Microfluidics for Positron Emission Tomography (PET) Imaging Probe Development

Abstract

Due to increased needs for Positron Emission Tomography (PET) scanning, high demands for a wide variety of radiolabeled compounds will have to be met by exploiting novel radiochemistry and engineering technologies to improve the production and development of PET probes. The application of microfluidic reactors to perform radiosyntheses is currently attracting a great deal of interest because of their potential to deliver many advantages over conventional labeling systems. Microfluidic-based radiochemistry can lead to the use of smaller quantities of precursors, accelerated reaction rates and easier purification processes with greater yield and higher specific activity of desired probes. Several ‘proof-of-principle’ examples, along with basics of device architecture and operation, and potential limitations of each design are discussed here. Along with the concept of radioisotope distribution from centralized cyclotron facilities to individual imaging centers and laboratories (“decentralized model”), an easy-to-use, standalone, flexible, fully-automated radiochemical microfluidic platform can open up to simpler and more cost-effective procedures for molecular imaging using PET.

Introduction

Molecular imaging with Positron Emission Tomography (PET)

Positron Emission Tomography (PET) is an exquisitely sensitive, non-invasive imaging technique for visualizing molecular processes in humans and animals.^{1, 2} It measures the spatial and temporal distribution of positron emitters in the human body by coincidence detection of annihilation photons upon positron decay. Based on data collected from those events, PET imaging systems generate three-dimensional pictures from tracer locations in the tissue or organs of a patient. In addition, PET allows real-time studies of pharmacokinetics and pharmacodynamics of tracers or radiolabeled drugs inside the human body. Clinically, PET is heavily used in oncology for tumor imaging and detection of metastases. In neurology, PET has found application for diagnosis of certain diffuse brain abnormalities and neuro-degenerative diseases. Equivalently, PET has become an important research tool to map human brain and heart function, and visualize specific biological processes in normal or diseased animal models to facilitate drug discovery and development. Over the past four decades, advances in instrumentation, basic biochemical research and clinical investigations as well as radiochemistry for probe production have merged to make PET a truly powerful scientific and clinical tool to directly evaluate many critical in vivo biological functions.

In order to obtain PET images, a radiolabeled probe targeted toward a specific biological entity or incorporated in certain metabolic pathways is administered to the subject under study. So specific PET probes are developed to meet criteria required for imaging selected biochemical and physiological processes in vivo. Several short-lived positron emitters, such as O-15, N-13 and C-11 (half-lives (T_1/2)= 2, 10 and 20.4 minutes, respectively), are isotopes of key elements of living organisms that extend the possibility of synthesizing radiolabeled endogenous tracers for imaging studies of specific biochemical processes. Alternatively, radiohalogens, like F-18, Br-76 and I-124, with longer half-lives (110 minutes, 16 hours and 4.3 days, respectively) allow different synthetic strategies and imaging protocols. Positron-emitting metals, like Cu-64 (T_1/2 = 12.8 hours) produced by charged-particle nuclear reactions and Ga-68 (T_1/2 = 68 minutes) obtained from a generator, are also gaining interest and usage.^{3, 4} Despite not being ideal for all applications, F-18 is easily produced in biomedical cyclotrons and its half-life is sufficient to permit complex or multistep radiochemical syntheses, delivery to off-site imaging centers, and extended in vivo PET imaging. Therefore, currently F-18 is widely used clinically because its low positron energy (0.64 MeV) limits the patient’s absorbed dose and offers high-quality PET images.

2-[¹⁸F]fluoro-2-deoxy-D-glucose ([¹⁸F]FDG),^5–8 a glucose analog, is the most frequently used PET probe to access abnormal glucose metabolism in patients. [¹⁸F]FDG-PET is widely applied in oncology, cardiology and neurology.^{9, 10} These procedures are performed routinely in major hospitals and medical centers that are equipped with PET or PET/CT scanners. In parallel, an extensive network of cyclotron and radiopharmacy sites provide convenient sources for positron-emitting radionuclides and tracers, such as [¹⁸F]sodium fluoride,¹¹ [¹⁸F]FDG,¹² and 3′-deoxy-3′-[¹⁸F]fluoro-L-thymidine ([¹⁸F]FLT).^{13, 14} Meanwhile, in a research or clinical setting, several PET probes aiming to visualize specific receptors or target different in vivo biochemical processes have also been developed and applied in imaging studies, such as O-2- [¹⁸F]fluoroethyl-L-tyrosine ([¹⁸F]FET) for measuring transport rates of amino acids, ^15–18 L-3,4- dihydroxy-6-[¹⁸F]fluorophenylalanine ([¹⁸F]FDOPA) for probing cerebral dopamine meta bolism and neuroendocrine tumors,¹⁹ [¹⁸F]fallypride for estimating expression level of dopamine subtype-2 (D2) receptor in brain,^{20, 21} and [¹⁸F]fluoromisonidazole ([¹⁸F]FMISO) for mapping tumor hypoxia in vivo.^22–25 However, despite being extremely versatile, more widespread use of PET is currently hindered by the limited availability of specific radiolabeled probes that can clearly visualize human disease processes.

Current Technologies for PET Probe Production and their Limitations

Due to the realization of PET as a powerful molecular imaging tool for clinical studies in disease diagnosis and receptor mapping, automated syntheses of radiopharmaceuticals labeled with positron-emitters is becoming increasingly important. Driven by the necessity for carrying out repetitive synthetic steps and cycles, many different types of automated radiochemistry synthesis modules/platforms have been designed and implemented to produce PET probes in an efficient, reproducible and safe manner. Generally, because of short half-lives of these PET probes, their syntheses must be accomplished within two half-lives after the isotope is produced. Besides achieving high specific activity (SA) required for certain probes, large amounts of radioactivity are often used in order to compensate for radioactive decay and, sometimes, low radiochemical yields of the final products. Careful consideration of shielding, remote operations, and automation are necessary and integrated into the experimental planning and production protocols. Therefore, in spite of the advantages PET offers, reliable and routine production of PET probes is indeed a major challenge to radiochemists as well as engineers.

