Analysis of [68Ga]NODAGA-exendin-4 PET data
Exendin-4 is a subcutaneously administered incretin peptide drug, used in the treatment of type 2 diabetes. [68Ga]NODAGA-exendin-4 is an high-affinity agonist of glucagon-like peptide 1 receptor (GLP-1R), and has been validated for imaging pancreatic β-cells in rats (Mikkola et al., 2014). [68Ga]NODAGA-exendin-4 can be produced with high specific activity, which is especially important for imaging GLP-1 receptors that are usually expressed in low levels; only very low mass of ligand can be administered, especially in small animals, to avoid pharmacological and receptor blocking effects (Brom et al., 2010; Mikkola et al., 2014; Nalin et al., 2014; Rydén et al., 2016; Mikkola et al., 2016).
Chelators are commonly conjugated to the free amine of a C-terminally attached lysine or cysteine of exendin-4. In addition, exendin-4 possesses naturally two lysines at positions 12 and 27, which could be used as conjugation sites, although that may affect the binding affinity, internalization, metabolism, and excretion rate (Jodal et al., 2014). NODAGA can chelate many radiometals, including 68Ga, 111In, 64Cu, and Al18F. When the two lysine positions (12 and 27) where used to attach NODAGA, and compared to ligand where NODAGA was attached to terminally added lysine (40), all ligands showed favourable kinetics and specific uptake (Jodal et al., 2014).
Exendin-4 labelled with 68Ga and other radiometals show high non-specific uptake in the kidneys, which leads to high radiation dose. Iodine-labelled exendin-4 ligands cause lower dose to kidneys, but radiation dose to thyroid may become significant. 18F-labelled exendin-4 tracers show lower radioactivity in the kidneys (Mikkola et al., 2016; Dialer et al., 2018). Although the absorbed dose to the kidneys is of concern in 68Ga-labelled exendin-4 studies, a dosimetry study has shown that repeated scans in humans with [68Ga]Ga-DO3A-VS-Cys40-exendin-4 are possible (Selvaraju et al., 2015), and automated GMP compliant production for this tracer has been developed (Velikyan et al., 2017).
High 68Ga activity in the kidneys hampers also image analysis in small animals; especially the pancreas is difficult to locate in the images (Mikkola et al., 2014). Bandara et al (2016) could identify pancreas in rats with the help of the CT (Inveon small animal PET/CT scanner); the tracers were 68Ga-labelled DO3A-VS-Cys40-exendin-4 and NODA-VS-Cys40-exendin-4.
[68Ga]NOTA-exendin-4 and [68Ga]DOTA-exendin-4 can be used to localize insulinomas (Luo et al., 2016; Antwi et al., 2018). Previously, [68Ga]Ga-DO3A-VS-Cys40-exendin-4 PET had detected a metastatic insulinoma (Eriksson et al., 2014), and [Lys40(Ahx-DOTA-68Ga)NH2]-exendin-4 had already shown promise in insulinoma imaging in a mouse model (Bauman et al., 2015). Sensitivity and specificity of preoperative imaging of insulin producing pancreatic neuroendocrine tumours with [68Ga]NODAGA-exendin-4 is being studied in a clinical trial.
Ex vivo autoradiography has shown that [68Ga]NODAGA-exendin-4 uptake is localized in macrophage-rich, GLP-1R-positive atherosclerotic lesions in the aorta of atherosclerotic and diabetic mice (Ståhle et al., 2017). PET imaging in myocardial infarction model in rats has shown that GLP-1R is upregulated during healing, and autoradiography confirmed that [68Ga]NODAGA-exendin-4 uptake correlated with the amount of CD68-positive macrophages in the infarcted region (Ståhle et al., 2018).
[68Ga]Ga-DO3A-VS-Cys40-exendin-4 PET study has shown striking variations among the species, caused by differences in β-cell mass and GLP-1R expression in β-cells and exocrine pancreas (Eriksson et al., 2017).
