Two decades of [¹¹C]PiB synthesis, 2003-2023: a review

Abstract

Because carbon-11 (¹¹C) radiotracers cannot be shipped over long distances, their use in routine positron emission tomography (PET) studies is dependent on the production capabilities of individual radiochemistry laboratories. Since 2003, ¹¹C-labeled Pittsburgh compound B ([¹¹C]PiB) has been the gold standard PET radiotracer for in vivo imaging of amyloid β (Aβ) plaques. For more than two decades, researchers have been working to develop faster, higher-yielding, more robust, and optimized production methods with higher radiochemical yields for various imaging applications. This review evaluates progress in [¹¹C]PiB radiochemistry. An introductory overview assesses how it has been applied in clinical neurologic imaging research. We examine the varying approaches reported for radiolabeling, purification, extraction, and formulation. Further considerations for QC methods, regulatory considerations, and optimizations were also discussed.

Introduction

Alzheimer’s disease (AD) is the most common neurodegenerative disorder associated with progressive memory loss and cognitive impairment AD is linked to the majority of dementia cases [1,2]. Globally, AD care provision was estimated to be over $600bn USD per annum worldwide [2].

Amyloidal plaques have been identified as a biomarker for AD in the brain [3]. Amyloid plaques are strings of amino acids known as amyloid-β (Aβ) peptides, having a common length of 40 or 42 [4,5]. Aβ₄₂ is far more prevalent in AD patients than in healthy individuals and tends to aggregate together and eventually forms larger β-pleated sheet [5]. These Aβ proteins can aggregate to form different shapes and patterns [4,6]. These Aβ aggregates have been shown to precede the onset of neurodegenerative symptoms by several years [7]. Thus, the ability to selectively image Aβ plaques could enable early diagnosis and risk prediction for AD prevalent individuals [8,9], there are a number of studies that demonstrate the feasibility of early AD diagnosis [10-13] which would also allow patients and caregivers to be better prepared.

[¹¹C]PiB was the first broadly applied radiotracer with specificity for Aβ aggregates in humans. It has a high affinity and selectivity for Aβ plaques and other Aβ containing lesions [14]. PiB is an analog of thioflavin-T, a dye used for staining amyloid in brain tissue [15].

Mathis et al. first reported [¹¹C]PiB in 2003, as part of a preclinical evaluation of structurally similar compounds built on a 2-arylbenzothiazole backbone, abbreviated as [¹¹C]6-OH-BTA-1 and also known as [¹¹C]N-methyl-2-(4’-methylaminophenyl)-6-hydroxybenzothiazole [16]. The first clinical [¹¹C]PiB PET study with was published a year later; the radiotracer provided quantitative information on Aβ plaques in 25 participants [15]. It is now regarded as the gold standard for Aβ plaque PET [2,17].

Radiopharmaceutical synthesis requires specialized equipment such as a cyclotron, hot cell, an automated radiochemistry module, preparative high-performance liquid chromatography (HPLC), radiation detectors, and a sterile (ISO 5) environment for peripheral use [18]. Several challenges impede commercialization of ¹¹C-radiotracers; the most prominent is the short half-life of radioactive decay (T_1/2 = 20.4 min) necessitating onsite radiosynthesis (refer to Figure 1). The most significant optimizations have focused on improving yield, shortening the radiochemistry duration or circumventing a step altogether minimizing points for failure and reducing reaction time. From a radiochemistry and automation point of view, fewer synthesis steps will facilitate automation and reduce potential failure points [19-21].

Figure 1.

Typical timeline of [¹¹C]PiB radiosynthesis. The image shows a simplified timeline of how [¹¹C]PiB is produced and highlights important time stamps. The extended portion of PiB synthesis shows each step, such as the radiolabeling and HPLC steps. References of step removing optimizations are quoted.

Optimizations reported in the literature

Molar activity (MA) or specific activity (SA) is expressed as the ratio of radioactivity (GBq) per amount of tracer or compound (μmol or µg respectively) and are thus interchangeable units. ¹¹C decays exponentially, leading to a drastic loss of activity and thus, MA over time [22]. Generally, the entire radiotracer synthesis and QC process should take no longer than 60 minutes to avoid low counts on the PET scanner that can result in noisy data [22]. Thus, shortening the total synthesis time and optimizing the QC and release processes significantly increases MA at the injection time [23]. Further, isotopic dilution of the ¹¹C-compound with a “cold” carbon-12 equivalent during the radiolabeling procedure or synthon generation would decrease MA. While isotopic dilution of [¹¹C]CO₂ with atmospheric CO₂ is hard to avoid [19], other synthesis pathways for example starting with [¹¹C]CH₄ are far easier to keep isotopically pure. “Cold” methane would have to be introduced into the system for isotopic dilution to occur [20,24]. A higher MA improves the images quality of the eventual PET scan and has an impact on subject safety [25,26] and can vary between production sites [25].

