[68Ga]NOTA-Galactosyl Human Serum Albumin: a Tracer for Liver Function Imaging with Improved Stability

Purpose Non-invasive techniques allowing quantitative determination of the functional liver mass are of great interest for patient management in a variety of clinical settings. Recently, we presented [68Ga]DTPA-GSA to target the hepatic asialoglycoprotein receptor for this purpose. Here, we introduce [68Ga]NOTA-GSA to improve metabolic stability of the radiopharmaceutical and compare the imaging properties with [68Ga]DTPA-GSA. Procedures Labeling of the compounds was carried out at room temperature using 1.9 M sodium acetate as buffer. For quality control, thin-layer, high-performance liquid, and size exclusion chromatographies were used. Metabolic stability was studied in rat and human serums. For in vivo evaluation, Fischer rats were scanned by positron emission tomography and magnetic resonance imaging and subsequently sacrificed for biodistribution studies. Time activity curves (TACs) for heart and liver were generated and corresponding parameters (T50, T90, LHL15, HH15) were calculated. Results [68Ga]NOTA-GSA can be produced in high radiochemical yield and purity (>95 %) within 15 min. Stability studies revealed almost no metabolite formation over the 2-h observation period. Analysis of the TACs showed comparable results for most of the investigated parameters. The only significant difference was found in the T90 value, where [68Ga]NOTA-GSA showed slower uptake in comparison with 68Ga-DTPA-GSA (123 ± 10 vs. 89 ± 3 s, p < 0.01). Conclusions [68Ga]NOTA-GSA showed a significant increase of the metabolic stability and in most organs lower background activity. However, comparison of LHL15 and HH15 indicates that the increased stability did not further improve the diagnostic value. Thus, [68Ga]NOTA-GSA and [68Ga]DTPA-GSA can be used equivalent for imaging hepatic function with positron emission tomography.

On the other hand, non-invasive methods allowing quantitative determination of the functional liver mass are of great interest for patient management in a diversity of clinical settings comprising liver surgery and liver transplantation [4][5][6] as well as diagnosis [7,8] and therapy monitoring [9] of cancer. Additionally, it has been revealed that the evaluation of remnant liver function can help to discriminate different stages of alcoholic liver cirrhosis [10] and could be used to differentiate areas of steatosis, fibrosis, and cholestasis [11]. Moreover, control of liver status before and during peptide receptor radionuclide therapy (PRRT) [12] could lead to an optimized patient management. Furthermore, patients who are potentially suitable for selective internal radiation therapy (SIRT) [13] may also benefit from such a diagnostic method, allowing stratifying patients according to their peri-interventional risk.
[ 99m Tc]diethylenetriamine-pentaacetic acid galactosyl human serum albumin ([ 99m Tc]GSA) [14] binds to the asialoglycoprotein receptor (ASGP-R), a hepatic cell surface receptor specific for galactose terminated glycoproteins [4]. This receptor is exclusively expressed on the sinusoidal surface of mammalian hepatocytes. It has been proven that [ 99m Tc]GSA and dynamic single-photon emission computed tomography (SPECT) allows estimation of regional hepatic function based on the determination of the ASGP-R density [15,16].
To combine the superior performance of PET compared with SPECT concerning imaging resolution and quantifying properties with the good properties of GSA in ASGP-R targeting, we developed a Ga-68-labeled analog [17].

