Automated synthesis of radiopharmaceuticals for positron emission tomography: an apparatus for labelling with [11C] methyl iodide (MIASA)

A fully automated apparatus for the routine synthesis and formulation of short-lived 11C (t1/2 = 20 min) labelled radiopharmaceuticals for positron emission tomography (PET) has been developed. [11C]Carbon dioxide is converted to [11C]methyl iodide, which can be used to label a wide variety of substrates by methylation at C, N, O, or S electron rich centres. The apparatus, MIASA (methyl iodide automated synthesis apparatus), was designed to operate as part of an automated labelling system in a shielded ‘hot’ laboratory. The apparatus was designed without the size constraints of typical instrumentation used in hot cells, although it is compact where necessary. Ample use of indicators and sensors, together with compact design of the reaction flasks for small dead space and efficient evaporation, led to good reliability and performance. The design of the hardware and software is described in this paper, together with a preparation of 3-N-[11C]methylspiperone as a sterile injectable solution in physiological saline.


Introduction
Positron emission tomography (PET) has developed into a unique tool for obtaining quantitative physiological images ofbiofunctions. A variety of applications are being pursued--these are based on the fact that positron emitters, such as 11C, 13N or 150, can be incorporated into almost any biologically active tracer without altering the chemical behaviour.
The labelling of pharmaceuticals with the positron emitter tC tbr use in PET studies is frequently accomplis.hed using [tiC]methyl iodide as a labelling agent. Figure   shows the synthetic scheme used. The short half life of 11C (20"4 min) means it is essential to synthesize the radiopharmaceuticals regularly and consistently on site. It is usually necessary to start with relatively high levels of radioactivity to obtain useful amounts of the required products. So automation of the synthesis apparatuses, for example 3-6, within a radiation-shielded facility is important for safe, reliable and efficient production of radiopharmaceuticals for PET.
A number of remote controlled and semior fully automated systems have been developed for labelling [1-5-1. However, although these systems have begun to address the need for producing labelled compounds, PET has begun to move out of the research arena to become a routine clinical tool and this is putting ever greater demands on the automated apparatus. A reliable system that can reproducibly deliver a variety of radiopharmaceuticals on a routine, repetitive basis would be desirable. The commercial synthesis instruments on the market are generally limited in their scope; need manual washing before re-use; and/or require the attendance of specialist operators or maintenance personnel.
A total system for routine production of PET radiopharmaceuticals is being developed at the Institute for Biofunctional Research (IBR). This paper reports on the design and construction of an apparatus for producing radiopharmaceuticals by labelling with [ttC]methyl iodidemMIASA (methyl iodide automated synthesis apparatus)--and describes its application to synthesis of 3-N-[11C]methylspiperone as a sterile injectable solution in physiological saline.
General features Figure 2 shows the general layout of the facilities at IBR which were designed for the development of a total production system from the cyclotron to the PET camera.
The production of 11C is accomplished by the nuclear reaction of cyclotron-accelerated protons with nitrogen gas, 14N(p,z)tlC, in a target chamber. A Sumitomo Heavy Industries HM-18 cyclotron is used for this. Pico mole quantities of 11C undergo rapid oxidation to [11C]carbon dioxide in the target chamber or by passage over a CuO catalyst. Incorporation of the positron emitters into the radiopharmaceuticals takes place in the hot laboratory, which is designed to contain several fully automated synthesis instruments under computer control.
Communication with a network of OPTOMUX (Opto 22, USA) modules allows flexible control and monitoring of all devices [4,6]. After quality assurance procedures have been completed, the sterile product is passed through to the PET camera room.  sequence or reading the status of sensors. An example of the layout of blocks and their connections for the control of the HPLC injection procedure is shown in figure 3. Figure 4 is a flowchart of the same procedure and a more detailed explanation is given below for the injection procedure during the synthesis of 3-N-[llC]methyl spiperone.
The flowchart for checking the integrity of the apparatus, the diagnostic check, is shown in figure 5. The status of the apparatus is confirmed to be ready for synthesis by intrachecking photosensors, micro-switch sensors, rotary valve positions and performing leak tests, thus helping to maintain reproducible and safe operation of the apparatus. A typical program flowchart for a synthesis and formulation procedure is shown in figure 6.
The synthesis apparatus General As the apparatus did not need to be operated in the confines of a hot cell, it was possible to organize the hardware for easy maintenance and for good reliability and reproducibility. The reaction unit was kept compact in order to minimize dead space in the two reaction flasks (F1 and F2) and flow lines, which can lead to dilution and losses of the 11C-labelling agents, but the peripheral supply and service units were laid so that they were easily accessible. The organization of the units on the two racks (main and supply, 60 x 40 x 180 cm) is shown in figure 7, and a schematic diagram of the apparatus is shown in figure 8. The general appearance of the apparatus and the reaction unit are shown in figures 9 and 10. Step 3 Step 4 Wait 40 s 6WV to Inject; Auto zero UV; Step 5   The reagents, reactants and solvents that are to be used in the reaction flasks are stored in glass reservoirs, from which they are dispensed as required. Figure 11 shows the layout of the reservoir supply unit shelf. The small volumes of liquids required, 100-200 gl from reservoirs R1 and R4, are measured out using fixedposition infra-red photosensors (EE-SX670, Omron) on the Teflon tube delivery lines. A block of black PVC was used to make a light shield which fits tightly into the well of the photosensor; holes for the Teflon tubing (1"6 or 3"0 mm ) and the light beam (1"0 mm ) cross in the centre of the light path. The basic circuit for control of dispensing from a reservoir is shown in figure 12. When a reservoir is opened (Va and Vb on) and the photosensor detects liquid, a signal is sent directly to the computer. Simultaneously, in order to achieve good reproducibility, the signal also switches a relay to immediately cut power to the solenoid valve at the bottom of the reservoir (Vc).
A relay ($64) can switch off the operation of the photosensors to allow washing. Supply Rack Figure 7. Arrangement of the units on MIASA's two racks.

