Synthesis and Evaluation of ¹⁸F-Labeled Choline Analogs as Oncologic PET Tracers

Synthesis of Radiotracers

[methyl-¹⁴C]Choline (CH) was obtained from NEN Research Products (Boston, MA). No-carrier-added FCH was synthesized by reacting [¹⁸F]fluorobromomethane ([FBM] isolated by gas chromatography [GC]) with dimethylethanolamine as modified from DeGrado et al. (11). In an analogous fashion to the synthesis of FCH, [¹⁸F]FEC was synthesized through the intermediate 1-[¹⁸F]fluoro-2-bromoethane (FBE) by radiofluorination of 1,2-dibromoethane rather than dibromomethane. For isolation of FBE, the temperature of the preparative GC column (Porapak Q, 80–100 mesh, 7.8 × 700 mm; Alltech Associates Inc., Deerfield, IL) was maintained at 135°C. [¹⁸F]Fluoromethyl-methylethyl-2-hydroxyethylammonium (fluoromethylethylcholine [FMEC]) (Fig. 1) was synthesized by reaction of FBM with ethylmethylethanolamine (Pfaltz and Bauer, Waterbury CT) by the same procedure as for synthesis of FCH. [¹⁸F]Fluoropropyl-dimethyl-2-hydroxyethylammonium (fluoropropylcholine [FPC]) (Fig. 1) was produced through the intermediate [¹⁸F]fluorobromopropane (FBP), which was synthesized and purified by high-performance liquid chromatography (HPLC) as described (12). The syntheses of FBE and FBP were not optimized, resulting in poorer radiochemical yields of FEC (<3%) and FPC (<1%), respectively, than for FCH or FMEC (30%–40%).

The radiochemical purity of the ¹⁸F-labeled choline analogs was monitored by cation-exchange HPLC as described for measurement of CH metabolites (13). The column was Partisil SCX (250 × 4.6 mm; Alltech Associates) eluted by 0.25 mol/L sodium dihydrogen phosphate solution (pH 4.8) and acetonitrile (90:10) at a flow rate of 1.8 mL/min. Radioactivity and ultraviolet absorbance (206 nm) of the eluent were measured in series. The retention times were 5.0, 5.4, 5.5, and 6.0 min for FCH, FEC, FMEC, and FPC, respectively (Fig. 2). The cation-exchange HPLC system was found to be preferable to the reverse-phase HPLC system originally developed for quality assurance of FCH (11) because peak resolution was superior, and retention of radioactivity on the column was negligible.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Cation-exchange HPLC radiochromatograms of FCH (A), FEC (B), and FPC (C) final products. UV = ultraviolet.

Two modifications were implemented to the FCH synthesis procedure since its original publication (11). To ensure removal of trace quantities of dimethylethanolamine from the product, the transfer of material from the fluoroalkylation reaction vessel to the cation-exchange Sep-Pak cartridge (ACCEL Light CM; Waters, Milford, MA) was performed using ethanol instead of water, and the Sep-Pak cartridge was washed with an additional 5 mL ethanol. The subsequent washing of the cartridge with 10 mL water and elution of FCH from the cartridge with isotonic saline were performed as described (11). The second modification was the introduction of a cleansing procedure for the GC column used to purify FBM or FBE. After each run, 20 mL ethanol were back flushed through the GC column followed by 2 h of back purging of the column with helium. This procedure ensures that remains of dibromomethane and its decomposition products do not build up on the GC column and eventually contaminate the FBM or FBE product.

Stability of FCH

The stability of FCH in its prepared form was evaluated by monitoring the radiochemical purity using the HPLC system described above. Furthermore, the stability of FCH in blood was examined by incubating approximately 3.7 MBq FCH in a 3-mL sample of heparinized whole blood taken from healthy human subjects (n = 5). After a 2-h incubation at 37°C, the plasma was separated, the plasma proteins were precipitated by adding 2 volumes of methanol, and the supernatant was analyzed for intact FCH by cation-exchange HPLC.

