Upper and lower endoscopy enable the detection of esophageal, gastric, colorectal, and some terminal ileal NETs. Double-balloon or push enteroscopy is more efficient for small intestine lesions but remains a time-consuming technique and uncomfortable for the patient [14]. Capsule endoscopy seems an attractive technique for esophagus, stomach, and small intestine examination but does not permit lesion biopsy [15]. The main advantage of video capsule is that it is a noninvasive technique able to analyze the entire small bowel. However, it has only moderate sensitivity (46–65 %) for tumor detection [16]. The use of transabdominal ultrasonography (TAUS) in GI NET diagnosis and staging is mainly limited to the solid viscera and suffers from inter-operator variability. Overall, US has limited sensitivity (13–27 %) for the detection of gastrointestinal NETs [17]. Intraoperative US could be also performed but it prolongs operating time, and it is not free of potential risk of iatrogenic injury during the diagnostic procedure. Conversely, endoscopic US (EUS) has a remarkable role. EUS is a highly sensitive technique for the detection of gastric, duodenal, and rectal NETs permitting not only the identification of small submucosal tumors but even to perform diagnostic procedures such as biopsy or fine-needle aspiration [18]. However, EUS remains a more invasive imaging method. In patients suspected of having small-bowel tumors, the enteroclysis and barium-contrast examinations may show fixation, separation, thickening, and angulation of the bowel loops, but they are rarely diagnostic. Small primary tumors are difficult to visualize unless there is secondary fibrosis. At present, barium-contrast studies have been largely replaced by multi-planar contrast-enhanced CT or MR imaging followed by small-bowel distention before a focused CT or MR enterography or enteroclysis to improve tumor detection. Small-bowel tumors are multifocal in about 30 % of the cases [19], and they manifest as polypoid lesions or hypervascular parietal concentric or asymmetrical thickening. The sensitivity and specificity for tumor detection of CT and MR enterography or enteroclysis have been estimated at 100 % and 86–94 %, and 96 % and 95–98 %, respectively. Overall sensitivity increases with rising tumor size [20–24]. Artifacts from bowel motion and lower spatial resolution compared with CT may be limiting factors of MR enterography or enteroclysis.

CT remains the first-choice diagnostic procedure for tumor staging allowing the assessment of local extent to the mesentery as well as metastases to lymph nodes, liver, and lungs. Typically, the appearance of mesenteric invasion by carcinoid tumor on CT is a spiculated mass with heterogeneous contours, variably associated to central calcifications and usually near the primary tumor (“cartwheel” pattern) [25]. Metastatic lymph nodes are frequently in the mesentery or retroperitoneal and paracardial gastric lymph nodes. For the assessment of metastatic hepatic disease, all available imaging modalities frequently miss lesions sized less than 5 mm [26]. Nevertheless, multiphasic imaging is recommended, with the hepatic arterial phase being the most informative for lesion detection [27]. On multiphasic contrast-enhanced CT or MR images, hepatic lesions are usually hypervascular in the arterial phase and demonstrate washout in the late phase [28, 29]. MRI is a very sensitive technique for the detection of liver metastases, and it is considered more accurate than both US and CT [26]. Multiphasic MRI with fat-suppressed contrast-enhanced T1-weighted imaging provides the best accuracy [29, 30]. Typically, NET lesions show T2 hyperintensity and T1 hypointensity. The use of diffusion-weighted imaging (DWI) and apparent diffusion coefficients (ADC) could improve the detection ability of MRI for malignant liver lesions [31]. An added advantage of using DWI is its ability to reflect lesion changes in response to treatment [32]. Finally, the advantages of MRI over CT are the lack of ionizing radiation and the use of gadolinium chelate contrast agents, which have a better safety profile in terms of allergic reactions and nephrotoxicity than iodine-contrast agents. Metastases to the bone arise more commonly from foregut primary tumors. The preferred site of metastatic spread is the axial skeleton and radiographic signs may be easily missed. On standard radiography and CT, bone metastases frequently demonstrate either an osteosclerotic or a mixed osteolytic–osteosclerotic pattern. MRI is the most sensitive technique for detection of bone marrow metastases [33–35].