In this review, we evaluate progressive optimizations of [¹¹C]PiB radiochemistry. We report different published approaches on all critical aspects including radiolabeling, purification, and formulation. Further, we discussed the QC requirements and regulatory guidelines in clinical research.

Literature review

Radiolabeling pathways

The literature reports three general radiolabeling reactions (Figure 2) for [¹¹C]PiB, employing either N-methylation agents [¹¹C]methyl Iodide ([¹¹C]MeI); [¹¹C]methyltrifluoromethanesulfonate ([¹¹C]MeOTf); or alternatively [¹¹C]CO₂ fixation-reduction, depending on the auto synthesis unit and the desired level of radioactivity [27]. The original [¹¹C]PiB synthesis method used [¹¹C]MeI as the synthon; a methoxymethyl-protected alcohol group on the precursor was later removed through acid hydrolysis [16]. The second approach used [¹¹C]MeOTf as the methylation agent, eliminating the need for deprotection and significantly reducing the required synthesis time, while improving the overall radiochemical yield [28]. More than a decade passed before a third radiolabeling pathway for [¹¹C]PiB synthesis was reported, using [¹¹C]CO₂ in a one-pot catalyzed reaction [19].

Figure 2.

Three general radiolabeling approaches for [¹¹C]PiB synthesis using different ¹¹C-sources: (a) [¹¹C]MeI; (b) [¹¹C]MeOTf; (c) [¹¹C]CO₂. The diagram depicts the three major synthesized pathways for [¹¹C]PiB synthesis.

¹¹C-N-methylation synthesis pathways

Through proton bombardment of N₂ gas resulting in ¹⁴N(p,α)¹¹C nuclear reaction, a cyclotron can produce [¹¹C]CO₂ or [¹¹C]CH₄ based on the amount of O₂ (0.5-1%) or H₂ (5-10%), respectively, introduced into the bombardment cell. Target-produced [¹¹C]CH₄ has very high specific activity compared to target-produced [¹¹C]CO₂, which is later reduced to [¹¹C]CH₄ [20]. Oxygen and other radio impurities can be removed using a cryogenic trap (liquid nitrogen) in combination with molecular sieves prior to the radiosynthesis procedure [29,30]. Cyclotron-produced [¹¹C]CO₂ [31] and [¹¹C]CH₄ [24] have both been used for [¹¹C]PiB radiosynthesis.

[¹¹C]MeI is an established starting material in the production of ¹¹C labeled tracers [32-34] and it has been used for [¹¹C]PiB radiolabeling both directly [16] and as an intermediate [30,31,35]. The longstanding “wet” method for [¹¹C]MeI synthesis involves LiAlH₄ dissolved in tetrahydrofuran (THF) reacted with [¹¹C]CO₂ bubbling, resulting in [¹¹C]MeO^-/[¹¹C]MeOH which was then reacted with hydroidic acid (HI) to yield [¹¹C]MeI [33]. The inherent drawback to the “wet method” is the fact that the LiAlH₄ solution can be a source of carbon-12 isotopic dilution, considerably lowering the MA of the final [¹¹C]PiB (with MA typically ranging 50-150 GBq/μmol) [20]. Still, in the original [¹¹C]PiB synthesis, the “wet” method was used to produce [¹¹C]MeI and the reported 210 GBq/µmol is among the higher MA’s in the review [16]. Despite the “wet method” producing higher radiochemical yields an unidentified radiochemical impurity co-eluting with [¹¹C]PiB during SPE purification was identified. Boudjemeline et al. suspected a double methylation byproduct might form. Such an unidentified radiochemical impurity co-eluting with [¹¹C]PiB would make clinical radiotracer batches unsuitable for imaging applications without the use of prep-HPLC. The formation of this unidentified radiochemical purity was resolved by producing [¹¹C]MeOTf separately, using the gas method, and transferring it to the employed kit. The specific activity was also improved when using the gas method over the wet method for [¹¹C]MeOTf production [35]. Note that our group reported a clinically-viable high-quality [¹¹C]PiB wet method with adequate specific activity and purity [36].