Analysis of Hydrophilic Choline or FCH Metabolites in Biologic Samples

The primary hydrophilic metabolites of choline in mammals are phosphocholine and betaine. If FCH is handled similarly biochemically to CH in cancer cells, then ¹⁸F-labeled phosphorylfluorocholine (P-FCH) and fluorobetaine (FB) may be formed. To measure the levels of FCH, FB, and P-FCH in biologic samples, a modification of the gradient-HPLC method of Pomfret et al. (14) was used. ¹⁴C-Labeled betaine and phosphocholine were prepared enzymatically from commercially available CH (NEN Research Products) using choline oxidase (13) and CK (15), respectively. The HPLC system used a microprocessor-controlled solvent delivery system and a silica column (Adsorbosphere silica [10 μ], 250 × 4.6 mm; Alltech Associates). The column was kept at room temperature and the flow rate was maintained at 1.5 mL/min. Buffer A contained acetonitrile, ethanol, acetic acid, 1.0 mol/L ammonium acetate, water, and 0.1 mol/L sodium phosphate (800:68:2:3:127:10 [v/v]) and buffer B contained the same constituents but in different proportions (400:68:44:88:400:10 [v/v]). Fractions of effluent were collected every 0.5 min and counted first for ¹⁸F radioactivity in a well counter for experiments involving FCH and then transferred to scintillation vials for counting ¹⁴C radioactivity. The column was equilibrated for 6 min with buffer A before injection. After the injection (<100 μL), buffer A was delivered for 6 min, which eluted betaine (3 min) from the column. Over a period of 10 min, solvents were switched to 100% buffer B using a linear gradient, during which time FCH (12.5 min) and choline (15 min) were eluted. Buffer B was then delivered for a further 9 min, eluting phosphocholine (19.5 min). Reequilibration of the column with 100% buffer A for 6 min preceded the next injection.

In Vitro Phosphorylation of FCH by Yeast CK

To determine whether FCH is a substrate for CK, FCH (0.9–1.8 MBq) and CH (74–150 kBq) were incubated in a test tube with yeast CK (25 milliunits [mU]/mL), choline (1–10,000 μmol/L), MgCl₂ (12.5 mmol/L), and ATP (10 mmol/L) in Tris . HCl buffer (0.1 mol/L, pH 8.75) for 10 min at 23°C. The test tube was gently agitated throughout the incubation period. Tubes were placed in a boiling water bath for 2 min to stop the reaction. To serve as controls, some samples were placed directly in the boiling water bath after addition of all substrates. The phosphorylated fraction of each radiotracer was isolated from the nonmetabolized fraction by anion-exchange chromatography according to the method of Ishidate and Nakazawa (15) and counted in a well counter or a scintillation counter. The percentage of radioactivity converted to the phosphorylated form was calculated. Preliminary studies showed the phosphorylated fraction to rise linearly with time for incubations of <15 min.

Uptake and Metabolism of Radiotracers by PC-3 Human Prostate Cancer Cells

Cells (2.5 × 10⁵ per well) were seeded on 6-well plates and incubated for 2 d, at which time >90% confluence was reached. The incubation medium was Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% calf serum. The medium contained 15 mmol/L glucose. On the day of the study, the medium was refreshed using a volume of 1 mL in each well. Cells were incubated under control conditions or with the addition of metabolic and growth factor receptor inhibitors to test the sensitivity of uptake of the radiotracers to specific inhibition. The inhibitor of choline uptake and phosphorylation, hemicholinium-3 (HC-3) (Research Biochemicals, Natick, MA), was added to give a concentration of 5 mmol/L. The concentration of the inhibitor was approximately 10 times the literature in vivo inhibitory concentration of 50% (IC₅₀) for choline phosphorylation (16). After a 30-min incubation period, the radiotracers were added (∼74 kBq per well) in separate wells and the cells were incubated at 37°C for 2 h. The radioactive medium was removed and the cells were washed 3 times with phosphate-buffered saline solution, released from the plates by incubating briefly with 0.05% trypsin in DMEM, transferred to test tubes, and counted for radioactivity in a γ-counter (¹⁸F) or a liquid scintillation counter (¹⁴C). The amount of radioactivity in the cells was normalized by the dose administered to each well.

