Biodistribution and radiation dosimetry of [11C]choline: a comparison between rat and human data
- First Online:
- Cite this article as:
- Tolvanen, T., Yli-Kerttula, T., Ujula, T. et al. Eur J Nucl Med Mol Imaging (2010) 37: 874. doi:10.1007/s00259-009-1346-z
- 336 Views
Methyl-11C-choline ([11C]choline) is a radiopharmaceutical used for oncological PET studies. We investigated the biodistribution and biokinetics of [11C]choline and provide estimates of radiation doses in humans.
The distribution of [11C]choline was evaluated ex vivo in healthy rats (n=9) by measuring the radioactivity of excised organs, and in vivo in tumour-bearing rats (n=4) by PET. In addition to estimates of human radiation doses extrapolated from rat data, more accurate human radiation doses were calculated on the basis of PET imaging of patients with rheumatoid arthritis (n=6) primarily participating in a synovitis imaging project with [11C]choline. Dynamic data were acquired from the thorax and abdomen after injection of 423±11 MBq (mean±SD) of tracer. Following PET imaging, the radioactivity in voided urine was measured. The experimental human data were used for residence time estimations. Radiation doses were calculated with OLINDA/EXM.
In rats, the radioactivity distributed mainly to the kidneys, lungs, liver and adrenal gland. The effective dose in a human adult of about 70 kg was 0.0044 mSv/MBq, which is equivalent to 2.0 mSv from 460 MBq of [11C]choline PET. The highest absorbed doses in humans were 0.021 mGy/MBq in the kidneys, 0.020 mGy/MBq in the liver and 0.029 mGy/MBq in the pancreas. Only 2.0% of injected radioactivity was excreted in the urine during the 1.5 h after injection.
The absorbed radiation doses after administration of 460 MBq of [11C]choline were low. Except for the pancreas, biodistribution in the rat was in accordance with that in humans, but rat data may underestimate the effective dose, suggesting that clinical measurements are needed for a more detailed estimation. The observed effective doses suggest the feasibility of [11C]choline PET for human studies.
KeywordsPositron emission tomography[11C]CholineRadiation dosimetryWhole-body distribution
Malignant transformation is associated with the induction of choline kinase activity, which results in increased levels of intracellular phosphorylcholine, a key intermediate in the biosynthesis of phosphatidylcholine. Choline-based phospholipids are formed in the S phase of the cell cycle. Consequently, rapidly proliferating cells, such as tumour cells, contain large amounts of phospholipids, particularly phosphatidylcholine [1, 2]. Intravenously (i.v.) administered methyl-[11C]choline ([11C]choline) is taken into cells by specific transporters. In particular, there are separate carrier systems for choline uptake at the blood–brain barrier and at the cell membrane . Subsequently, it is phosphorylated to [11C]phosphorylcholine by choline kinase and trapped inside the cells . [11C]Choline PET has been used for the detection of various neoplasms, such as prostate, oesophageal and bladder carcinoma, and brain tumours [5–8]. Recently, [11C]choline PET has also been used for the imaging of synovial proliferation associated with rheumatoid arthritis [9, 10]. While [11C]choline PET is being used more widely in the assessment of tumours and proliferative disorders, it is important to consider the radiation safety of this particular radiopharmaceutical. Estimating the effective dose of a new agent is essentially mandatory when evaluating the risk-benefit ratio of medical radiation exposure. To date, only one review article reports an estimate of radiation exposure associated with injection of [11C]choline, which was based on measured biodistribution in one healthy subject and fragmented data obtained from ten other subjects . The study in question did not, however, systematically estimate the effective dose of [11C]choline.
The critical organs in humans for the use of a new radiopharmaceutical in the clinical setting are generally determined in preclinical studies in rodents and other mammalian species (dogs, nonhuman primates). Usually, these studies list the organs that receive the highest uptakes and compare the absorbed doses to the FDA guidelines 21 CFR 361.1. Findings from animal experiments should always be followed by a detailed study in humans for a reliable estimate of effective dose. However, animal-derived estimates have often been considered sufficient for the purpose of granting researchers permission for a human study. Our goal was to calculate the absorbed radiation doses for [11C]choline in humans using both animal data derived from preclinical studies and data from human PET measurements. We determined the biodistribution and critical organs in rats ex vivo and in vivo, and calculated the radiation dose estimates in humans. We compared radiation dose estimates derived from rats with those in humans calculated from PET measurements.