Presently, common PET radiopharmaceuticals are mainly produced by specialized automated synthesis modules. In fact, the utilization of those modules has enabled large-scale [¹⁸F]FDG production and made a substantial contribution to the widespread use of [¹⁸F]FDG PET in cancer diagnostics. For example, typical large-scale [¹⁸F]FDG production can provide enough doses for 50 to 100 patients which can supply several PET centers. After quality control and assurance (QC/QA) tests, [¹⁸F]FDG is distributed for clinical uses from production site to imaging facilities that are within a distance of 10 to 100 miles (i.e. 0.5–2 hours transportation time).

These automated synthesis modules are designed to: (1) reduce personnel exposure to harmful levels of radiation; (2) increase user efficiency and minimize operator error by reducing manual operation; (3) increase batch-to-batch reproducibility; (4) facilitate good laboratory practices (GLP) and current good manufacturing practices (cGMP) compliance for PET drugs. However, there are several drawbacks to current practices. First, radiolabeling reactions are inherently carried out on a small scale due to the minute amounts of radioactive material (nano- to microgram masses) produced from a biomedical cyclotron. The reactor volume of these modules is typically 0.3–3.0 mL and the entire system is designed to handle macroscale volume. Reaction rate is significantly reduced by the dilution of radioisotopes. In order to compensate for short half-lives, a large (10² to 10⁶-fold) excess of the probe precursor is applied to promote rapid and efficient incorporation of the radioisotope into the target radiotracer. These conditions present a challenge for rapid purification/separation of an extra-small quantity of radioactive product from a mixture composed of a large excess of unreacted/decomposed precursor or byproducts (mg scale). Second, automated radiochemistry modules are generally expensive and must be hosted in a lead-shielded hot cell or mini-cell for operation by highly-skilled personnel. Third, expanding the capability of PET imaging requires the development of new PET probes for imaging novel biological targets or unique biochemical processes relevant to specific diseases.

To accelerate the discovery process, a highly flexible synthesis platform is required to perform various radiochemistry processes to make a reasonable number of probe candidates that can be applied for both in vitro and in vivo screening and characterization. Currently, most commercial radiochemical synthesizers are probe-specific and often have very limited flexibility for adopting other synthetic protocols for known or new PET probes. Although some custom-built robotic platforms have a certain degree of flexibility, they generally require a long reconfiguration interval for a new synthetic process.²⁶ The lack of highly versatile radiochemical systems, which can handle a variety of precursor in minute quantity and small volumes to facilitate diverse radiochemical operations, poses a major bottleneck in the discovery and development of new PET probes.

Microfluidic-based Radiochemical Synthesis Technologies

Microfluidic reactors are generally defined as devices consisting of a network of micron-sized channels (typically 10–500 μm) embedded in a solid substrate such as glass, metal or plastic. They are capable of controlling and transferring small amounts of liquids which allow chemical processes and biochemical assays to be integrated and carried out on a microscale. Microfluidic platforms have been widely used in the fields of biology and chemistry. For example, in biology, applications using microfluidic devices include immunoassays, Polymerase Chain Reaction (PCR) amplification, DNA/RNA analyses, chemical gradient formation, cell manipulation, separation and patterning, etc.^27–29 Many of applications are aimed to improving aspects of critical basic biological analyses and clinical diagnosis. On the chemistry side, many examples have demonstrated that microfluidic reactors are extremely useful in performing a wide variety of chemical reactions with purer products, higher yields, greater selectivities and shorter reaction times than those obtained in batch-scale reactions. Furthermore, as one of the major attributes of the microfluidic technology, more benign and milder reaction conditions can be found for certain reactions within microfluidic devices. Manufacturing processes to fabricate microfluidic devices are relatively inexpensive and easily amenable to mass production, even with highly elaborate, complicated device designs. Analogous to microelectronics, microfluidic technologies enable the implementation and production of highly integrated, yet relatively inexpensive and disposable devices for performing several different functions on the same chip. Biology and chemistry carried out in integrated microfluidic chips have great promise as a foundation for new chemical technology and processes.

A typical microfluidic platform³⁰ has a ‘lab-on-chip’ configuration that can manipulate small volumes (from fL to μL) of fluids constrained within microfluidic channels. In general, platforms have a series of generic components for: (1) introducing and removing multiple reagents and samples within microfluidic channels; (2) efficient mixing of separate reagents; (3) performing reactions with precise temperature control; and (4) carrying out other functionalities, such as purification for chemical syntheses or detection for product characterization. At the microscale, heat and mass transfer are very rapid due to high surface area-to-volume ratios and large heat exchange coefficient so reaction conditions within such an environment can be precisely controlled. As a consequence, the yield and selectivity of many biological reactions and chemical syntheses carried out in microfluidic channels can be dramatically improved, while the consumption of reagents and reaction/process time is significantly reduced.^{31, 32} In addition, unlike devices used in macroscale chemical and biological reactions, microfluidic platforms exhibit a high degree of scalability, reproducibility, modularity and automaticity.

Miniaturization of PET radiosynthesis devices is an emerging technology that has the potential to address many of current issues associated with PET probe production.^31–35 For example, it allows the use of very small quantities of expensive precursors than currently required in conventional radiosyntheses, thus reducing production cost while simplifying purification processes. With improved yield and specific activity, much less starting radioactivity is required, lowering the shielding cost while enhancing safety. In addition, the ability to manipulate reagent concentrations, reaction interfaces and rapid mixing in both space and time within the microchannel network of a microfluidic device provides the fine level of control that is highly desired in PET probe production processes. As a result, the microfluidic platform is very practical for synthesizing PET probes³³ on demand because the quantity of radioisotope required for single patient injection is extremely minute. In general, microfluidic radiochemistry platforms can offer several advantages: (1) reliable production of PET probes with high radiochemical purity (RCP) and radiochemical yield (RCY); (2) stand-alone and self-shielded devices without full-sized hot-cells; (3) production of various probes using custom-designed microfluidic chips; (4) utilization of lower starting activity and less probe precursor, and (5) significant reduction in separation challenge during purifications. Most important, this approach provides a suitable direction for the development of cost-effective, interchangeable, disposable and quality-assured microfluidic chip-based radiochemistry processes.