Preliminary analysis of radiolabeled metabolites of FCH and CH in PC-3 cells was performed. Cells were incubated in 6-well plates with a mixture of FCH (∼3.7 MBq) and [¹⁴C]CH (∼74 kBq) for 2 h as described above, followed by removal of radioactive medium and 3 rinses with phosphate-buffered saline solution. Methanol (1 mL) was added to each well and the cells were lysed by maintaining the temperature at 37°C for 30 min. The methanol phase was transferred to a glass test tube. Each well was rinsed with an additional 0.5 mL methanol that was added to the original fraction. To each tube, 3 mL chloroform and 1 mL 0.25 mol/L sodium phosphate (pH 4.5) were added to separate lipids from water-soluble molecules. After vigorous mixing of the samples for 1 min, the 2 phases were separated, and a 0.5-mL aliquot of each phase was counted for ¹⁸F and ¹⁴C radioactivity. The aqueous phase was further analyzed for water-soluble metabolites using the gradient-HPLC method described earlier. Radioactivity in the metabolite fractions (lipid, CH/FCH, [¹⁴C]-betaine/FB, [¹⁴C]-phosphocholine/P-FCH) was expressed as the percentage of total radioactivity administered to each well.

Human Dosimetry Estimation

Tissue distribution data (percentage dose per gram of tissue) obtained in a mouse model after injection of FCH (11) were converted to the percentage dose per organ using the method of Kirschner et al. (17). The distribution was assumed to be static at >10 min after injection, consistent with the long retention of the tracer in tissue (11). These data were entered into the MIRDOSE 3.1 program (18) to calculate dose estimates. Urinary radioactivity was assumed to be retained within the urinary bladder. Thus, assumptions made in these calculations would tend to overestimate the radiation dose in human imaging studies in which urinary radioactivity may be voided after the imaging study is performed. Because urinary excretion patterns in rodents are commonly more rapid than those in humans, the assumption of no urinary clearance of radioactivity was precautionary. Bone uptake was distributed at the bone surfaces. The look-up table corresponding to a 70-kg adult male Oak Ridge National Laboratory phantom was used because this would best reflect the primary study population (prostate carcinoma).

Toxicity Study

Four unanesthetized BALB/c nude mice were administered 1 mg/kg [¹⁹F]FCH through tail vein injection and monitored for 48 h. This dose represented an approximately 300,000-fold excess of FCH in comparison with the normal dose that a 70-kg person would receive in a FCH PET study. The mice were killed at 48 h.

PET Studies

The biodistribution of FCH was investigated in a patient with prostate cancer, a patient with metastatic breast cancer, and a patient with a brain tumor. The FCH PET studies were approved by the Duke University Medical Center Institutional Review Board and Cancer Research Committee. The subjects were informed of all risks associated with the study and written informed consent was obtained. Patient 1 had undergone prostatectomy and radiation therapy and was being evaluated for recurrence of prostate cancer. Patient 2, having previously undergone resection of an astrocytoma, was also being evaluated for recurrence of disease. Patient 3 was a 50-y-old breast cancer patient who had undergone mastectomy 3 y earlier, followed by tamoxifen chemotherapy. Imaging was performed using an Advance PET scanner (General Electric Medical Systems, Milwaukee, WI) with an intrinsic spatial resolution of ∼5 mm in all directions (19). A transmission scan of the pelvic region was obtained before administration of radiotracer. [¹⁸F]FCH (93–185 MBq) was administered intravenously. In the first patient dynamic imaging of the lower pelvic region commenced for 10 min (frame sequence = 12 × 10 s, 8 × 1 min). After the dynamic scan, a whole-body emission scan (4 min per bed position) was obtained without attenuation correction. In the patient with a brain tumor, a 4-min transmission scan of the head preceded the injection of radiotracer. For the breast cancer and brain tumor patients, FDG PET images were acquired within 1 wk of the FCH PET scan. Approximately 370 MBq FDG were administered intravenously, and emission imaging commenced at 45 min after injection. The emission imaging of the brain was performed in 3-dimensional scanning mode (septa out), whereas imaging of the body was performed in 2-dimensional mode (septa in). The images were reconstructed using an ordered subset expectation maximization algorithm. Regions of interest (ROIs) were drawn manually on the attenuation-corrected images for evaluation of FCH kinetics in tissues. Standardized uptake values (SUVs) of FCH uptake in tissues were calculated using the attenuation-corrected images according to the equation: where C_FCH is the concentration of FCH in the tumor ROI. In the breast cancer patient, SUVs for FDG uptake within tumors were calculated similarly using similar ROIs drawn on the FDG scans.