Materials and methods
Preparation of [11C]choline
[11C]Choline was synthesized from N,N-dimethylethanolamine and [11C]methyl triflate as described previously . An alternative method for synthesizing [11C]choline has previously been presented by Lehikoinen et al. . The radiochemical purity exceeded 99.2% and the average specific radioactivity was 75.5 GBq/μmol at the end of the synthesis (approximately 19 GBq/μmol at the time of injection).
Ex vivo biodistribution studies in normal rats
All animal studies were approved by the local University Laboratory Animal Committee. Nine Sprague-Dawley rats (Harlan Sprague-Dawley, Indianapolis, IN) weighing 289±40 g were used for the study of the biodistribution of [11C]choline ex vivo. The animals were allowed access to food and tap water ad libitum. All animals were anaesthetized with sodium pentothal intraperitoneally (Mebunat, Orion Oy, Espoo, Finland) and a purpose-made plastic cannula was inserted into the jugular vein for the injection of radiotracer. Each rat was given a bolus of 17±6 MBq of [11C]choline (<0.7 ml). Rats were killed in a CO2 box 5, 10 or 15 min after injection. Tissue samples were rapidly excised and blood collected, and the samples were weighed and measured for radioactivity in a well-type gamma counter (3×3-inch NaI(Tl) crystal; Bicron, Newbury, OH). All data were corrected for background radioactivity and radioactivity decay. The amount of radioactivity was expressed as percentage of injected dose per gram of tissue or blood (%ID/g). Measured organs/tissues were the adrenal gland, blood, bone + marrow, cerebellum, cortex, eyeball, mesenteric fat, heart, intestine, kidney, liver, lung, marrow, skeletal muscle, pancreas, ear skin, skull, spleen, stomach and urinary bladder.
In vivo PET imaging of [11C]choline distribution in tumour-bearing rats
Four male athymic Hsd:RH-rnu/rnu rats were obtained from Harlan, The Netherlands, at the age of 6 weeks, and allowed access to food and water ad libitum. Human BxPC-3 pancreatic adenocarcinoma cells (ATCC, Rockville, MD) were injected subcutaneously into the flank region (107/rat). Tumours were measured with external callipers every 2 or 3 days, and allowed to grow to 1 cm in diameter.
For dynamic PET imaging, the rats (weight 340±50 g) were anaesthetized by intraperitoneal injection of a mixture of midazolam-fluanisone-fentanyl citrate (Dormicum, Roche, Espoo, Finland; Hypnorm, Janssen Pharmaceutica, Beerse, Belgium). PET studies were performed with an HRRT camera (CPS Siemens, Knoxville, TN). The imaging field of view in the axial direction was 252 mm, which enabled whole-body imaging of a rat. All PET studies started with a 5-min transmission scan. Following the transmission scan, 33±14 MBq of [11C]choline was injected as an i.v. bolus via a tail vein. Dynamic imaging was started simultaneously with the injection. The acquisition frames were as follows: six at 60 s and ten at 300 s (total duration 60 min). All transaxial image slices were iteratively reconstructed using the three-dimensional ordered subsets expectation maximization (3D OSEM) algorithm. Image pixel size was 1.219×1.219 mm in a 256×256 matrix.
Regions of interest were drawn on major organs and on the tumours by two observers (T.U. and A.A.). Pharmacokinetic curves, representing the radioactivity concentrations vs. time after injection, were determined accordingly.
PET imaging of [11C]choline distribution in humans
Six patients with rheumatoid arthritis (four men, two women; age 60±10 years, weight 78±19 kg) were injected i.v. with 423±11 MBq of [11C]choline and imaged over the thorax or abdomen region to acquire dynamic data from the heart, lungs, liver, kidneys, spleen, pancreas, stomach, vertebral bodies, muscles and intestinal tissues. Tissues used for dosimetry were distant from the site of the inflammation, and the distribution of [11C]choline in the whole body was assumed to be undisturbed by the uptake in the inflamed joints.