There are two major types of microfluidic platforms used for radiochemical preparation of PET probes: continuous-flow systems and batch-based reactor systems:

1. Continuous-flow Microfluidics

A continuous-flow microfluidic platform^36–39 is based on internally-connected microfluidic channel networks and is perhaps the simplest microfluidic configuration. Since no valve or other isolation mechanism involved, it is easily constructed by mechanical assembly of capillary tubes or an “on-chip” microfluidic network fabricated using common lithography techniques. Generally, the individual microfluidic channel is made of glass, metal, silicon or plastics that are resistant to organic solvents. In a typical operation, reagents and solvents are introduced into a continuous-flow microfluidic reactor by syringe pumps, peristaltic pumps or by simply applying back-pressure through tube connections. After reagents are mixed, reactions can occur by flowing a reaction mixture through the heating/cooling areas with well-controlled temperature and residence time. Finally, the products are collected on the other end of the channels. Continuous-flow microreactors are particularly well suited to conducting multiphase gas–liquid or gas–liquid–solid reactions because of their capability to sustain high-pressure. The pressurization allows reactions to be increased in rate at elevated temperature beyond solvent boiling point and it also facilitates the dissolution of reactant gas into the flow stream. Therefore, the characteristics of a continuous flow microfluidic system confer great efficiency and reproducibility for various chemical and radiochemical reactions. However, without valve control, it is rather difficult to isolate processes from one another to avoid cross-contamination. Even though the application of external valves can provide certain degree of controllability, serious detrimental effects such as possible contamination, transfer delays, or material loss in dead volumes, can jeopardize the entire operation.³⁴

2. Batch-based Microfluidic Reactors

In contrast to a continuous-flow system, a batch-based microfluidic reactor platform^40–44 represents a scalable integration of microfluidic channel networks with functional microfluidic modules, enabling the execution and automation of complicated chemical reactions and biological operations within a single device. Among the current designs, pioneered by Quake et al., the poly(dimethylsiloxane) (PDMS, a silicone polymer)-based^45–48 microvalve-controlled integrated system⁴⁹ is one of the most representative examples. Such microvalve-based chips offer several advantages over continuous-flow microfluidics: (1) precise reagent loading and delivery; (2) facilitated mixing with on-chip peristaltic pumps;⁵⁰ (3) isolation of distinct regions in a microreactor to avoid cross-contamination; (4) automated programmable control for sequential synthesis. Several multi-step radiochemical reactions have been successfully conducted using batch-based microfluidic reactors. Despite those proof-of-concept demonstrations, some issues still remain.

Currently, batch-based microfluidic reactors use PDMS as the chip-making material because it is inexpensive, easy to fabricate and the microvalve systems require a flexible membrane. However, the current chip material and architecture lack high-pressure capability and they cannot be used in radiochemical reactions under high-pressure. In addition, PDMS is hydrophobic and swells in many organic solvents, restricting its use to reactions that do not involve organic solvents or reagents that cause swelling or damage.^{51, 52} Another major concern, especially after heating in the presence of organic solvents, is that radiolabeled probes, unreacted F-18 complex/salt and solvent can be extracted from solution and absorbed into PDMS crevices. These detrimental effects can potentially cause dose loss and reduce reaction yield.

Single Patient Dose on Demand

PET imaging assessment provides the modern physician unprecedented opportunities for disease diagnosis and therapeutic monitoring at the molecular level.¹ As personalized medicine continues to evolve into more specific, tailored treatment regimens to target particular molecular characteristics of a patients’ disease, specific PET probes to improve diagnosis and guide treatment selection are crucial and will become indispensable. However, this goal cannot be achieved without easy and reliable access to different PET probes specific to particular diseased states. At present, the PET probe production based on a cyclotron-centered distribution model (i.e. “centralized model”) allows [¹⁸F]FDG and several other common probes to be synthesized in large quality within radiopharmacies equipped with biomedical cyclotrons and automated synthesis modules, then transported to imaging centers where PET scans are performed (Figure 1). Since each scan is performed in one patient at a time, at ca 0.5–1 hour intervals, significant radioactive decay can occur between patient administration from nominal storage and transport from the radiopharmacy. At the same time, the RCP of PET probes might decrease continuously. For example, it is well known that [¹⁸F]FDG can undergo radiolysis upon storage and this might confound imaging results and application potential.⁵³ Therefore, as the demand for a diverse array of PET probes specific to certain biological processes continuously increases, it may become impractical to produce many different PET probes in high activity within centralized facilities and dispense single to a few patient doses for individual PET imaging without increasing the burdens of isotope production and risk of radiolysis.

Figure 1.

Schematic illustration of two PET probe distribution models: (1) centralized model for delivery of PET probes; (2) decentralized mode for delivery of radioisotopes.

Single-dose on demand has emerged as a concept to overcome these limitations, whereby probe synthesis is conducted for each patient (or study) just prior to imaging. In this way, patients can be scheduled according to imaging need rather than dose availability. This also reduces the amount of decayed products and probe radiolysis.^{53, 54} One idea to fulfill this concept of single-dose on demand is to employ a “decentralized” model whereby probe production is transitioned to individual PET imaging centers or preclinical research facilities with the supply of radioisotopes, such as F-18, from off-site cyclotrons (Figure 1). This approach can be realized through the use of novel on-site microfluidic radiochemistry platforms capable of performing different radiochemical syntheses to yield desired probes in a microscale environment. In addition, the microfluidic on-site PET probe production platform can also enable biologists and clinicians in the basic research community along with R&D scientists in the pharmaceutical and biotechnology sectors to synthesize small-molecule based tracers or prosthetic groups to label biological molecules or drugs of interest by technicians.

Ideal PET Probe Production Platform

So what are the requirements for a versatile, microscale on-site PET probe production platform? It should fulfill the following specifications: (1) ability to handle liquids in small volumes to produce single patient dose on demand; (2) kit-based disposable reactors, fluid tubing and reagent cartridges to avoid contaminations; (3) universal base platform for different probes to reduce instrument cost and simplify operation; (4) easy to use, low maintenance and reliable mechanical performance; (5) minimum and inexpensive shielding; (6) possible automated purification, formulation and final QC/QA assessment.

Microfluidics allows decreased reaction volumes, avoids reagent waste, reduces costs and increases reaction speed, and enables new probe distribution models that are not possible at the macroscopic scale. Since many of the milestones in delineating biochemical processes and mapping receptor density in the human body and brain have involved ¹¹C- and ¹⁸F-labeled probes, we focus on discussing microfluidic radiochemistry using these two isotopes.