The somatostatin receptor (SSTR) is a seven-transmembrane G-coupled receptor, of which six human subtypes have been described and which upon activation reduces endocrine secretion and cell growth and increases apoptosis [36]. Upon binding of an agonist to the receptor, the receptor/ligand complex will internalize in the cell, and the receptor will be later recycled to the cell surface. A vast majority (>85 %) of intestinal NETs will have moderate to high overexpression of one of the six SSTR subtypes (1, 2A, 2B, 3, 4, and 5) [36], most frequently subtype 2A [37]. The SSTR has been used as a molecular target in nuclear medicine since more than 25 years [38]. Radioligands binding to the SSTR detected by gamma camera scintigraphy and SPECT were the first to be developed and are based on synthetic somatostatin analogues (SSAs) such as the 8 amino acid derivative octreotide or further optimized chemical variants of this vector backbone. Different chelators have been developed to allow labeling of these vector molecules with a whole range of radioisotopes. One gamma camera radiopharmaceutical is commercially available (¹¹¹In-pentetreotide; Octreoscan™), and technetium-99 m-labeled tracers have been described as well, using both octreotide and octreotate backbones with a HYNIC chelator/linker. The use of somatostatin receptor scintigraphy (SRS) is recommended in the guidelines at diagnosis [1] and during follow-up [39]. It offers high sensitivity and specificity in detecting tumoral lesions in grades 1 and 2 NET, outperforming other nuclear medicine tracers such as ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) and ¹²³I-meta-iodobenzylguanidine (¹²³I-MIBG). In a study by Binderup and colleagues, SRS detected more lesions in all organs studied than ¹⁸F-FDG or ¹²³I-MIBG with a sensitivity of 90 % vs. 57 % and 53 %, respectively [40]. Of note, in highly proliferative NET (Ki-67 > 15 %), ¹⁸F-FDG had a sensitivity of 92 % vs. 69 % for SRS [40]. Besides tumor detection, SRS allows to determine functional target presence for treatment with cold SSAs. Two randomized controlled trials (RCTs) have shown a prolongation of progression-free survival (PFS) in NET patients treated with SSAs; one trial included well-differentiated metastatic midgut NETs (PROMID) [41], and the other included well-differentiated nonfunctional tumors of the pancreas, midgut, hindgut, or unknown origin (CLARINET) [42]. A positive SRS was an inclusion criterion in the CLARINET study, whereas in the PROMID trial, 73 out of 85 patients had a scintigraphy performed, with 86 % positive and 14 % negative patients [41], highlighting the importance of the scintigraphy as a selection tool for this treatment. For peptide receptor radionuclide therapy (PRRT), good uptake on pre-therapy SRS has always been considered a prerequisite. For example, in the Netter-1 RCT, a positive scintigraphy on all evaluable lesions was an inclusion criterion [43].

Uptake on whole-body planar ¹¹¹In-pentetreotide scintigraphy is typically scored according to the Krenning scale, with grade 1 representing tumor uptake lower than the liver, grade 2 tumor uptake equal to the liver, grade 3 uptake higher than liver, and grade 4 tumor uptake as the most intense site in the body (usually meaning higher than spleen uptake). A grade 3 uptake is typically required for PRRT candidates.

In the last decade, a novel class of radiopharmaceuticals for SSTR imaging has emerged as the current gold standard (Fig. 15.1). These tracers are based on octreotide derivatives labeled with the positron-emitting radionuclide gallium-68. Gallium-68 is a radioisotope of the metal gallium. It decays primarily through positron emission (abundance 89 %). It has a half-life (T_1/2) of 68 min and can be obtained from a germanium-68/gallium-68 generator (T_1/2: 271 days), thus allowing in-house production of labeled radiopharmaceuticals. Gallium can be chelated by a number of different chelators to label peptides or other biomolecules for detection with PET, PET/CT, or PET/MR. Labeling has to be done at the site of administration as the yield of the generator, and the half-life usually does not allow for off-site transport.