Statistical Methods

Results are expressed as mean ± SD. Statistical analysis was performed using the Student t test, and statistical significance was inferred at P < 0.05.

Stability of FCH

The radiochemical purity of the FCH preparation was >99% as monitored by cation-exchange HPLC. Radioactivity concentration was 37–185 MBq/mL. The radiochemical purity remained >99% after maintenance of the FCH preparation at room temperature for 7 h. FCH was also found to be stable in blood samples taken from healthy human subjects (n = 5). HPLC analysis showed FCH to be completely intact after 2-h incubations in whole-blood samples at 37°C.

In Vitro Phosphorylation of FCH by Yeast CK

To determine whether FCH is a substrate for CK, FCH and [methyl-¹⁴C]choline (CH) were incubated with yeast CK (25 mU/mL) and choline (1–10,000 μmol/L) for 10 min at 23°C. Samples quenched by boiling before incubation showed negligible phosphorylation activity. Figure 3 shows that, after separation from unreacted FCH and CH using an anion-exchange resin (15), a single, more polar chemical product is formed from FCH and CH. The ¹⁴C-labeled product exhibited the same retention time as an authentic standard for unlabeled phosphocholine, seen by in-series ultraviolet detection. The presence of a single ¹⁸F-labeled product and the similar relationship of retention times between FCH and CH (Fig. 3A) and their phosphorylated products (Fig. 3B) suggest that the ¹⁸F-labeled product is P-FCH. Figure 4 shows the dependence of phosphorylation rate on choline concentration, displaying a similar sigmoid relationship for both radiotracers that is typical of Michaelis–Menten-type kinetics. The apparent IC₅₀ values for inhibition of phosphorylation of radiotracer by choline were ∼0.4 mmol/L for FCH and CH. The phosphorylation rates were similar for FCH and CH at all choline concentrations. At a low choline concentration of 1 μmol/L, phosphorylation rates of FCH and FMEC were equivalent to that of CH, whereas phosphorylation rates of FEC and FPC were approximately 30% and 60% less than that of FCH, respectively (Table 1).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Normal phase, gradient-HPLC analysis of FCH and [¹⁴C]CH (A); P-FCH and [¹⁴C]phosphoryl-CH enzymatically synthesized using yeast CK according to method of Ishidate and Nakazawa (15) (B); and hydrophilic radiolabeled metabolites in cultured PC-3 prostate cancer cells after incubation with FCH and CH (C). Close correspondence of chromatograms B and C indicates extensive intracellular phosphorylation of FCH and CH in cancer cells.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

In vitro phosphorylation of FCH and CH by CK. Incubations were performed at room temperature with 25 mU/mL CK over choline concentrations of 0.001–10 mmol/L. Each data point represents results as mean ± SD of 3 samples. Similar inhibition of FCH and CH phosphorylation at higher choline concentrations is indicative of competitive inhibition of FCH phosphorylation by choline.

Accumulation of Choline Analogs by Cultured PC-3 Prostate Cancer Cells

Under control conditions, cultured PC-3 human prostate cells accumulated FCH similar to that of CH (Table 1). However, the fluoroethylated analog, FEC, showed only one fifth of the uptake of FCH (P < 0.01). The fluoromethylethyl analog, FMEC, showed an accumulation higher than that of FEC (P < 0.05) but lower than that of FCH (P < 0.01). Uptake of the fluoropropyl analog, FPC, was not significantly different from that of FCH. Specific inhibition of choline transport and phosphorylation by HC-3 resulted in 89% (P < 0.001), 45% (P < 0.01), and 41% decreases in uptakes of FCH, CH, and FPC, respectively.