Material and Methods
All reagents were obtained from VWR International GmbH (Vienna, Austria) or Sigma-Aldrich Handels GmbH (Vienna, Austria) and were used without further purification. DTPA-conjugated galactosyl human serum albumin (GSA) was a kind gift from Nihon Medi-Physics (Tokyo, Japan) and was supplied as a freeze-dried technetium kit formulation. The kit was reconstituted with water and directly used for 68 Ga labeling without isolation of the glycoprotein. The mean number of DTPA groups per human serum albumin (HSA) is 6 and the mean number of galactosyl units is 35 resulting in an estimated MW of 80,730 g/mol (for details, see [17]). NOTA-GSA was supplied from piCHEM (Graz, Austria) and includes six NOTA, conjugated via isothiocyanate chemistry and 19 galactosyl moieties per HSA molecule resulting in an estimated MW of 76,398 g/mol. The Ge-68/Ga-68 generator was purchased from Eckert & Ziegler (Berlin, Germany) with a nominal activity of 1100 MBq and was eluted with 0.1 N HCl (Biochemical grade, FLUKA, Switzerland).
Additionally, size exclusion chromatography (SEC) was carried out to control for potential colloid formation. Therefore, Sephadex G-25 PD-10 columns (GE Healthcare Europe GmbH, Vienna, Austria) were incubated with 15 ml bovine serum albumin solution (BSA, 1 % in isotonic saline), loaded with 0.2 ml reaction mixture, and washed with 7 ml isotonic saline. Eluate and column were measured in a dose calibrator.

Stability of [ 68 Ga]DTPA-GSA and [ 68 Ga]NOTA-GSA
The metabolic stability in human serum was determined after 2, 30, 60, and 120 min incubation of [ 68 Ga]NOTA-GSA at 37°C. Hundred microliter [ 68 Ga]NOTA-GSA labeling solution per 1 ml serum were used for incubation. For HPLC analysis, 400 μl samples of the human serum were passed through a 0.20 μm sterile filter (Millex LG, EMD Millipore, Bellerica, MA; activity remaining on the sterile filter was between 1 and 2 %) and 20 μl aliquots were injected to the HPLC (gradient A). Stability assay for [ 68 Ga]DTPA-GSA (gradient B) in human serum was carried out following the same protocol and is described in [17]. Assays were carried out in duplicate.
The metabolic stability of [ 68 Ga]NOTA-GSA and [ 68 Ga]DTPA-GSA in rat serum was determined after 2, 30, and 60 min at 37°C. Therefore, 20 μl of the corresponding tracer was added to 200 μl of serum. At the corresponding time points, 20 μl aliquots were removed and diluted with 80 μl water, passed through a 0.20 μm sterile filter, and the filter was washed with 100 ml water. Eighty microliter samples were analyzed using HPLC and gradient A. Assays were carried out in duplicate.
ACN  For biodistribution studies, blood samples were taken and animals were sacrificed immediately after imaging studies were finished. Organs (liver, lung, kidneys, spleen, heart, gut, stomach, and brain) as well as muscle were excised. The samples were weighed and activity in the samples was measured using a γ-counter (Wizard single-detector-counter; PerkinElmer, Waltham, MA, USA). Results are expressed as the percentage injected dose per gram of tissue (%ID/g). Each value represents the mean and SD of four animals. All animal experiments were approved by the German competent authorities (Regierungspräsidium Tübingen; R12/13). For analysis, the images of PET and MRI were fused and regions of interest were drawn on the basis of the MRI, covering the entire liver and the left ventricle for the cardiac input curve using Inveon Research Workplace (IRW, Siemens Healthcare). Time activity curves (TACs) were analyzed and characterizing parameters were calculated using MatLab (Mathworks, Natick, MA, USA). Briefly, a second-degree polynomial was fitted into the five points around the maximal value of the cardiac input function, defining t 0 for the experiment. Subsequently, a decreasing exponential function y = A ⋅ exp (−b ⋅ x) + c was fitted into the cardiac TAC, starting at the peak value of the polynomial. Liver uptake was also characterized by an exponential function, but was fitted over the entire dataset. Based on these functions, characteristic parameters for blood clearance (T 50 = time to reach 50 % of the maximum heart uptake; HH15 = blood activity at 15 min divided by the blood activity at 3 min; decaying constant b of the fit) and functional liver reserve (T 90 = time to reach 90 % of the maximum liver uptake; LHL15 = liver activity at 15 min

Synthesis of [ 68 Ga]NOTA-GSA
Labeling of NOTA-GSA (for proposed structure, see Fig. 1) with gallium-68 could be carried out in radiochemical yield and purity 995 % determined by HPLC as well as TLC. Additional analysis of potential colloid formation using SEC demonstrated that the formulation contains 997 % [ 68 Ga]NOTA-GSA and indicating that if at all only low amounts of colloid had been formed. The specific activities of the studied preparations were in the range of 38-77 MBq per nmol NOTA-GSA. Due to the fact that no separation of the unlabelled compound is possible, the specific activity strongly depends on the amount of precursor used.

