pretreatment improves image interpretation and visualisation of carcinoid tumours with 11C-5-hydroxytryptophan positron emission tomography
.
Eur J Nucl Med Mol Imaging
2006
;
33
:
60
65
[PubMed]
25.
Herrero
P
,
Markham
J
,
Bergmann
SR
.
Quantitation of myocardial blood flow with H2 15O and positron emission tomography: assessment and error analysis of a mathematical approach
.
J Comput Assist Tomogr
1989
;
13
:
862
873
[PubMed]
26.
Pisania
A
,
Weir
GC
,
O’Neil
JJ
, et al
.
Quantitative analysis of cell composition and purity of human pancreatic islet preparations
.
Lab Invest
2010
;
90
:
1661
1675
[PubMed]
27.
Hellman
B
.
Actual distribution of the number and volume of the islets of Langerhans in different size classes in non-diabetic humans of varying ages
.
Nature
1959
;
184
(
Suppl. 19
):
1498
1499
[PubMed]
28.
Rahier
J
,
Goebbels
RM
,
Henquin
JC
.
Cellular composition of the human diabetic pancreas
.
Diabetologia
1983
;
24
:
366
371
[PubMed]
29.
Rahier
J
,
Guiot
Y
,
Goebbels
RM
,
Sempoux
C
,
Henquin
JC
.
Pancreatic beta-cell mass in European subjects with type 2 diabetes
.
Diabetes Obes Metab
2008
;
10
(
Suppl. 4
):
32
42
[PubMed]
30.
Goda
K
,
Sasaki
E
,
Nagata
K
,
Fukai
M
,
Ohsawa
N
,
Hahafusa
T
.
Pancreatic volume in type 1 and type 2 diabetes mellitus
.
Acta Diabetol
2001
;
38
:
145
149
[PubMed]
31.
Williams
AJ
,
Thrower
SL
,
Sequeiros
IM
, et al
.
Pancreatic volume is reduced in adult patients with recently diagnosed type 1 diabetes
.
J Clin Endocrinol Metab
2012
;
97
:
E2109
E2113
[PubMed]
32.
Hirshberg
B
,
Qiu
M
,
Cali
AM
, et al
.
Pancreatic perfusion of healthy individuals and type 1 diabetic patients as assessed by magnetic resonance perfusion imaging
.
Diabetologia
2009
;
52
:
1561
1565
[PubMed]
33.
Maffei
A
,
Liu
Z
,
Witkowski
P
, et al
.
Identification of tissue-restricted transcripts in human islets
.
Endocrinology
2004
;
145
:
4513
4521
[PubMed]
34.
Jahan
M
,
Eriksson
O
,
Johnström
P
, et al
.
Decreased defluorination using the novel beta-cell imaging agent [18F]FE-DTBZ-d4 in pigs examined by PET
.
EJNMMI Res
2011
;
1
:
33
[PubMed]
35.
Fagerholm
V
,
Mikkola
KK
,
Ishizu
T
, et al
.
Assessment of islet specificity of dihydrotetrabenazine radiotracer binding in rat pancreas and human pancreas
.
J Nucl Med
2010
;
51
:
1439
1446
[PubMed]
36.
Eriksson
O
,
Jahan
M
,
Johnström
P
, et al
.
In vivo and in vitro characterization of [18F]-FE-(+)-DTBZ as a tracer for beta-cell mass
.
Nucl Med Biol
2010
;
37
:
357
363
[PubMed]
37.
Virostko
J
,
Henske
J
,
Vinet
L
, et al
.
Multimodal image coregistration and inducible selective cell ablation to evaluate imaging ligands
.
Proc Natl Acad Sci U S A
2011
;
108
:
20719
20724
[PubMed]
38.
Lubberink M, Lilja J, Lindsjö L, et al. Low-dose, quantitative images of cerebral glucose metabolism using principal component analysis of dynamic [18F]FDG PET images. Presented at the SNM Annual Meeting, 9–13 June 2012, Miami Beach, FL