Analysis of Metabolites of FCH and [¹⁴C]CH in Cultured PC-3 Prostate Cancer Cells

A preliminary analysis of the chemical form of radioactivity present in PC-3 cells incubated with FCH and CH was performed. Cells incubated in 6-well plates for 2 h with radiotracers were washed thoroughly to remove all extracellular radioactivity. The cells were lysed in methanol, and the methanol solution was added to chloroform to solubilize all lipophilic metabolites. The hydrophilic metabolites were extracted with 0.25 mol/L sodium phosphate solution (pH 4.8) and subjected to gradient-HPLC analysis for measurement of radioactivity in the form of FCH/CH, FB/betaine, and P-FCH/phosphocholine. Table 2 shows the results expressed as percentage of administered dose to each well. CH and FCH were found to undergo extensive metabolism within the PC-3 cells. Approximately 72% of ¹⁸F radioactivity was indicated to be P-FCH, whereas 91% of ¹⁴C radioactivity was found as [¹⁴C]phosphocholine (Table 2; Fig. 3). Radiolabeled lipophilic metabolites of FCH comprised 25% of the total ¹⁸F radioactivity, whereas lipophilic metabolites of CH represented only 7% of the total ¹⁴C radioactivity. A 5-fold higher incorporation of radioactivity into lipophilic species for FCH was found relative to CH. Neither [¹⁴C]betaine nor its ¹⁸F-labeled counterpart was detected by HPLC analysis, indicating that oxidation of FCH and CH in the cancer cells was negligible.

Human Dosimetry Estimates

Table 3 gives the radiation dose estimates to human organs as determined from calculations based on mouse FCH biodistribution data (11). To produce conservative estimates, the total-body residence time was assumed to be determined solely from radioactive decay (1.44 × half-life = 2.6 h). The effective dose equivalent from a 370-MBq administration was estimated to be 0.01 Sv, which is below the Food and Drug Administration (FDA) single-study limit of 0.03 Sv for research subjects. However, the largest organ radiation dose (kidney) was 0.081 Gy/370 MBq injected, which is above the 0.05-Sv single-organ dose per study established by the FDA for research studies. Therefore, it was determined that the maximum administered dose in the initial studies with FCH would be 220 MBq. Refinement of these dosimetry estimates is needed using biodistribution data in humans in a subsequent study with larger numbers of subjects.

Toxicity Study

Acute toxicity of [¹⁹F]FCH (1 mg/kg of body weight) was determined in 4 untreated BALB/c mice. No deaths were observed in the mice up to 48 h after administration of [¹⁹F]FCH. Nor were any behavioral or movement abnormalities observed by casual inspection during the monitoring period. On the basis of our estimates of specific activity of the FCH (∼74 GBq/μmol), the normal dose of FCH in the radiotracer preparation would be a factor of ∼300,000 times lower than the dose given in this toxicity study.

PET in Patients

Patient 1.

A 66-y-old man was initially diagnosed 5 y earlier with locally advanced clinical stage T3 N0 M0 prostate cancer and underwent radical prostatectomy followed by external beam radiation to the prostate bed. His serum prostate-specific antigen (PSA) was 2.0 ng/mL after therapy. The patient then underwent androgen-deprivation therapy. Because his serum PSA began rising at approximately 1 y before PET scanning, the chemotherapy was discontinued. The current serum PSA of the patient was 40.1 ng/mL. Recent radionuclide bone scans showed no abnormalities, whereas CT of the chest, abdomen, and pelvis showed only small lymph nodes (∼1 cm) in the pelvis and mediastinum. FCH PET images acquired at 2–4 min revealed accumulation of FCH in the prostate bed before the arrival of activity at the bladder (Fig. 5A). Figure 5B shows the kinetics of FCH in the prostate bed and ROIs drawn within the urinary bladder and iliac artery. The arterial concentration of FCH peaked early and fell rapidly to <5% of the peak level within 2 min after injection, indicating extremely rapid blood clearance. Radioactivity concentration rose rapidly in the prostate bed and reached a plateau by 3 min after injection. Radioactivity began to arrive in the bladder at 4–5 min after injection and the concentration increased rapidly thereafter. The whole-body images showed abnormally high foci of FCH uptake in the chest suggestive of prostate cancer within mediastinal lymph nodes (Fig. 5C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

(A) Attenuation-corrected FCH PET image (coronal projection, 2–4 min after injection) of pelvic region of patient 1 having biopsy-proven recurrent local prostate carcinoma. Slice thickness is 12.9 mm. In this early image, radioactivity had not yet arrived at urinary bladder, allowing excellent delineation of recurrent disease in prostate bed (arrow). (B) Time–activity curves for FCH in same patient as in A show very rapid clearance of radioactivity from ROI placed on iliac artery, rapid accumulation of tracer in local prostate bed, and arrival of radioactivity in urinary bladder at >4 min after injection. (C) Attenuation-corrected whole-body scan (coronal projections) shows several foci of high FCH uptake in mediastinum suggestive of prostate cancer in hilar and paraaortic lymph nodes. L = left.