Reaction unit
The front of this unit contains four flasks on four stirrers. These are for: (1) Synthesizing methyl iodide.
(3) Collecting the separated product, evaporating eluant and dissolving in saline solution.
(4) Adjusting the pH and final concentration.
Flask F1 has an outer jacket for cooling and an inner jacket for heating. Coolant, fluorinert FC77, is circulated from a cold bath and heating is performed with a nichrome wire heater (c. 15 f) in silicone oil. The flask is designed so that all the tubing connections to the inner reaction vessel pass through the heating jacket; the flask is compact which helps the efficient removal of solvent from the connections during evaporation and drying processes. Flask F2 has a single jacket containing fluorinert FC77, which can be heated by a nichrome wire heater (c. 20 or cooled by circulation from the cold bath. The four tubing connections pass through the jacket for compactness and efficient drying of the flask. Flask F3 has a single jacket filled with silicone oil, heated by a nichrome wire (c. 20 ). The magnetic stirring bar is supported on a Teflon disk on the Teflon tubing that dips down to the bottom of the flask. The tube is made rigid by an outer Teflon tube collar, so the stirrer bar spins freely. The Teflon tubing passes through a Teflon-coated silicone disk (septum) in the screw connector to minimize the dead space in the flask top (this is difficult to dry during evaporation). Flask F3 was also designed to minimize the loss of product through bumping. A glass protuberance is set just below the tubing connection to the vacuum line, blocking the direct loss of solution to the drain. Flask F4 is fitted with a small pH electrode (CE105-C, Nissin) and a level sensor to detect when the volume of the final solution reaches a pre-set value, for example 10 ml. Motor) for operating a syringe pump (25 ml)--this pushes HPLC eluant from the injection loop to F2 and pulls the reaction solution back into the injection loop.
To ensure the reliability of the injection procedure, two photosensors are used for detecting the point when all the reaction solution has gone into the sample loop. When both photosensors go off, the sample is injected onto the HPLC column by a motor-operated six-way micro rotary valve (E010, Uniflows; 0"75 ml loop). Using two photosensors means that if a small air bubble were to pass through the Teflon tube, it would not accidentally cause premature injection. A third photosensor is fitted to the Teflon flow line just above F2 to detect when the push of eluant fiom the sample loop should be stopped. The set up shown in figure 13 was thus used to obtain reproducible injection.
A manual six-way injection valve (7125, Rheodyne; 0"6 ml loop), for calibration, and two manually operated six-way selection valves (7060, Rheodyne) for changing the HPLC flow line, and thus the HPLC column, are also located on the same shelf. Five columns and one bypass line can typically be connected. When methyl iodide is passed from F1 to methylation flask F2 it is necessary to remove traces of excess acid (HI) and water by passing it through a trap containing soda lime and P205. The trap is fitted to a motor-operated six-way micro rotary valve (E010, Uniflows) to allow it to be by-passed during the washing and drying procedure, which can then follow immediately after synthesis. When the bypass is a second, empty, trap, it can be cleaned and made ready for use on the following run.

Purification unit
The two shelves below the reaction unit contain the purification apparatus, consisting of a compact HPLC pump (PU-980,Jasco) and UV detector (UV-970,Jasco), and a radioisotope detector (positron detector, Aloka). Figure 14 shows the layout of the unit.
Vacuum, wash and drainage unit Also at the bottom of the main rack is a diaphragm type vacuum pump, a.stainless-steel waste drain with a pressure sensor to allow monitoring of the vacuum, three polyethylene tanks for containing wash solvents and one tank to collect solvents from the vacuum pump exhaust line.     AC 100 V supply A series of AC 100 V outlets with safety breakers supply power to both racks.