Internalization of GLP-1R tracers, even Exendin-4 tracers, is different between tracers (Jodal et al., 2014), and possibly also in different organs. In vitro, 35% of [Lys40(Ahx-DOTA-68Ga)NH2]-exendin-4 was internalized after 2 h (Jodal et al., 2014). In another cell line, ∼5% and ∼10% [Lys40(Ahx-DOTA-68Ga)NH2]-exendin-4 was internalized at 2 and 4 h, respectively (Wild et al., 2010). Radioligands can also be slowly released from the cells (Wild et al., 2010).
In cynomolgus monkeys, pancreatic [68Ga]Ga-DO3A-VS-Cys40-exendin-4 TACs were fitted to one-tissue compartment model to estimate the total volume of distribution (VT = K1/k2), fitting also blood volume (VB) as the third parameter (Selvaraju et al., 2013). The same model was later used to study diabetic (type 1 diabetes model) and non-diabetic pigs: VT was ∼2.3 and K1 ∼0.034 in the pancreas of both diabetic and non-diabetic pigs, although perfusion was decreased by almost 50% in diabetic pigs. Blocking with exendin-4 caused over 80% decrease in VT (Nalin et al., 2014). In occupancy study in cynomolgus monkeys, K1 decreased with increasing mass, from 0.07 to 0.01, while k2 remained relatively constant (Selvaraju et al., 2013). VT decreased from ∼8.4 to ∼0.3 at maximum ligand mass. VB estimates were in the range 0.15-0.45 (Selvaraju et al., 2013). In the liver, increasing ligand mass led to decrease in both K1 and k2, and changes in VT were negligible (Selvaraju et al., 2013).
[Nle14,Lys40(Ahx-NODAGA-68Ga)NH2]-exendin-4 ([68Ga]NODAGA-exendin-4) is rapidly cleared from the blood. Uptake in the heart is clearly lower than the radioactivity concentration in the blood even 1 h after administration (Mikkola et al., 2014), suggesting that input function can be derived from VOI placed on heart cavity, without any marked activity spill-over from the heart muscle.
Plasma vs blood
Exendin-4 is a relatively large peptide, and it, or its label-carrying metabolites do not pass the membranes of red blood cells, as indicated by the low activity in the blood cell pellet and the stable serum-to-blood ratio in a rat study with [64Cu]NODAGA-exendin-4 (Mikkola et al., 2014). Therefore the image-derived blood time-activity curve (TAC) is easy to convert to plasma TAC. Conversion can be based on haematocrit value, preferably measured individually.
In the rat study with [64Cu]NODAGA-exendin-4 (Mikkola et al., 2014) the median plasma-to-blood ratio was 1.793 (not published), which, if concentration in RBC is 0, means that haematocrit was ∼0.44.
Exendin-4 has an in vivo halflife of about 2.4 h.
[68Ga]NODAGA-exendin-4 has been shown to be stable in vitro and in vivo in rats: one hour after administration, 70±5% of the serum radioactivity was due to the intact radioligand (Mikkola et al., 2014; Ståhle et al., 2018). In pancreas, 32±5% of the 68Ga radioactivity represented intact tracer. Only minimal amounts of intact tracer were detected in the kidneys and urine (Mikkola et al., 2014). Two radiometabolites of [68Ga]NODAGA-exendin-4 can be detected in the urine and plasma (unpublished data). Power function can be used to fit the fractions of parent radioligand in plasma (Fig. 1); function parameters a=0.0452, b=1.5, c=0.218, and e=0.0 were determined from three rats, and parameter d (initial fraction at t=0) can be set to the measured radiochemical purity of each batch.
PET images of rats are dominated by the very high uptake in the kidneys and liver (Fig. 3 and 4). Tissue-to-blood ratio (Figure 5) and tissue-to-plasma ratio (Figure 6) is low, as expected in organs with low GLP-1R expression. The uptake in the liver is probably mostly due to radioactive metabolites.
It is difficult to determine from the shapes of the TTACs and tissue-to-plasma ratios whether the tracer kinetics is strictly reversible or if there is a tiny component of irreversible uptake. Model-independent graphical analyses, Logan plot and Patlak plot, are both linear (Figures 7 and 8), and thus do not help in deciding between reversible and irreversible model. Small part of the uptake in muscles and lungs may also be due to the radioactive metabolites. The very low slope in the Patlak plot demonstrates that the impact of active metabolites or irreversible uptake component during the 60-min study is small.