The study protocol was approved by the joint Ethics Committee of the University of Turku and Turku University Hospital. Each subject provided informed consent before entering the study.
PET studies were performed with an Advance PET scanner (General Electric Medical Systems, Milwaukee, WI) operated in two-dimensional mode. The scanner has 18 rings of bismuth germanate (BGO) crystals, which form 35 imaging planes spaced by 4.25 mm. The imaging field of view in the axial direction is 152 mm. All PET studies started with a 9-min transmission scan. After transmission scanning, the tracer was injected and dynamic imaging over the thorax (n=3) or abdomen (n=3) was performed with the following time frames: eight at 15 s, six at 30 s, five at 180 s and four at 300 s. Images were reconstructed with the OSEM algorithm.
Whole-body PET scanning was performed on a female patient with rheumatoid arthritis (age 39 years, weight 83 kg) using a GE Advance PET scanner operated in two-dimensional mode. Imaging was performed with the patient in the supine position and arms alongside the body. Following 40-min dynamic imaging of the knee joint area and after i.v. bolus injection of 430 MBq of [11C]choline, whole-body PET proceeded from the head to the pelvic floor, excluding the legs. Six bed positions were required for this measurement, with a 5-min acquisition time for each position. The acquired data were iteratively reconstructed with attenuation correction using the OSEM algorithm.
Measurement of [11C]radioactivity in human urine and blood samples
Urine samples from 18 participants had already been collected during another [11C]choline PET study . Subjects were asked to void before the PET study. The total radioactivity in the whole urine volume was determined for the 1.5 h after tracer administration. Urine samples of 2.5 ml were measured with a well-type gamma counter (Bicron), cross-calibrated with the PET scanner via a dose calibrator (VDC-404; Veenstra Instruments, Joure, The Netherlands). The fraction of injected radioactivity excreted in the urine was calculated. The accumulated radioactivity in the urinary bladder was estimated by fitting the formula for exponential in-growth AB(1−e−bt) to the measured data, where AB represents the fraction of the total injected radioactivity excreted in the urine. We estimated the rate coefficient for clearance (b) by adjusting AB values between 0.7% and 2.0%. Physical decay was removed from the fitted curve, and the data were normalized to injection of 1 MBq. The area under the curve represents the accumulated radioactivity in the bladder from injection of 1 MBq. The average measured volume (239 ml) was used for calculation of the residence time of the urinary bladder contents.
Arterial blood samples from nine participants had already been obtained in another [11C]choline PET study . During the PET imaging session, approximately 20 sequential arterial blood samples were collected. At the beginning of the session, samples were taken more frequently to detect the peak of the blood radioactivity curve. Plasma was separated by centrifugation and the radioactivity was measured with an automatic gamma counter (Wizard 1480; Perkin Elmer, Turku, Finland), cross-calibrated with the PET scanner via the dose calibrator. The plasma time–radioactivity curves (TAC) were normalized to injection of 1 MBq before analysis. The data from the plasma curve were used to represent radioactivity in the heart contents.
Radiation dose calculations
Absorbed doses were calculated with OLINDA/EXM version 1.0 software (organ level internal dose assessment and exponential modeling; Vanderbilt University, Nashville, TN), which applies the MIRD schema (developed by the Medical Internal Radiation Dose Committee of the Society of Nuclear Medicine) for radiation dose calculations in internal exposure [14, 15]. The software includes [11C] radionuclide information and selection of human body phantoms. Standard phantoms are crude representations of the real situation in the body. The OLINDA/EXM software requires the number of disintegrations of [11C]choline in source organs as input. Disintegrations from human PET scans were derived from TACs. Regions of interest were first drawn in summed frames of the dynamic scan and later copied to the dynamic scan in order to get the TAC. The automated decay correction was removed and the curve was normalized to injection of 1 MBq and a 70-kg reference man. Each normalized TAC was fitted with exponential functions and extrapolated to infinity. The number of disintegration in all source organs was calculated as the integral under the measured and extrapolated curve multiplied by the organ volume of the reference man .
and processed to the number of disintegrations as described above.