Microfluidics in C-11 Radiochemistry

The progress of making PET a powerful imaging technique in nuclear medicine, and drug research and development is driven by an increasing demand for novel imaging probes, especially for those labeled with C-11 probes.^{55, 56} The unique feature of C-11 in PET studies comes from its short half-life of 20.4 minutes, a time interval that permits repeated PET studies and, to some extent, simple radiochemical transformations. Additionally, unlike ¹⁸F-labeled probes, which often required functional group replacement during the labeling process (e.g. fluorine-18 to a hydroxyl group) and can change the target binding affinity or other pharmacological properties, ¹¹C-labeled probes can be very similar, if not identical, to the original compound to be labeled. Therefore, radiolabeling through substitution of a carbon atom with C-11 makes the corresponding ¹¹C-labeled tracers indistinguishable from their stable counterparts within the biological system.

On the other hand, the short half-life of C-11 constrains the synthetic strategy of ¹¹C-labeled compounds. For example, it is preferred to introduce the C-11 radioisotope at the latest stage of syntheses, if possible. In addition, the final ¹¹C-labeled probes have to be purified and pass quality control rapidly before being used in vivo. Although a wide variety of ¹¹C-labeled PET probes exist (e.g. [¹¹C]methionine,^{57, 58} [¹¹C]choline,^{59, 60} [¹¹C]acetate,^{61, 62} [¹¹C]raclopride,^63–65 [¹¹C]-2β-carbomethoxy-3β-(4-fluorophenyl)tropane (β-CFT, WIN 35,428),^66–68 etc.), they can be only routinely available for those privileged PET imaging centers equipped with on-site cyclotron and radiochemistry facilities.

Currently the most dominant source of C-11 generated in biomedical cyclotrons is [¹¹C]CO₂. Several reactions, such as [¹¹C]methylation via [¹¹C]CH₃I and [¹¹C]carbonylation via [¹¹C]CO, have been carried out using microfluidic reactors, because [¹¹C]CH₃I and [¹¹C]CO are the most popular secondary precursors derived from [¹¹C]CO₂. It matches well with the requirements and characteristics of C-11 radiochemistry. The promising initial results reported by Brady and Lu et al.^{69, 70} make microfluidic-based ¹¹C-labeling reaction a very attractive approach for producing ¹¹C-labeled PET probes. Once widely utilized, microfluidic-based C-11 radiochemistry will certainly have significant impacts on radiopharmaceutical chemistry and drug discovery in the future. We have summarized important examples of ¹¹C-labeled compounds generated by microfluidic reactors in Table 1.

Table 1.

Summary of ¹¹C-labeled compounds generated using microfluidic reactors.

Compound	11C-Labeling reagent	Reaction condition	Radiochemical yield	Reference
1	[11C]CH3I	NaOH, acetone RT	not reported	Brady et al. 2003
2	[11C]CH3I	acetone RT	10%	ibid
3	[11C]CH3I	NEt3, acetone RT	5–19%	ibid Jeffrey et al. 2004
4	[11C]CH3I	nBu4NOH, DMF RT	56%a, 88%b	Lu et al. 2004
5	[11C]CH3I	nBu4NOH, DMF RT	45%a, 65%b	ibid
6	[11C]CO	silica-supported Pd catalyst 75 °C	6a R=H 72% 6b R=CN 68% 6c R=CF3 57% 6d R=OMe 39%	Miller et al. 2007

Note:

^ainfusion rate=10 μL/min;

^binfusion rate=1 μL/min

1. [¹¹C]Methylation in Microfluidic Reactors

In 2004, Lu and Pike et al.⁶⁹ reported the synthesis of ¹¹C-labeled carboxylic ester via [¹¹C]methylation using a simple hydrodynamically-driven micro-reactor. The design of this T-shaped glass chip is illustrated in Figure 2. It was composed of two inlet ports for reagent delivery, a microfluidic channel for mixing and radiolabeling (with dimensions: 220 μm (width) × 60 μm (height) × 14mm (length) and total volume of approximately 0.2 μL), and one outlet port for product collection. Precise control of flow rate was attained by syringe pumps that delivered [¹¹C]CH₃I and 3-(3-pyridinyl)propionic acid accurately into the microchannel. The best RCY (RCY mentioned here is “decay-corrected”) of the corresponding [¹¹C]methyl ester 4 was 88% when the infusion rate of reagents was set at 1 μL/min at room temperature. This microreactor was further applied to prepare compound 5, a brain peripheral benzodiazepine receptor ligand via ¹¹C-methylation. The highest RCY was 65% when the infusion rate was also at 1 μL/min. The same device can be used for the synthesis of ¹⁸F-labeled esters as well (described below). This simple device demonstrated the unique advantages and promises of using a microfluidic for PET imaging probe synthesis.

Figure 2.

Schematic illustration of T-shape microreactor via [¹¹C] methylation and [¹⁸F]fluoroethylation.

Brady et al.⁷⁰ also described a similar glass microreactor for ¹¹C-labeling of various substrates containing –NH, –OH, and –SH functional groups in their patent, which also described the direct application of [¹¹C]CH₃I for ¹¹C-methylation. For example, the reported RCYs of 1–3 were 5–19% in the case of substrates containing secondary amines with the infusion rate from 10 to 100 μL/min. It should be noted that the radioactive reaction mixtures obtained from the microreactor were easily and rapidly separable on an analytical High-Performance Liquid Chromatography (HPLC) column, since only low amounts of material are present in a low volume. The radiotracer may be obtained from the HPLC in a smaller volume which, in turn, facilitates easier formulation for safe intravenous administration. These results exemplify some of the potential advantages of this methodology for radiotracer synthesis, which, in successive generations, should be amenable to greater sophistication to encompass entire radiosyntheses in a versatile high throughput manner.

2. ¹¹C-Carbonylation in Microfluidic Reactors

For PET chemistry, ¹¹C-carbonylation reaction using [¹¹C]CO is a very effective means to introduce a ¹¹C-labeled carbonyl group into a probe since many biologically active substances often contain carbonyl groups or functionalities that can be derived from a carbonyl group, such as amides, esters, lactams, and lactones.⁷¹ However, direct introduction of gaseous [¹¹C]CO into a microfluidic reactor is challenging. For example, the low [¹¹C]CO trapping efficiency and poor reagent/substance contact in reaction mixture often results in low RCY. In addition, the harsh reaction conditions, such as the elevated temperature, high pressure and use of organic solvents, make the current plastic-based (e.g. PDMS) microfluidic reactors suboptimal. Therefore, until now, only the microfluidic reactors based on micro-capillary format have shown promise for [¹¹C]CO-based radiochemistry.