There are several gallium-68-labeled tracers for SSTR imaging that have been described and are in clinical use in Europe of which ⁶⁸Ga-DOTATOC, ⁶⁸Ga-DOTATATE, and ⁶⁸Ga-DOTANOC are the most commonly used. “DOTA” refers to the chelator, which is covalently bound to a slightly modified synthetic octapeptide with high affinity for the SSTR (all have high affinity for SSTR₂, the most overexpressed subtype). They are collectively referred to as ⁶⁸Ga-DOTA-peptides or ⁶⁸Ga-OctreoPET.

These ⁶⁸Ga-DOTA-peptides have advantages over gamma scintigraphy-based imaging. They have a much higher affinity for SSTR₂ compared to ¹¹¹In-pentetreotide (IC₅₀ = 2.5 nM for ⁶⁸Ga-DOTATOC, 0.2 nM for ⁶⁸Ga-DOTATATE, and 1.9 nM for ⁶⁸Ga-DOTANOC, vs. 22 nM for ¹¹¹In-pentetreotide) [44]. This better affinity profile combined with the physical advantages of current clinical PET cameras over gamma cameras, with higher spatial resolution and higher sensitivity (detected events per unit of radioactivity), allows detection of smaller lesions and detection of lesions with even low or moderate SSTR expression, resulting in a higher sensitivity (Fig. 15.2). There is strong scientific data that the ⁶⁸Ga-DOTA-peptides are significantly better diagnostic agents for performing SSR imaging than ¹¹¹In-pentetreotide. Gabriel et al. [45] compared ⁶⁸Ga-DOTATOC with ¹¹¹In-pentetreotide in 84 NET patients and found a sensitivity of 97 % for ⁶⁸Ga-DOTATOC vs. 52 % for ¹¹¹In-pentetreotide. ⁶⁸Ga-DOTATOC showed better performance for small lesions in lymph nodes and for bone metastases (Fig. 15.3). Buchmann et al. [46] compared ⁶⁸Ga-DOTATOC with ¹¹¹In-pentetreotide in 27 NET patients and found a sensitivity of 100 % and 66 %, respectively. They concluded that the PET ligand was superior for detection of lung and bone metastases. Van Binnebeek et al. [47] have recently shown similar results with sensitivity of 99.9 % for ⁶⁸Ga-DOTATOC with only 60 % for ¹¹¹In-pentetreotide in metastatic NET patients scheduled for PRRT. Furthermore, they showed that in up to 80 % of the patients, the PET ligand would detect at least 20 % more lesions, showing that the benefit of ⁶⁸Ga-DOTA-peptide PET reaches a large fraction of patients. Recent data from Northern America have shown a lesion sensitivity of 95.1 % vs. 45.3 % for anatomic imaging and 30.9 % for ¹¹¹In-pentetreotide SPECT/CT in 131 GEP-NETs or NETs from unknown primary [48]. In four out of 14 patients (29 %), ⁶⁸Ga-DOTATATE PET/CT found a previously unknown primary tumor. The therapy recommendation was changed on the basis of the ⁶⁸Ga-DOTATATE PET/CT in one third of the patients [48]. Again, large differences between ⁶⁸Ga-DOTATOC and ¹¹¹In-pentetreotide were reported by Morgat and colleagues in the follow-up of MEN1 patients [49].

