Study participants
The study included 6 healthy volunteers (3 men and 3 women) with a mean age of 56.3 years (range, 52–67 years) and a mean BMI of 27.7 kg/m2 (range, 24.6–30.9 kg/m2). All participants were screened for renal and cardiac diseases prior to inclusion and signed an informed consent for participation and publication, before being included in the study. The study site was The Department of Nuclear Medicine & PET Centre at Aarhus University Hospital.
Radiopharmaceutical preparation
The (R)-[1-11C]β-hydroxybutyrate was produced as previously reported with some modifications (21). [11C]cyanide was produced from [11C]CO2 in a process cabinet (GE Healthcare) and delivered as NH4[11C]CN to an in-house built synthesis module. Initially, NH4[11C]CN (10–20 GBq) was trapped on a “cyanide trap” as previously reported (22), to facilitate the removal of ammonia, the presence of which would otherwise lead to hampering of the following hydrolysis step and formation of side products. Following elution of K[11C]CN from the “cyanide trap” to the reaction vessel with 400 µL sterile H2O, 400 µL of the precursor, (R)-propylene oxide, was added and the mixture was heated at 40oC for 4 minutes in which time approximately 90% conversion to (R)-[1-11C]β-hydroxybutyronitrile was obtained. Next 2.4 ml 12 M hydrochloric acid was added to reach a final molarity of > 10 M of the hydrolysis mixture, which was subsequently heated at 150oC for 4 minutes to achieve a conversion average of 50% (n = 19) from the intermediate (R)-[1-11C]β-hydroxybutyronitrile to the product (R)-[1-11C]β-hydroxybutyrate (Fig. 1). The crude acidic product mixture was purified by semi-preparative HPLC (PolymerX 10x250 mm, 10 µm, Phenomenex, Torrance) and the product fraction (~ 3 ml) was collected and passed through a 0.22 µm sterile filter and formulated in 7 ml sterile saline. (R)-[1-11C]β-hydroxybutyrate have been produced in radiochemical yields up to 2.8 GBq with decay corrected yields of up to 50%, radiochemical purities was > 97% and enantiomeric purities larger than 98%.
Whole-Body PET Imaging
Intravenous study. Dynamic whole-body (D-WB) PET with continuous-bed-motion was performed using a Biograph Vision 600 PET/CT system (Siemens) with 26.3 cm axial field of view. A low-dose CT (25 Ref mAs, 120 Ref. kV, CARE Dose4D, CARE kV, ADMIRE level 3) from vertex cranii to mid-thigh was acquired first. A mean bolus of 204.2 MBq (range, 182–222 MBq) of [11C]OHB was injected, and immediately afterwards 24 consecutive whole-body passes with increasing durations (to account for the short half-life of [11C]OHB) were obtained. The 24 passes had the following structure: 5 · 60 s, 5 · 120 s, 5 · 180 s, 6 · 300 s, and 3 · 600 s. The time intervals for the PET scans were therefore approximately 1–5, 6–15, 16–30, 31–60, and 61–90 min. The PET images were reconstructed using TrueX + TOF, 4 iterations and 5 subsets, 2-mm Gaussian post filter, and 440 × 400 matrix with voxel size 1.653 mm3.
Oral study. An oral study was performed 120 min after the intravenous study using the same PET scanner. Participants rapidly ingested a mean oral dose of 96.11 MBq (range, 59.15–114.5 MBq) of [11C]OHB dissolved in water. A low-dose CT (as in the intravenous study) was performed, and 5 min after ingestion of the radiotracer, 23 consecutive whole-body continuous bed motion scans with increasing durations were obtained. The 23 scans had the following structure: 5 · 120 s, 5 · 180 s, 7 · 300, and 6 · 600 s. The time intervals were therefore approximately 6–15, 16–30, 31–65, and 66–125 min.
Blood sampling and metabolite correction
During the D-WB PET scans, manual blood samples were taken to measure arterial radioactivity for the input function. Both plasma and whole-blood radioactivity were measured in a well counter (Hidex AMG). We collected 40 manual blood samples during the intravenous studies and 23 manual blood samples during the oral studies. In addition, 8 manual venous blood samples were taken to correct for conversion of [11C]OHB to [11C]CO2, the only quantitatively important metabolite that has to be taken into account during ketone PET studies of < 120 minutes.
To determine [11C]CO2 in blood, two 0.5 mL venous whole blood samples were drawn simultaneously during the intravenous study (5, 10, 15, 20, 30, 50, 70, and 90 minutes after administration of [11C]-OHB). They were then placed in tubes containing 1.5 mL of isopropyl alcohol and 0.5 mL of 0.9 M sodium bicarbonate. Sample 1 was treated with 0.5 mL of 0.1 N sodium hydroxide and sample 2 was mixed with 0.5 mL of 6 N hydrochloric acid and vigorously stirred with a magnet for 10 minutes at room temperature. The radioactivity of each sample was then measured in a well counter for 1 minute, and the data was corrected for decay to a common time point. Radioactivity in sample 2 divided by sample 1 was then considered the parent fraction. A correction curve was established by fitting the parent fraction curve using a monoexponential function and applied to the blood input function.