Recently, Miller et al.⁷² described a simple, low-cost and effective microreactor employing a reusable silica-supported palladium catalyst packed into standard Teflon tube⁷³ for the rapid multiphase ¹¹C-carbonylation reactions (Figure 3a). It combined three phases, gaseous [¹¹C]CO, substrates in solution and solid-phase catalyst in the ¹¹C-labeling process. The main component of the microreactor is the catalyst-immobilized Teflon microtube (1mm (diameter) × 45 cm (length)) connected with precision syringe pumps/injectors and a mass flow controller to meter the flow of [¹¹C]CO through a mixing T-shaped connector (Figure 3a). The active catalyst within the microtube is a palladium-phosphine complex attached to the silica-support material, where the surface area-to-volume is drastically increased. It is estimated to be about 20,000 times larger than that by coating only the catalyst inside the surface of the microchannel. A feature such as this has dual properties in contributing to the improved yields of ¹¹C-labeled amides. It provides large active surface area (coated with Pd catalyst) for the catalytic reaction to occur. As well, it increases the turbulence of both [¹¹C]CO and substrate solution, enhancing the mixing efficiency. This platform is, indeed, a very elegant and highly efficient design.

Figure 3.

(a) Schematic illustration of microfluidic reactor based on microtube for ¹¹Ccarbonylation and (b) amide formation reactions via a ¹¹C-carbonylative cross coupling reaction.

In a typical experiment, the substrate solution (aryl halide and benzylamine in THF) was infused into the microtube reactor with a steady stream of [¹¹C]CO gas controlled by a mass-flow controller through the T-shaped piece. Temperature in the microtube was maintained at 80°C for 6 min to generate the target ¹¹C-labeled compounds. Several aryl halides were investigated to test the applicability of this microtube reactor to [¹¹C]CO labeling for amide formation (Figure 3b). The ¹¹C-labeled amides 6a and 6b were produced in good RCY (65–79%) whereas the modest RCYs (33–46%) were obtained for 6c and 6d. Undoubtedly, ¹¹C-carbonylative cross coupling reactions utilizing a microtube reactor with immobilized catalysts represent an efficient method to facilitate the development of a wide variety of ¹¹C-labeled PET probes useful for imaging human diseases.

In conclusion, a microtube reactor packed with palladium-supported catalyst (which proved to be reusable) is an effective method for continuous flow carbonylation reactions. This methodology has been successfully applied towards radiolabeling by [¹¹C]CO carbonylative cross-coupling reactions and obtained modest to good radiochemical yields and purities of labeled amides. RCYs can be improved by increasing the residence time of the substrates within the microtube reactor to give better synthetic and radiochemical output. It has great potential to be very useful methodology to discover a diverse array of ¹¹C-labeled PET probes from [¹¹C]CO radiolabeling.

Microfluidics in F-18 Radiochemistry

Significant research efforts have been directed to developing microfluidic reactors capable of producing PET imaging probes, especially for [¹⁸F]FDG. As with ¹¹C-radiolabeling, microfluidic devices provide an excellent match for utilizing minute amounts of precursors to yield ¹⁸F-labeled PET probes. In the following section, we will highlight some of the most notable examples of ¹⁸F-radiofluorination reactions performed in microfluidic devices.

[¹⁸F]FDG Synthesis using Microfluidic Devices

[¹⁸F]FDG, an ¹⁸F-labeled glucose analog, is the most widely available PET probe for imaging normal and elevated glucose utilization states of disease processes in brain, heart and cancer.⁷⁴ Its radiosynthesis is typically composed of three sequential steps^{75, 76} (Scheme 1): (1) concentration and drying of the [¹⁸F]fluoride from [¹⁸O]H₂O solution (1–10 ppb), which is produced in a biomedical cyclotron using proton bombardment; (2) radiofluorination of the precursor (D-mannose triflate 9) using the nucleophilic ¹⁸F-fluorinating reagent ([K⊃2.2.2][¹⁸F]F, 8), which is an organic solvent-soluble metal-fluoride complex made of a cryptand, K222 (4,7,13,16,21,24-hexaoxa-1,10-diazabicyclo[8.8.8]hexacosane), potassium salt and [¹⁸F]fluoride ion; (3) deprotection of the intermediate, tetraacetylated-[¹⁸F]fluorodeoxyglucose ([¹⁸F]FTAG 10) to afford [¹⁸F]FDG 11 after either aqueous acidic or basic hydrolysis. Presently, [¹⁸F]FDG is produced using automated macroscopic radiochemical synthesizers⁷⁷, and a single production run can usually yield tens to hundreds of patient doses depending on the starting radioactivity. However, this practice is impractical for preclinical and clinical research, which requires much smaller doses and more flexible delivery schedules. Additionally, from an economic and safety perspective, the consumption of significantly less precursor and starting with lower radioactivity in each run is highly desired. Given these advantages, microfluidics-based platforms are indeed very suitable for PET probe syntheses. The two types of microfluidic platforms, continuous flow-based or batch-based, have been used to synthesize [¹⁸F]FDG. We summarized several representative examples below.

Scheme 1.

Radiochemical synthesis of 2-deoxy-2-[¹⁸F]fluoro-D-glucose ([¹⁸F]FDG).

1. Continuous Flow-based Microfluidics for [¹⁸F]FDG Synthesis

Gillies^{78, 79} et al. reported using two identical disc-shape polycarbonate microreactors for [¹⁸F]FDG synthesis (Figure 4a). This device was constructed by three layers of polycarbonate bonded together using SU-8, a UV-sensitive polymer as an adhesive, to form microfluidic channels and a reaction chamber. The top layer was used for reagent delivery. The middle and bottom layers were etched by hydrofluoric acid to form discs 100 μm in diameter and 1 mm in depth with total internal volume of approximately 16 μL in each microreactor. These three layers are thermally bonded into a single microfluidic device and connected in sequence by fused silica capillary to allow multi-step reactions. Reagent 8 (13.5 mCi) and 9 (25 mg) in DMF were primed into the first chip to form 10. The mixture was then moved into the second chip for deprotection by sodium methoxide in methanol solution to afford the crude [¹⁸F]FDG 11 (RCY: ca 50%). The extremely short synthesis time (4–6 seconds on chip) is a unique product of the well-controlled microenvironment provided by the mircoreactor. Additionally, this platform does not require heating.