Stability Assays
Incubation of [ 68 Ga]NOTA-GSA with human serum up to 120 min demonstrated high metabolic stability of the tracer (Fig. 2). Over the whole observation period of 120 min, more than 98 % intact tracer was found if samples were analyzed via HPLC. In contrast, rapid degradation was found for 68 Ga-DTPA-GSA. Already after 30-min incubation, only approx. 70 % intact tracer was found. After 120 min, the amount of intact tracer was further reduced to 30 %. Due to the low retention time, it can be assumed that the observed Bdegradation product^is mainly due to release of the radiometal from the chelating system.
The same behavior is found when analyzing the metabolic stability in rat serum. Again, [ 68 Ga]NOTA-GSA was stable over the observation period of 60 min (imaging data where recorded not longer than 30 min; thus for the rat study, observation time was reduced) whereas a significant reduction of intact [ 68 Ga]DTPA-GSA was observed. After 1-

Comment
Due to the physiological role of the asialoglycoprotein receptor, which transfers the glycoproteins into liver cells where they are degraded stability analysis in the liver would not supply evaluable information about the stability of the chelating systems and were thus not included.

Imaging and Biodistribution Data
Both tracers showed comparable uptake patterns, i.e., a very fast accumulation throughout the entire liver with low background activity in all other organs and a very fast blood clearance (Fig. 3). In direct comparison, [ 68 Ga]DTPA-GSA reached a slightly higher liver uptake level, but also showed a slow decrease over time (Fig. 3a). Mean TACs generated at the left ventricle were very similar, showing a very fast decrease to baseline in the first 5 min with [ 68 Ga]NOTA-GSA reaching a lower final level (Fig. 3c). The biodistribution confirmed the in vivo results, showing only relevant uptake in liver tissue, while all other organs remained below 0.5 % ID/g tissue (Fig. 4a). Indepth analysis of the TACs, based on exponential fits, also showed comparable results for most of the investigated parameters. Overall fit-quality was excellent (R 2 for liver TAC is 0.99 ± 0.01 and for blood TAC 0.94 ± 0.03 averaged over all animals). Most parameters were comparable ( Table 1). The only significant difference was found in the T 90 value, where [ 68 Ga]NOTA-GSA showed slower  Table 1. Characteristic parameters for blood clearance (T 50 = time to reach 50 % of the maximum heart uptake, HH15 = blood activity at 15 min divided by the blood activity at 3 min, and decaying constant b of the fit) and functional liver reserve (T 90 = time to reach 90 % of the maximum liver uptake and LHL15 = liver activity at 15 min divided by the sum of liver and blood activity at 15 min). Data are given as mean ± standard deviation