Patient 2.

A 23-y-old woman was evaluated by FCH PET and FDG PET for recurrence of a previously resected anaplastic astrocytoma (Fig. 6). The PET images were coregistered with T1-weighted magnetic resonance (MR) images using Gd-diethylenetriaminepentaacetic acid contrast enhancement. The FCH PET images showed marked uptake in regions of the wall of the surgical cavity. The low concentration of tracer in the normal cerebral cortex allowed excellent delineation of the tumor from normal brain. The maximal tumor-to-cortex ratio was ∼10:1 and was attained within 5 min after injection. Most, but not all, of the FCH-avid regions corresponded with Gd-enhanced regions on the MR images. Biopsies taken from the cavity wall confirmed recurrence of the tumor. The FDG PET images indicated a thin rim of increased FDG uptake located posteriorly on the cyst wall, but the boundary of the tumor could not be reliably determined because of high uptake of FDG by normal cortex.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Patient with biopsy-confirmed recurrent anaplastic astrocytoma was imaged by T1-weighted Gd-diethylenetriaminepentaacetic acid–enhanced MRI, FCH PET (5–10 min after injection), and FDG PET (30–36 min after injection). MRI shows nodular enhancement posteriorly at postoperative cyst wall. FCH scan shows diffuse abnormal accumulation posteriorly and medially to cyst with focal areas of accumulation corresponding to nodular areas of enhancement on MRI scan. Note absence of normal cortex accumulation that is seen with FCH. FDG scan shows thin rim of abnormal accumulation that would support recurrent tumor, but abnormality is difficult to detect compared with FCH and MRI scans.

Patient 3.

A 50-y-old woman had undergone reexcision of right breast intraductal carcinoma 4 y earlier and then right modified radical mastectomy 3 y earlier followed by tamoxifen chemotherapy. Six months before the PET scans, metastatic breast cancer was detected by palpation (right breast sternal mass) and CT (mediastinal and hilar adenopathies, mass in right anterior pelvis). Tamoxifen therapy was stopped. The FDG PET scan showed uptake of tracer in the sternal mass (SUV, 5.0), mediastinal and hilar adenopathies (SUVs, ∼5.5), and the pelvic tumor (SUV, 9.5) (Fig. 7). Uptake of FCH was also observed in the corresponding tumors: sternal mass (SUV, 14.6), mediastinal and hilar adenopathies (SUVs, ∼8.5), and pelvic mass (SUV, 6.5). The parasternal mass appeared significantly larger on the FCH PET scan relative to that on the FDG PET scan. Smaller metastases were indicated on the FCH PET scan in the right chest wall and left lung that were not seen on the FDG PET scan. In reference to the anterior pelvic tumor, a homogeneous distribution of FDG was observed across the tumor in the FDG PET scan, whereas the FCH was preferentially distributed on the periphery of the tumor.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Patient with metastatic breast cancer underwent FCH PET and FDG PET. Myocardial uptake is observed only with FDG, whereas more prominent uptake in salivary glands, liver, and kidneys is seen with FCH, which is consistent with normal uptake of choline by these tissues. Uptake of FDG and FCH is indicated in large metastases associated with sternum, right hilar and paratracheal lymph nodes, and right anterior pelvis. Volume of submanubrial metastasis was significantly larger on FCH PET scan. Smaller regions of focal uptake are observed on FCH PET scan in right chest wall and left lung (arrows) that are not seen on FDG PET scan. Uptake pattern of FDG was homogeneous across anterior pelvic metastasis, whereas FCH was taken up preferentially by periphery of this tumor.