Figure 4.

Schematic illustration of two continuous flow microfluidic (a) polycarbonate based and (b) glass-based microreactor for [¹⁸F]FDG synthesis.

Steel et al.⁸⁰ reported the automated synthesis of [¹⁸F]FDG using a two-stage, serpentine-channeled microfluidfic device (Figure 4b). With a flow rate of 250 μL/min, 40% RCY of [¹⁸F]FDG is obtained. The on-chip reaction time includes: 2 minutes for ¹⁸F-fluorination at 70°C and 2 minutes for deprotection at 20°C. Using wide range of starting activities, the authors claimed successful radiosyntheses of [¹⁸F]FDG without any observed radiolysis. This system demonstrated the feasibility of using a microfluidic design for routine [¹⁸F]FDG production and resulting RCY comparable to commercially available [¹⁸F]FDG synthesizers.

Wester et al.⁸¹ recently reported a simple and low-cost capillary tube-based microfluidic device which can be integrated into a fully automated module for fast batch production of [¹⁸F]FDG. The synthesis module includes several features: (1) a simple T-shaped mixer to mix reagents for radiofluorination and deprotection reaction; (2) a microtube-based reactor for in-capillary radiofluorination, and (3) on-column basic deprotection via ion-exchange cartridge. The diameter of the capillary tube is approximately 300 μm, its length is 0.7 m, and the internal volume is approximately 50 μL. The capillaries can be made of chemoresistive plastics, such as polytetrafluoroethylene (PTFE), fluoroethylenepropylene (FEP) or polyetheretherketone (PEEK) without affecting the radiofluorination. It is reported that the optimized RCYs of 11 is 88 ±7 % at 105°C at a total flow rate of 0.3 mL/min with a reduced amount of precursor 9 of 1 mg.

Compared to the bench-scale reaction with the same radiofluorination condition (RCY: 42±5 %), it demonstrated the significant advantage of in-capillary reaction using the microreactor. After deprotection under a basic condition (0.3 M NaOH, 40°C), the crude [¹⁸F]FDG was purified and then formulated to produce patient-injectable [¹⁸F]FDG with an overall RCY of 81±4 % (37.3 mCi). The total synthesis time is about 10 minutes (excluding F-18 processing). So far, this microfluidic platform is one of the most promising microscale solutions for on-site synthesis platform that can enable a decentralized distribution model.

2. Batch-based Integrated Microfluidic Device for [¹⁸F]FDG Synthesis

In contrast to the continuous-flow microfluidic platform, Lee et al.⁴¹ reported a batch reactor-based microfluidic platform utilizing PDMS chips (Fig 5a). It exemplified a scalable integration of a simple microchannel network with functional microfluidic modules. This setup enables the execution and automation of complicated chemical reactions and biological operations in a single device. The integrated microfluidic chips possess several critical features for conducting sequential chemical processes. First, microvalves⁸² provide precise control of liquid movement. Second, reagent mixing is accomplished by a peristaltic pump improving the trapping efficiency and concentrating F-18 by nearly three orders of magnitude compared to the original solution. Third, the PDMS matrix is gas permeable allowing solvent exchange directly within the microfluidic channel upon heating. During the [¹⁸F]FDG synthesis, five processes occur in sequence: (1) F- 18 is concentrated via miniaturized anion exchange column located in a square-shaped F-18 concentration loop: the diluted [¹⁸F]fluoride from a cyclotron is trapped by anion exchange beads, followed by releasing via aqueous potassium carbonate solution into a ring-shaped reaction loop; (2) F-18 drying and solvent exchange is accomplished by removing water via evaporation and solvent exchange into anhydrous CH₃CN; (3) radiofluorination of the precursor 9 with the intermediate 10 produced by heating the reaction mixture in the reaction loop; (4) solvent exchange back to water and (5) acidic hydrolysis is accomplished by the deprotection of 10 in the ring-shaped reaction loop to produce [¹⁸F]FDG 11. The operation of each microfluidic channel is controlled by a combination of pressure-driven microvalves (i.e. (1) regular microvalves for site isolation; (2) pump microvalves composed of on-chip peristaltic pumps for fluidic metering/circulation, and (3) sieve microvalves for trapping anion exchange beads to form a on-chip F-18 concentration module).

Figure 5.

Batch reactor-based integrated microfluidic reactor for [¹⁸F]FDG Synthesis. (a) Schematic representation of a PDMS-based microfluidic reactor used in the production of [¹⁸F]FDG. (b) Four sequential steps of the [¹⁸F]FDG production performed in this device: (A) F-18 concentration. (B) drying/water evaporation. (C) [¹⁸F]fluorination. (D) deprotection.

It is worthwhile to note that platform operation is highly computer-automated, further improving and accelerating chemical synthesis with high RCY and RCP. The entire process takes about 14 minutes from loading the diluted [¹⁸F]fluoride solution, and it produces 272 μCi [¹⁸F]FDG with a RCY of 38% and a RCP of 97.6%. A simplified and enlarged microfluidic device was designed and fabricated for scaled-up production.⁸³ The device has a coin-shaped reactor with an internal volume of 5 μL. It has been reported that it can handle up to 100 mCi of [¹⁸F]fluoride and produce up to 3 mCi of [¹⁸F]FDG, an amount sufficient for several microPET studies with high RCP.

Comparison Between Different Microfluidics Platforms in [¹⁸F]FDG Syntheses

Table 2 summarizes the main points of comparison between continuous flow-based microreactors and batch reactor-based integrated microfluidic devices for [¹⁸F]FDG synthesis. Continuous-flow microfluidic platforms can easily be constructed and have parallel array configurations for additional function. They can also be integrated with purification modules (solid-phase extraction (SPE) or HPLC) for streamlining the entire PET probe production process. In most continuous-flow microfluidic chips, depending on the starting activity and reagent volume, the overall [¹⁸F]FDG synthesis time is generally less than 15 minutes, starting with radiofluorination reagent preparation [K⊃2.2.2][¹⁸F]F 8. On the other hand, the batch reactor-based microfluidic systems are highly integrated and allow incorporation of technologically innovative modules, such as ion-exchange columns, active peristaltic pump/rotary mixers and even external high-pressure reactors, to achieve favorable reaction conditions. They can routinely perform chemical reactions at a nano-liter scale with microvalve mechanisms to manipulate liquid movement and prevent cross-contamination. The first proof-of- concept device shown in Figure 5a has demonstrated the capability to yield [¹⁸F]FDG in good RCY and RCP.