Although there are some differences in the affinity profiles (⁶⁸Ga-DOTATOC: SSTR₂ > SSTR₅; ⁶⁸Ga-DOTATATE: SSTR₂; ⁶⁸Ga-DOTANOC SSTR₂ > SSTR₅ > SSTR₃), clinical comparative studies have only shown minor if there are any differences in lesion detection rate and lesion uptake when comparing head to head two of these PET SSTR ligands (Fig. 15.4). The consensus is that for the vast majority of clinical indications, these tracers are equivalent [50]. In a head-to-head comparison in the same patients (n = 40) between ⁶⁸Ga-DOTATOC and ⁶⁸Ga-DOTATATE, 262 and 254 (97 %) lesions were detected with an average standardized uptake value (SUV) of 20.4 and 16.0, respectively [51].

The clinical impact of ⁶⁸Ga-DOTA-peptide PET is mainly due to:

When comparing a ⁶⁸Ga-DOTA-peptide PET image with a previous scan of the same patients, one needs to be careful that visualizing more lesions does not necessarily mean that the patient is progressive.

The physiological biodistribution in normal patients is as follows: high uptake in spleen (hottest organ in normal patient), kidneys, adrenal glands, liver, pituitary gland, and the bladder (tracer excretion). There is also moderate uptake within the pancreas and more notably the uncinate process, the thyroid gland, the intestine. There is low uptake in the lungs, muscles (including myocardium), and the bone marrow.

Typical pitfalls in SSTR PET include misinterpretation of the physiological uptake in the uncinate process of the pancreas as a pancreatic NET, misinterpretation of a small meningioma (which can also have high SSTR expression) as a bone metastasis or brain metastasis, misinterpretation of accessory spleens, intrapancreatic spleen, or splenosis in operated or trauma patients as metastases of a NET, and classifying the mild to moderate uptake seen in inflammatory processes due to activated white blood cells as metastatic disease, in postsurgical scar tissue, or inflammatory or reactive lymph nodes, or in facet joint osteoarthritis, for example.

Several developments are currently ongoing to optimize SSTR PET imaging. A relatively recent paradigm shift is coming from radiolabeled antagonists that allow achieving higher uptake than with all previously mentioned ligands, which are agonists [55, 56]. Because some antagonist analogues are independent of the conformational status of SSTR, they can bind to a much higher fraction of the receptors present than agonists, which only bind certain specific conformations. This could lead to the emergence of second-generation ⁶⁸Ga-DOTA-peptides with even higher uptake ratios and potentially higher clinical detection rate. Another evolution is optimization of the labeling process. One major drawback of gallium-68-labeled peptides is the necessity to perform an on-site radiolabeling step, which requires specialized radiopharmaceutical personnel that is lacking in most nonacademic PET centers. Some groups are developing SSTR ligands labeled with radioisotopes that would allow centralized production and distribution to peripheral sites, using, e.g., copper-64 (T_1/2: 12.7 h) [57] or aluminum-fluoride-18 (T_1/2: 110 min) complexes that can be chelated and bound to octreotide derivatives [58]. The latter isotope is produced in curie quantities in cyclotrons throughout the world to produce fluorine-18 for ¹⁸F-FDG synthesis and would allow scanning 10–30 patients per production, compared to 1–4 for gallium-68-labeled peptides. For gallium-68, kits are currently being developed that would allow a more facile on-site labeling. It will be interesting to see which of these improvements will be the more adopted in clinical practice.

The cellular ability to take up, accumulate, and decarboxylate amines and amine precursors has been taken into account for the development of amino acid-based diagnostic radiotracers for nuclear medicine imaging. Accordingly, dihydroxyphenylalanine radiolabeled with fluorine-18 (¹⁸F-FDOPA) has been successfully proposed for in vivo nuclear imaging of NETs [59]. Once internalized via the sodium independent system L, ¹⁸F-FDOPA is decarboxylated to ¹⁸F-dopamine by the aromatic L-amino acid decarboxylase (AADC) enzyme, transported and stored into secretory vesicles. Therefore, the high uptake of ¹⁸F-FDOPA in NETs is the result of the cellular increased synthesis, storage, and secretion of biogenic amines [60, 61].