Cooling system
The lower half of the supply rack contains the cooling system, consisting of a 'cool pipe' cooler with a minimum temperature of -50C, a circulating micro pump, a voltage controller and a 3 Dewar tank of coolant (fluorinert FC77). Solenoid valves are used to stop and direct the flow of coolant to F1 or F2.
The software was developed to be compatible with the total automated-labelling production system, including the cyclotron and the PET camera. MIASA reliably completes the link between the production of the positron emitter radionuclide, 11C, and the delivery of the lC-labelled radiopharmaceutical for PET study.  Reservoir #6 (R6)--c. 5 ml Na2CO3(aq) (0"1 M). The P2Os/sodalime trap was filled and the six-way valve holding the trap was set to direct the flow through the trap. The drying tube on the argon line was filled with 2-3 g ofP2Os. The HPLC eluant was prepared by mixing disodium hydrogen citrate (0"04 M) and methanol in the ratio 52"5:47"5 v/v, followed by degassing under reduced pressure and sonication. An authentic sample of 3-Nmethylspiperone in the eluant was manually injected (c. 500 lal; 50 nmol), and the retention time confirmed to be about 10min using the following chromatographic conditions: column, Capcell pak SG 120 (15 x 150 mm + 15 x 30mm precolumn); flow rate, 9"5 ml/min; wavelength, 254 nm.

Diagnostic check of the apparatus
The status of the apparatus was checked to be ready tbr synthesis, as outlined in figure 5. The position of the six-way rotary valves was checked and set to the correct starting position if necessary. Similarly, the position of the syringe pump was checked and photosensors on the reservoir lines were confirmed to be off.
Leak tests on F1-F4 and the sodalime/P205 trap were performed by closing all outlets, opening them to the argon flow line and monitoring the mass flow controller reading. If zero flow could not be obtained, the source of the leak was searched for and remedied. Get ready to start a synthesis Flask F1 was cooled to about 20C ($63 on) and flushed with dry argon gas (V5 on) before the LAH/THF solution (100 gl, 10 gmol LAH) was injected. Cooling was continued while the target was irradiated for up to 40 min.

Collection of [elc_]carbon dioxide
The contents of the target were swept to the cryotrap and [11C]carbon dioxide was trapped in a coiled tube dipping into liquid argon. The start of transfer, SOT, to MIASA was begun by raising the coil from the liquid argon and flushing with He gas (10ml/min). The radioisotope detector on the inlet line was used to monitor the release of radioactivity from the trap, and valves V9/V9a were opened to direct the flow to F1 when the peak of 11COz was detected. The outlet from F1 (V40) was opened to a sodalime trap to prevent any leakage of radioactive gas. The peak of radioactivity was collected for 2-3"5 min for a 20-40 min irradiation of the target (Nz: 14"7 kg/cm2; 15 gA).
During the collection of CO2, spiperone solution was added to F2. The outlet of F2 was opened by switching on V29 and measurement from R2 was performed by opening V12 and V24 until photosensor #2 (PS2) switched on, and then opening V23 to flush the measured volume to F2. Addition was repeated three times to add a total of 300 lal, mg spiperone.
[11CJMethyl iodide synthesis and labelling After the collection of 1CO2 in F1 the THF was evaporated by evacuation and heating to 130C for 2 min. Flask F1 was then cooled to c. 45C by switching on the cooling pump. Valves V15 and V29 were opened and HI(aq) solution  Transfer from F1 was stopped after 3 min by closing V15, and TBAOH base (200 gl) was added to F2 from R3. Flask F2 was completely closed and then heated to 65C, set with the temperature controller (AOUT1). The HPLC pump was also started in preparation for the next step.

HPLC injection
After 4 min reaction, the injection of the reaction mixture into the HPLC column was started. The six-way micro rotary valve, holding the injection loop, was turned to the load position and valves V45 and V29 were opened. The eluant in the loop was pushed by the syringe pump to fill the Teflon tube between the loop and F2. When photosensor # 11 (PS11) detected the liquid just above F2, a short time (3 s) was allowed for the tubing to be completely filled, and then the push was stopped by closing V45. The reaction mixture in F2 was neutralized by addition ofdilute UClaq (1 M)/THF (56:44 v/v) from R4. Valves V14 and V28 were opened until photosensor # 4 (PS4) switched on, and then V27 was opened to flush the measured volume of acid (100 l.tl) to F2. Stirring in F2 was stopped and the two-phase reaction mixture allowed to separate. The addition of THF with the acid helped to speed up the separation and reduce the cloudiness of the mixture, which was essential for the operation of the photosensors. The reaction mixture in F2 was loaded into the injection loop by opening V45 and pulling with the syringe pump until photosensor # 6 (PS6) detected that no solution remained in the Teflon tube. The HPLC six-way valve was then turned to inject. The heater of F3 was then switched in preparation for the subsequent evaporation step.