Table 2.

Summary of [¹⁸F]FDG Radiosyntheses in Microfluidics.

Device Material	Microfluidic Reactor volume	Starting Activitya	Precursor Amount	Fluoriantion Conditions	Deprotection Conditions	Reaction Time
PDMS	40 nL	1 mCib	92 ng in 40 nL CH3CN	100–120 °C 80 sec	HCl 60 °C, 1 min	14 minc	97.6/38
PC	16 μL	13 mCi	25 mg in 500 μL DMF	RT 3 sec	NaOMe RT, 3 sec	6 s	50/50
glass	500 μL	1–2000 mCi	5–40 mg in 500 μL CH3CN	70 °C 2 min	NaOH 20 °C, 2 min	10 mins	N.A./20–40
capillary microtube	500 μl	48 mCi	1 mg in 125 μL CH3CN	105 °C 40 sec	NaOH 40 °C, 1 min	7 min	>95/1

Note:

^aStarting activity measured from fluorination reagent [K⊃2.2.2][¹⁸F]F;

^bStarting activity measured from the F-18 solution;

^cReaction time including F-18 concentration and radiofluorination reagent [K⊃2.2.2][¹⁸F]F preparation.

Other ¹⁸F-Labeled PET Probes Prepared by Microfluidics

Lu and Pike et al.⁶⁹ also applied the same T-junction type microfluidic chip illustrated in Figure 2 to perform ¹⁸F-fluoroethylation using 2-[¹⁸F]fluoroethyl tosylate. 3-(3-Pyridinyl)propionic acid and 2-[¹⁸F]fluoroethyl tosylate were passed by syringe pumps into the microfluidic channels via port A and B, respectively. Initially the radiofluoroethylation did not proceed at ambient temperature. Then the device was heated to 80°C and product 7 collected in port C was obtained in 10% RCY. The average residence time at the flow rate of 1 μL/min is about 12 seconds and total processing time 10 minutes. This further illustrated the flexibility and diversity of microfluidic reactors to perform the radiosynthesis of PET imaging probes.

Recently, the same group reported an efficient one-step fluorination approach to prepare N-[¹⁸F]fluoroacetyl-N-(2,5-dimethoxybenzyl)-2-phenoxyaniline 12⁸⁴, a ¹⁸F-labeled high-affinity ligand to translocator protein (TSPO) which is an 18 kDa protein located mainly in the outer mitochondrial membrane and its expression level is elevated in regions of brain inflammation associated with trauma (e.g., stroke) or neurodegenerative disorders (e.g., Alzheimer’s disease) (Scheme 2). They applied a commercially available continuous flow micro-reactor platform (Nanotek, Advion, Louisville, TN) with microcapillary tube (length: 2 mm, internal diameter: 100 μm and a total volume: 15.7 μL) as a reactor loop. It can be used alone or in series to carry out sequential reactions. In a typical experiment, ¹⁸F-fluorinating reagent 8 and substrate precursors in CH₃CN were fed into the microcapillary reactor with the same flow rate. Since the device can sustain pressure up to 300 psi, it can allow reactions under superheating conditions (e.g. up to 190°C in CH₃CN). The average residence time for a particular liquid plug inside the reactor loop is 94 seconds and the total operation time takes about 4 minutes. The optimal RCY can be obtained by screening different reactor temperature (from 30–150°C). The highest RCY (85%) of 12 was obtained at 110°C and it was used for microPET studies in rat and monkey. The major advantage in carrying out radiosynthesis of 12 in a microfluidic platform is the minimal precursor consumption that in turn reduces the amount of non-radioactive impurities produced and can greatly simplify the purification. The radiometabolite of 12, 2-[¹⁸F]fluoro-N-(2-phenoxyphenyl)acetamide 13, was synthesized with the same setup (RCY: 26–30% at 160 °C) as a standard for radio-HPLC metabolite analysis. Additionally, [¹⁸F]fallypride 14²¹ that is widely used for PET imaging of D2 receptors can also be produced in a similar fashion. The optimal RCY can be obtained by screening different conditions (e.g. the length of reactor loop (from 2 to 4 m), the reactor temperature (from 100°C to 190°C), the volume ratio of precursor to [¹⁸F]fluoride solution (from 0.5 to 3) by adjusting the flow rate (from 5 to 20 μL/min) (Scheme 2). Furthermore, 3-[¹⁸F]fluoro-5-[(2-(fluoromethyl)thiazol-4-yl]ethynyl]benzonitrile ([¹⁸F]SP203B, 16), an effective radioligand for imaging metabotropic glutamate receptor 5 (mGluR5) expression^{85, 86} was prepared from the diaryliodonium salt 15 using the same microreactor. Compared to the low RCY (4–6%) obtained from radiofluorinating the corresponding bromoaryl precursor (150 °C, 10 minutes), micro-reactor-assisted reaction gave 16 in 31% RCY (160 °C, 7.83 minutes).

Scheme 2.

Radiochemical synthesis of N-[¹⁸F]fluoroacetyl-N-(2,5-dimethoxybenzyl)-2-phenoxyaniline 12, 2-[¹⁸F]fluoro-N-(2-phenoxyphenyl)acetamide 13, [¹⁸F]Fallypride 14 and 3-[¹⁸F]fluoro-5-[(2-(fluoromethyl)thiazol-4-yl]ethynyl]benzonitrile ([¹⁸F]SP203B, 16 ) using a continuous-flow microreactor.

Other Positron-Emitter-Labeled PET Probes Prepared by Microfluidics

With the dramatic expansion of PET applications in clinical diagnosis and preclinical research, and the increasing demand of different types of PET probes over recent years, several other positron-emitting radionuclides have become available in high yields in small biomedical cyclotrons, such as non-metallic Br-76 and I-124 and metallic Ga-68, Tc-94m (t_1/2 = 4.9 hours), Cu-64 and Y-86 (t_1/2 =14.7 hours) that hold different half-life and chemical features. They are drawing great attention because either their half-lives or their production sources make their delivery significantly easier than ¹⁸F-labeled ones.^{3, 4} Their labeling conditions, especially for the metallic positron-emitting radionuclides, are often mild, which is very suitable for using PDMS-based microfluidic reactors.