Defining organ and tissue volumes of interest on the PET images
All D-WB PET images were visually inspected in PMOD 4.3 (Zurich, Switzerland). For PET scans following i.v. injection, the source organs for the dosimetry calculations were the brain, salivary glands, heart, liver, spleen, kidneys, pancreas, red marrow, and urinary bladder contents. For PET scans following oral ingestions, the source organs for the dosimetry calculations were the stomach contents, small intestine, heart, liver, kidneys, and salivary glands.
Volumes of interest (VOIs) were defined in PMOD 4.3. A sphere or box was drawn as large as possible for each organ to encompass the entire organ. In difficult cases, the CT scan was used to aid the correct anatomical placement of the sphere/box. Thresholding was subsequently used to segment the organs within the bounding sphere/box. For the heart (left ventricle), pancreas (cauda pancreatis), red marrow (one lumbar vertebra), and skeletal muscle (musculus vastus lateralis) a smaller VOI was drawn within the organ. The VOI representing the brain was drawn by utilizing the full brain atlas supplied by PMOD.
Biodistribution and Dosimetry
For each source organ, the time course of the non-decay-corrected total radioactivity was normalized to the administered activity and recalculated to time courses of percentage injected activity. Time-integrated activity coefficients (TIACs) were computed using the trapezoidal integration method to calculate the area under the curves, assuming only physical decay after the last scan without further biological clearance. For oral administrations, we assumed that 100% of the tracer was in the stomach contents at the scan start. The remainder TIAC was calculated by subtracting the individual source organ TIACs from the total body TIAC (without voiding), which for 11C is 0.49 h. TIACs for source organs and remainder were used in OLINDA/EXM 2.0 (HERMES Medical Solution AB, Sweden) (23) to compute organ absorbed doses (µGy/MBq) and the effective dose (µSv/MBq) using anthropomorphic human body phantoms with organ masses based on ICRP89 (Basic anatomical and physiological data for use in radiological protection: reference values). A report of age- and gender-related differences in the anatomical and physiological characteristics of reference individuals. ICRP Publication 89 (24) and ICRP103 tissue weighting factors (25). Organ doses and effective dose results are given for the reference gender-averaged adult according to ICRP103.
Kinetic analyses
All kinetic analyses were done using PMOD 4.3 software. Arterial plasma input functions were corrected for metabolites. To test whether organ and tissue radiotracer kinetics were best analyzed using a reversible or an irreversible model, we initially analyzed tissue data excluding the first 30 min using a 1-tissue-compartment model with k2 fixed at 0 or k2 estimated by the model. In all cases the Akaike Information Criterion (AIC) favored a reversible model i.e., k2 > 0 (Table 1). For all organs and tissues, we compared a 1-tissue with a 2-tissue compartment model using fixed estimates of tissue blood volume fraction (vB). The 2-tissue compartment model yielded the best AIC for all organs and tissues (except for the brain) and we therefore proceeded to perform full kinetic analysis using this model assuming reversible kinetics. Since we were limited by the relatively small axial field-of-view of the scanner, and utilized continuous-bed-motion to cover the whole body, we were precluded from obtaining sufficiently detailed radiotracer first pass data in all organs and tissues. Consequently, we opted to use a fixed tissue vB according to published data for the heart (26), brain (27), kidneys (28), liver (29), pancreas (30), red marrow (31), and skeletal muscle (32). Since no data exists on the vB in salivary glands, it was estimated to be 5%. In addition to compartment modelling, we estimated brain radiotracer uptake using the initial 10 minutes and assuming irreversible uptake to be able to compare our results with those previously obtained by other groups using [11C]-acetoacetate in particular (33, 34). Since all tissue radiotracer kinetics were considered reversible, we used also used a linear non-compartmental approach (Logan) to estimate volume of distribution (Vt).
Table 1
Test for reversibility using AIC
Target organ | AIC 1T | AIC 1T (k2 = 0) |
Heart | -22.89\(\pm\)8.68 | 32.26\(\pm\)4.92 |
Liver | -10.22\(\pm\)30.34 | 21.51\(\pm\)18.19 |
Pancreas | -1.11\(\pm\)4.96 | 22.38\(\pm\)5.50 |
Kidneys | -4.45\(\pm\)9.82 | 22.80\(\pm\)2.34 |
Salivary glands | -5.80\(\pm\)4.03 | 33.26\(\pm\)9.51 |
Skeletal muscle | -0.03\(\pm\)16.87 | 23.72\(\pm\)6.23 |
Brain | -17.95\(\pm\)22.38 | 28.46\(\pm\)2.82 |
Red marrow | -2.12\(\pm\)3.47 | 29.55\(\pm\)2.37 |
Notes: Comparison of a better fit illustrated by AIC was done by excluding the first 30 min of tissue kinetics. Values are presented as mean \(\pm\) SD. Abbreviations: AIC = Akaike Information Criterion; 1T = 1-tissue-compartment model. |