Reversible one-tissue and irreversible two-tissue compartmental models (1TCM and 2TCM, respectively) were fitted to the regional TTACs, both providing good fits (Figures 9 and 10). In skeletal and myocardial muscles the k3 was zero, suggesting that reversible models are appropriate. Estimates of the VB were reasonable, 5% for the skeletal muscle and 12% for the myocardium. The volume of distribution (VT), calculated as K1/k2, was 0.055 for the muscle and 0.31 for myocardium. The sum of VT and VB from the compartmental model fit is close to the VT computed using the Logan plot.
Although fits for the liver and lungs were good, the results are not expected to be physiologically valid, because of the metabolites in the liver, and because of the unknown air/tissue fraction in the lungs; additionally, the arterial input function is not fully valid for the lungs since it represents activity in plasma that has already circulated through lungs.
Kinetic analysis does not tell in what extent the calculated VTs represent specific binding of [68Ga]NODAGA-exendin-4 to the GLP-1 receptors, or non-specific binding to other targets. However, blocking studies have shown that specific binding component is substantial.
In conclusion, reversible 1TCM seems to be feasible method for assessment of VT. Logan plot gives overestimated values because of the relatively high contribution from the vascular activity, but could be used to assess relative changes in VT if simultaneous changes in VB are not expected.
In infarcted myocardial muscle the GLP-1R expression is increased during healing process. Internalization of [68Ga]NODAGA-exendin-4 leads to marked increase in the irreversible uptake component, allowing the usage of Patlak plot for the analysis (Ståhle et al., 2018).
Antwi K, Fani M, Nicolas G, Rottenburger C, Heye T, Reubi JC, Gloor B, Christ E, Wild D. Localization of hidden insulinomas with 68Ga-DOTA-Exendin-4 PET/CT: a pilot study. J Nucl Med. 2015; 56(7): 1075-1078.
Gao H, Kiesewetter DO, Zhang X, Huang X, Guo N, Lang L, Hida N, Wang H, Wang H, Cao F, Niu G, Chen X. PET of glucagonlike peptide receptor upregulation after myocardial ischemia or reperfusion injury. J Nucl Med. 2012; 53: 1960-1968.
Jodal A, Lankat-Buttgereit B, Brom M, Schibli R, Béhé M. A comparison of three 67/68Ga-labelled exendin-4 derivatives for β-cell imaging on the GLP-1 receptor: the influence of the conjugation site of NODAGA as chelator. EJNMMI Res. 2014; 4:31.
Mikkola K, Yim C-B, Fagerholm V, Ishizu T, Elomaa V-V, Rajander J, Jurttila J, Saanijoki T, Tolvanen T, Tirri M, Gourni E, Béhé M, Gotthardt M, Reubi JC, Mäcke H, Roivainen A, Solin O, Nuutila P. 64Cu- and 68Ga-labelled [Nle14,Lys40(Ahx-NODAGA)NH2]-exendin-4 for pancreatic beta cell imaging in rats. Mol Imaging Biol. 2014; 16: 255-263. doi: 10.1007/s11307-013-0691-2.
Mikkola K, Yim C-B, Lehtiniemi P, Kauhanen S, Tarkia M, Tolvanen T, Nuutila P, Solin O. Low kidney uptake of GLP-1R-targeting, beta cell-specific PET tracer, 18F-labeled [Nl114,Lys40]-exendin-4 analog, shows promise for clinical imaging. EJNMMI Res. 2016; 6:91. doi: 10.1186/s13550-016-0243-2.
Nalin L, Selvaraju RK, Velikyan I, Berglund M, Andréasson S, Wikstrand A, Rydén A, Lubberink M, Kandeel F, Nyman G, Korsgren O, Eriksson O, Jensen-Waern M. Positron emission tomography imaging of the glucagon-like peptide-1 receptor in healthy and streptozotocin-induced diabetic pigs. Eur J Nucl Med. 2014; 41: 1800-1810.
Selvaraju RK, Velikyan I, Johansson L, Wu Z, Todorov I, Shively J, Kandeel F, Korsgren O, Eriksson O. 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.
Updated at: 2018-12-18
Created at: 2018-08-15
Written by: Vesa Oikonen