Product collection
The UV and radioisotope detectors on the HPLC line were monitored and recorded on the computer, and the 3-N-[11C]methylspiperone peak eluting at c. 10 min was collected into F3 by opening V16. Nitrogen gas was bubbled through F3 to start the evaporation of the eluant as soon as collection proceeded. Collection was stopped by closing V16, and evaporation was continued under evacuation. Opening V39 at the same time ensured that the delivery line into F3 was emptied and also helped to prevent loss of product by bumping. In order to determine the radiochemical yield and specific activity at end of synthesis (EOS), the product could be collected into a volumetric flask placed in a Curie meter (runs HR6, HR8 and HR9), and no further formulation was performed. The collection fraction was made up to 25 ml with HPLC mobile phase after radioactivity had decayed and the amount of 3-N-methylspiperone was analyzed by HPLC.

Product formulation
After allowing 9 min for evaporation to dryness in F3, the heater and vacuum pump were switched off and saline solution (5 ml) added from R3. The product was dissolved in the saline by stirring and bubbling nitrogen gas.
Applying the vacuum pump for a short time (2 s) improved the dissolution of the product by making the bubbling of nitrogen gas more vigorous, which resulted in more effective washing of the top part of F3 with the saline solution.
Flask F4 was emptied of water by opening valves V42, V33 and switching on the vacuum pump, and then the saline solution was transferred from F3 for pH adjustment. Sodium carbonate solution (0"1 M) was added from R6 by repeatedly opening V 19 for s, until the pH was within the range 7"0 _ _ 1"0. The volume of the final solution in F4 was adjusted to 10 _ _ ml (set by the level sensor position) by adding more saline from R5. Valves V18 and V38 were opened for s, and then V43 and V37 were opened to flush saline to F4.
Finally, the product solution was filtered through a 0"22 lam sterile filter into a sealed vial by applying nitrogen gas pressure. The vial was placed in a Curie meter to allow the amount of radioactivity in the product to be measured. The specific activity of the product was calculated after HPLC analysis of the 3-N-methylspiperone in the final solution at 18 h after end of bombardment (EOB).

Washing
Washing of the apparatus could be started immediately after collection of the product. The position of the six-way valve holding the sodalime/P20 5 trap was changed to direct the flow through a bypass and the photosensors on the reservoir lines were switched off with a relay. F1 and R1 were washed with water and then the soda lime/P20 trap and F2 were washed by opening V15 and V29 to pass water from F 1. F2 was also washed from the reservoir line via each of V23, V25 and V27. Washing was then repeated with acetone. F3 and F4 were washed with methanol and water, leaving water in F4 to preserve the pH meter. The HPLC column was washed with water (c, 3 min, 9"5 ml/min), 90 methanol (c. 12 min, 9"5 ml/min) and then water again (c. 3 min, 9"5 ml/min), to remove dichlorobenzene and prevent build up of impurities.

Results and discussion
The total synthesis time from the end of bombardment to production of the sterilized saline solution was approximately 40 min. Table 2   Thus the synthesis and formulation of 3-N-[lC]methyl spiperone with the fully automated apparatus MIASA was sufficiently fast for a high specific activity product to be obtained, suitable for dopamine and serotonin receptor binding studies. The specific activity previously reported by Burns et al. [9], using a similar reaction condition, was significantly lower (270 mCi/lamol). The high specific activity is presumably a result of the precautions taken in the preparation and handling of [C]methyl iodide--notably in the preparation of the LAH/THF solution, the conditioning of the flow lines and F1 with dry argon gas before synthesis and the use of high purity (99"9999) nitrogen gas as the target material.
Improvements were made to the hardware and software during the development of the apparatus. For example the use of a mass flow controller, to monitor and regulate gas flows, aided the detection and elimination of leaks prior to synthesis runs. Safety and reliability were improved. The use of three photosensors to control the injection procedure, improved the reproducibility of this crucial step.
The computer interface unit (OPTOMUX) and software are compatible with the total production system of PET radiopharmaceuticals. In future, it will be possible (using the present hardware) to use feedback to monitor and control such processes as solvent evaporation (THF in F1 and temperature maintenance. [xXC]raclopride, has also been synthesized.