Gillies et al. also expanded the applications of their three-layer polycarbonate microfluidic platform (Figure 4a) using other isotopes, such as ¹²⁴I. The device was used for the preparation of the apoptosis imaging probe, ¹²⁴I-labeled Annexin-V. Three separate syringes were respectively loaded with the Annexin-V in buffer solution, [¹²⁴I]NaI in buffer solution and Iodogen (an oxidizing agent) in acetonitrile, and pumped through the microfluidic reactor simultaneously. The labeling efficiency was about 40% after 2 minutes and comparable with conventional methods over the same timeframe. The anti-cancer drug doxorubicin was labeled with ¹²⁴I via the one-step radioiodination of the corresponding butyl-tin precursor in a similar way within the microfluidic chip. The tributyl stannous precursor, [¹²⁴I]NaI and oxidizing agent N-chlorosuccinamide were pumped through the chip and the iodinated [¹²⁴I]doxorubicin iodobenzoate (I-DOXIB) with approximately 80% labeling efficiency was obtained within 2 minutes.^{78, 79}

Although reports on the preparation of PET probes labeled with metallic positron-emitter radioisotopes are scant in the literature, we believe the highly flexible batch-based integrated microfluidic reactors fabricated by PDMS will play an important role in this category because of their mild labeling conditions in aqueous solution.

Summary and Outlook

In the last few years, the utilization of microfluidic reactors in synthesizing PET probes has attracted great attention as a novel means of streamlining radiolabeling protocols. Various examples of microfluidic-based PET radiochemistry have focused on reactor design and proof-of- principle reactions, as in the reported syntheses of ¹¹C- and ¹⁸F-labeled probes. Furthermore, the multi-step radiosyntheses of [¹⁸F]FDG have been accomplished by several groups using microfluidic reactors with different designs. Using microfluidic reactors offer many benefits such as: increased product yields and purities; accelerated reaction kinetics; superior reaction control; high production reliability; improved safety; more efficient use of hot cell space or less shielding material; use of less precursor for saving precious material and a reduced separation challenge and waste; low-cost, interchangeable, disposable and quality-assured reagent cassettes and microreactors for radiochemical processes; facile automation and integration with downstream processing; and decreased use of reaction volumes to increase concentration of radioisotope that is difficult to achieve at macroscales. The combination of all of these features is critically important and essential for cGMP and future GMP for PET drug production of clinical applications. Additionally, an entirely new avenue for studying fundamental radiochemical reactions with similar stoichiometry of both radioisotope and precursor is opened.

The development of microfluidic reactors for performing continuous-flow organic reactions on a small scale have obvious potential in the area of radiolabeling for PET. Although some of platforms are amenable to routine use, several performance issues should be rectified prior to commercial deployment. For example, an inherently low Reynolds’ number causes low reagent mixing efficiency which requires additional mixing apparutus.⁸⁷ In addition, before establishing a desired continuous flow “equilibrium state”, significant amounts of reagents can be wasted. Furthermore, the possibility of cross-contamination between different reactions prevents sequential syntheses of different high-purity radiopharmaceuticals.^{88, 89} The development of batch-based microfluidic reactors can provide potential solutions to issues found in continuous flow platforms. However, despite their advantages over the continuous-flow systems, batch-based reactors are usually made of plastic elastomeric materials, such as PDMS, which are not compatible with most organic solvents or reagents and thus can’t sustain harsh chemical reaction conditions.⁵¹ Additionally, because of the serial nature of the system, the total amount of product generated is limited to one probe for a single batch production. Nevertheless, once the material requirements can be met, the direct transition to new chemoresistive elastomers or plastics can be made. It is conceivable that a highly automated batch-based microfluidic platform is likely to be the most promising solution for “single patient dose on demand” for onsite syntheses and will truly enable the decentralized model to fulfill clinical, translational and research PET needs.

An interesting idea to explore is to confer digital control over a continuous-flow system to generate sizable droplets whose volumes are similar to that of a batch microreactor. This can be accomplished by incorporating integrated microvalves into a microchannel network, thus enabling easy generation and manipulation of single droplets as well as optimizing reaction conditions within the microenvironment.⁹⁰ In a single droplet mode, this platform would work the same as batch microreactors; in a multiple droplet mode, it can assimilate continuous-flow reactions and can easily be scaled up to produce multiple single patient doses. This hybrid design could be modular and suitable materials can be chosen for different areas of the platform to fulfill operational requirements. This design is possible to circumvent the need to satisfy a broad range of very stringent requirements within a single material.

The over-arching purpose of microfluidic technologies for PET probe production is to create scalable, flexible and programmable platforms capable of preparing various specific PET probes to target and understand human diseases. Although a number of issues remain, many aspects of microfluidics have a great potential and are an ideal platform for performing the rapid productions of a wide spectrum of PET imaging probes. With continuous improvements in microfluidic technology, cost-effective, integrated, easy-to-use microfluidic radiochemistry platforms are beginning to enable a number of different medical and biological fields formerly limited from PET use. Along with a new distribution model supported by existing radiopharmacy networks to deliver common radioisotopes, such as F-18, the concept of “single-dose on demand” can indeed be realized. Next generation microfluidic platforms will have highly innovative designs and be composed of a number of advanced materials that satisfy diverse radiochemistry requirements for PET probe discovery. By enabling function of these technologies, a number of valuable existing and new probes will come to play an important role in advancing PET. This transition of current operational and logistical practices to accommodate decentralized PET probe production and distribution and the use of automated modular microfluidic platforms can potentially revolutionize the development and use of next generation of PET probes. The further development of microfluidic technology could empower biologists and clinicians to routinely use radiolabeled probes in their research could, which will be of tremendous benefit to fundamental science and medical research. This review is merely to herald a new era of microfluidic-based radiosyntheses. In summary, the numerous attributes of the microfluidic-based platforms suggest that it could become the core technology underlying the next generation of radiochemical synthesizers, making the outlook for microfluidic-based radiochemistry extremely promising.