All the results are expressed as means±SD. Correlations between the rat and human radiation dose estimates were calculated by linear regression.
Urinary excretion of [11C]choline was below 2% of injected radioactivity up to 1.5 h after injection. The rate of [11C]choline excretion to the urine was 0.014 ml min−1.
Number of disintegrations of [11C]choline for measured organs and the rest of the body (expressed as means±SD)
Upper large intestine
Urinary bladder contents
Rest of the body
Absorbed dose estimates of [11C]choline for target organs determined from rat or human PET imaging (expressed as means±SD)
Lower large intestine wall
Upper large intestine wall
Urinary bladder wall
Effective dose equivalent (mSv/MBq)
Effective dose (mSv/MBq)
Comparison of rat and human distribution data
Comparison of rat and human TACs for liver, kidney and muscle from in vivo PET imaging are shown in Fig. 2. Generally the uptake was lower in rats than in humans, the latter yielding a lower number of disintegrations and absorbed dose estimates (Tables 1 and 2).
[11C]Choline is widely used in routine PET examinations especially for imaging patients with prostate cancer. In this study we investigated the biodistribution, biokinetics and radiation dosimetry of [11C]choline in rats and humans. The observed effective doses using appropriate injected radioactivities were low in rats as well as in humans. Based on our measurements we confirm that [11C]choline can be safely used in clinical studies. Direct comparison of tracer biodistribution and dosimetry findings in humans and experimental animals are rare and this was a major strength of this study.
Comparison with previous studies and choline analogues
In the current study the overall distribution of [11C]choline was determined in rats ex vivo. The primary sites of radioactivity uptake in rats were the kidneys, lungs, adrenal gland and liver. This is in accordance with the findings of previous animal studies evaluating the biodistribution of radiolabelled choline [18, 19]. A high uptake of [11C]choline in rat kidneys, liver, pancreas, small intestine contents and salivary gland has been reported by Hara . These organs were clearly visualized in our rat PET images (data not shown).
Previously Hara has reported preliminary results of [11C]choline dosimetry in a review article . To our knowledge, we are the first to present comprehensive [11C]choline dosimetry in both rats and humans. The preliminary human study by Hara  adopted the MIRD method and found the highest absorbed doses in the kidneys (0.01803 mGy/MBq), liver (0.01731 mGy/MBq) and pancreas (0.01330 mGy/MBq). These results are somewhat lower than ours obtained using OLINDA/EXM. However, the differences in dosimetry between this study and that of Hara are most likely due to different observed values of uptake and clearance of the radiopharmaceutical and not due to major differences in the dose conversion factors between the MIRD values and the OLINDA/EXM values. In the current study, the dose to the kidneys was overestimated because the whole organ was assumed to have a uniform radioactivity concentration. However, the radioactivity from [11C]choline accumulates in the renal cortex rather than the pelvis because of low urinary excretion. The study by Hara did not, however, provide any estimate of the effective dose of [11C]choline. Our data suggest an effective dose of 0.0044 mSv/MBq for diagnostic [11C]choline PET scans.
Pharmacokinetics and radiation dosimetry of [18F]fluorocholine have been studied by DeGrado et al. . The dose-critical organ for [18F]fluorocholine is the kidney which receives 0.17 and 0.16 mGy/MBq in females and males, respectively. This limits the administration of [18F]fluorocholine to 4.07 MBq/kg resulting in an effective dose (whole-body) of approximately 0.01 Sv. Consequently, the effective dose of [18F]fluorocholine is 0.0225 mSv/MBq for females and 0.0178 mSv/MBq for males. The higher radiation dose of [18F]fluorocholine compared to that of [11C]choline (β+ decay 100%, Eβ+max 0.96 MeV, T1/2=20 min) is due to the longer half-life of the [18F] isotope (β+ decay 97%, Eβ+max 0.64 MeV, T1/2 110 min). In addition, [18F]fluorocholine does not undergo similar metabolism through the oxidative pathway and therefore does not exhibit a clearance pattern similar to that of [11C]choline .