Discussion
[ 99m Tc]GSA and dynamic SPECT including data analysis using corresponding kinetic models [19] has successfully been demonstrated to be able to determine non-invasively regional hepatic function in a great number of patients especially in Japan [14]. Recently, we introduced the Ga-68labeled analog [ 68 Ga]DTPA-GSA and showed that the technetium kit formulation can directly be labeled with Ga-68 [17]. Moreover, the new tracer showed comparable liver uptake as found for the 99m Tc-analog. However, the stability in human serum was inferior to [ 99m Tc]GSA. This might be due to the use of DTPA as gallium chelating system. In this study, we replaced the DTPA moiety by a NOTA derivative and could demonstrate that this modification resulted in a much more stable GSA derivative ready for imaging liver function in clinical settings. Initially, DTPA-GSA was designed to be labeled with Tc-99 m [20], for which DTPA is a well-established chelator. It is known that DTPA possesses also a high complex stability constant for Ga(III) [21]. Thus, it could be assumed that the corresponding [ 68 Ga]DTPA-GSA complexes should be sufficiently stable for non-invasive imaging of the hepatic function. However, in our previous study, we found significant degradation in human serum within 30 min [17]. The low stability of 68 Ga-DTPA complexes is confirmed by Anderson and Strand [22] who studied the stability of a Ga-67-labeled monoclonal antibody and found also low stability of the complex. To improve the metabolic stability of the tracer, we replaced the DTPA moiety by a NOTA derivative. It is known that the nine-membered ring structure of NOTA forms a cage with an optimal size for complexing Ga(III), resulting in a very high complex stability constant [21].
Synthesis was carried out to produce the labeling precursor with a similar average amount of chelating moieties as found for the DTPA-GSA. Subsequent labeling of NOTA-GSA was straightforward. Within 15 min at room temperature, [ 68 Ga]NOTA-GSA can be produced in high radiochemical purity and yield. No subsequent purification is required. However, it has to be mentioned that fixing of the product, independent if NOTA-GSA or DTPA-GSA is used, on a C-18 cartridge is not possible. Thus, it is recommended to use a pre-purification procedure for labeling to guarantee that no germanium breakthrough can reach any patient preparation in a clinical setting.
Corresponding stability assays carried out either in human or in rat serum both clearly demonstrated the desired improvement. [ 68 Ga]NOTA-GSA showed almost no release of the radiometal over the observation period of 2 h in human serum. Even the clinically established [ 99m Tc]DTPA-GSA revealed, over the whole observation period, more degradation as found for [ 68 Ga]NOTA-GSA [17]. However, the imaging studies comparing the activity distribution and pharmacokinetics of both 68 Ga-labeled compounds reveal, at first glance, only small differences in tracer uptake in the liver and in the elimination of the tracer from the body. Anyway, a more detailed analysis revealed that for [ 68 Ga]NOTA-GSA activity concentration in most organs beside the liver is even lower as the already low values found for [ 68 Ga]DTPA-GSA (e.g., activity concentration in blood for [ 68 Ga]NOTA-GSA is less than half of the concentration found for [ 68 Ga]DTPA-GSA). Additionally, TACs show a decrease in the liver uptake of [ 68 Ga]DTPA-GSA at later time points, whereas it remains constant for [ 68 Ga]NOTA-GSA. All this might be attributed to the lower stability of [ 68 Ga]DTPA-GSA. However, determination of the relevant parameter based on the TACs resulted in comparable values. An exception is the T 90 value. Due to the fact that the initial uptake of [ 68 Ga]DTPA-GSA is faster and also slightly higher than found for 68 Ga-NOTA-GSA, this value is smaller for [ 68 Ga]DTPA-GSA. It seems that the higher stability due to the introduced NOTA system reduces the activity concentration in blood and may slow down the tracer kinetics. As the T 90 time may be more sensitive to physiological parameters such as, e.g., blood flow, general cardiac output, or depth of anesthesia, a relevant influence on the determination of the hepatic function is not expected based on this slight difference, especially since the LHL15 values are almost the same. Therefore, if using the entire dataset to calculate functional liver reserve (LHL15) and blood clearance (HH15), both tracers yield identical results. Thus, the difference in metabolic stability does not significantly affect the final outcome. As the parameters LHL15 and HH15 were developed in a clinical setting using SPECT imaging [23], they may not be the best suitable parameters to characterize the fast metabolism of rats or mice. A recent study by Schnabl et al. investigated the functional liver reserve in an animal model of liver disease using T 90 values to characterize [ 68 Ga]DTPA-GSA uptake [24]. Here, we see small differences in the absolute values; however, their influence on the sensitivity has to be investigated in studies with corresponding disease models.

Conclusion
[ 68 Ga]NOTA-GSA can be easily produced in high radiochemical purity and yield within 15-min reaction time. The introduction of the NOTA moiety resulted in a significant increase of the metabolic stability and in most organs in lower background activity. However, comparison of LHL15 and HH15 indicates that the increased stability did not further improve the diagnostic value. Thus, [ 68 Ga]NOTA-GSA and [ 68 Ga]DTPA-GSA can be used as an equivalent for imaging hepatic function with PET.