Comparison with other [11C] radiopharmaceuticals
Effective doses of [11C] radiopharmaceuticals and [18F]FDG in humans
Effective dose (mSv/MBq)
Discrepancies between species
The absorbed doses and effective dose from the in vivo studies in rats were lower than those from the human studies because of the faster physiology of rodents, for example, in the kidneys. The kidney TAC is faster in rats than in humans. The muscle has low flow and normalized TACs in rats and humans are equal. This illustrates that blood flow has a remarkable influence on the curve shape, the area under the curve, and the measured number of disintegrations. While radioactivity concentration in blood is high immediately after injection, it is important to study the uptake directly after injection and continue to the plateau of the TAC.
The human effective dose was underestimated by 36% using the rat data, and exposure of individual organs to radiation was both over- and underestimated (Table 1). The dose to the pancreas was significantly (sixfold) lower in rats than in humans. For all other organs, the radiation dose estimates in rats correlated significantly with those in humans (Fig. 4; r=0.955).
Comparison of preclinical studies with human effective dose estimates
Different scaling methods
In this study, scaling between rat and human data was performed using the overall non-organ-specific weight. In general, interspecies extrapolation of biokinetic data is based on the fact that cellular structure and biochemistry are remarkably alike across the entire animal kingdom. Despite these similarities, extrapolation of biokinetic data from laboratory animals to humans is uncertain, particularly for the liver due to qualitative differences among species in the handling of many elements by this organ.
Allometric scaling from laboratory animals to humans on the basis of body weight or surface area is the most commonly used method. It is based on the assumption that the biokinetics of compounds depends primarily on the metabolic rate of the animal, and that the metabolic rate is a function of the body weight or body surface area of the animal. Yet, several other methods of scaling have been proposed , for example based on modelling of pharmacokinetic parameters where variation of serum protein binding between species is taken into account.
Ex vivo biodistribution studies in normal rats were performed at 5, 10 and 15 min after [11C]choline injection. However, for reliable dose estimation from rat-derived data, later measurements would have been helpful.
We failed to define the adrenal gland as a region of interest in PET images without an accurate anatomical reference image, and consequently, the adrenal gland was not defined as a source organ for in vivo human measurements. While the adrenal gland was considered only as a target organ, its self-radiation absorption dose was ignored. This resulted in an underestimation of the adrenal gland radiation dose. The radioactivity in the salivary glands, which have been noted as a relevant source region in human studies, was not considered in dose calculations because these organs are not included in OLINDA/EXM software. The absorbed dose to the lungs was also lower in the in vivo measurements, mostly because of technical differences between the ex vivo and in vivo distribution measurements. Lung tissue has a higher density ex vivo than in vivo, leading to an overestimation of the absorbed dose in the ex vivo measurements.
In conclusion, the effective dose of [11C]choline PET was found to be in accordance with those of other [11C]-labelled tracers fully justifying clinical studies within the range of 400–500 MBq of injected radioactivity. The preclinical distribution measurements with [11C]choline were in accordance with human PET measurements except for the pancreas in which tracer uptake was approximately sixfold higher in humans than in rats. Furthermore, the effective dose based on the preclinical estimate was 36% of the more accurate human-derived effective dose of 0.0044 mSv/MBq. Extrapolation of rodent data to humans is possible. In general, human and experimental data comparisons are rare and can be considered as the major strength of this study.
We thank the medical laboratory technologists and radiographers of the Turku PET Centre for their professional assistance and cooperation. We acknowledge Maija-Liisa Hoffren for excellent assistance with the animal studies. This study was funded by grants awarded by the Academy of Finland (no. 119048) and the Hospital District of Southwest Finland (no. EVO13856). This study was approved by the joint Ethics Committee of the University of Turku and Turku University Hospital, and the University Laboratory Animal Committee. All experiments were in compliance with Finnish law.