Gas phase [¹¹C]MeI synthesis results in a MA, exceeding 300 GBq/μmol, although it presents significant technical challenges [20]. A typical method for [¹¹C]MeI production involves gas phase halogenation of [¹¹C]CH₄ with iodine. In short, the cyclotron-produced [¹¹C]CO₂ was first trapped on 4 Å molecular sieve for purification. It was then reacted with hydrogen at 350-360°C under a Shimalite nickel catalyst to yield [¹¹C]CH₄, and was collected on a liquid nitrogen cooled Porapak N trap. A phosphorus pentoxide trap was required to remove impurities formed during proton bombardment and subsequent catalytic reactions. Next, the [¹¹C]CH₄ was desorbed through heating and pushed with helium through a quartz tube where the gases are mixed with vapors from iodine crystals. Heated to 60°C, the mixture was then pushed to a reaction chamber at 720 to 725°C with helium flow. The gas mixture then passed through a room-temperature quartz tube where most of the excess iodine crystallizes. An additional ascarite trap was used to remove the rest of the iodine. The resulting [¹¹C]MeI was trapped on the Porapak N trap at room temperature, and any unreacted [¹¹C]CH₄ was recirculated through the ovens for further [¹¹C]MeI synthesis. [¹¹C]MeI was desorbed from the Porapak N trap through heating to 190°C and helium carrier gas [27,32]. This method has also been deployed with [¹¹C]CH₄ produced directly from the cyclotron [32]. When [¹¹C]CH₄ from the target and a single pass iodination procedure is used, specific radioactivity as high as 4700 GBq/μmol has been reported [20,37]. Choosing between the wet and gas phase methods for [¹¹C]MeI synthesis is a tradeoff, as the wet method typically has a higher radiochemical yield [37].

Strong alkaline conditions are required to “activate” the precursor for methylation with [¹¹C]MeI [16]. Under such alkaline conditions the primary amine was probably negatively charged leading to the reaction mechanism proposed in Figure 3.

Figure 3.

A proposed SN₂ reaction mechanism for the radio methylation of the primary amine of 6-MOMO-BTA-0, which acts as a nucleophile under basic conditions.

[¹¹C]MeOTf activity was trapped in the reaction vial loaded with the corresponding precursor, after [¹¹C]MeI was passed through a column containing graphitized carbon impregnated with 50% weight silver triflate heated to 200°C, refer to Figure 4 [27,38]. With increasing demand for new radiotracers, [¹¹C]MeOTf became one of the two primary [¹¹C]synthons, along with [¹¹C]MeI [39], and it is still widely used for novel ¹¹C-tracer clinical applications [40]. Moreover, with higher reactivity, lower volatility, and shorter reaction times [¹¹C]MeOTf greatly improves [¹¹C]PiB radiochemical yields. Lower radiolabeling temperatures are required because [¹¹C]MeOTf is a stronger methylation agent than to [¹¹C]MeI, by a factor of ~10⁴. It can readily radiolabel compounds in cases where [¹¹C]MeI yield would be insufficient [39,41]. Further, N-alkylation was so strongly dominated over O-alkylation with [¹¹C]MeOTf that the protection group of 6-MOMO-BTA-0 was no longer required [28]. Figure 5 depicts a proposed reaction mechanism for ¹¹C-N-methylation using [¹¹C]MeOTf.

Figure 4.

[¹¹C]MeOTf preparation after [¹¹C]MeI synthesis using the wet method [36]. An example reaction where [¹¹C]MeOTf was prepared from [¹¹C]MeI.

Figure 5.

A proposed SN₂ reaction mechanism for N-alkylation radio methylation of the primary amine of 6-OH-BTA-0, which acts as a nucleophile.

The graphitized carbon impregnated with silver triflate was prepared by dissolving 1 g of silver trifluoromethanesulfonate in 150 mL of anhydrous diethyl ether. After adding 2 g of Graph-pack C (C80/100 mesh), the solution was stirred under vacuum in the dark. When ether evaporates, the resulting powder was dried under vacuum (0.2 Torr) for at least 2 more hours then stored in amber glass containers [27]. Alternatively, the reagent can be purchased from ABX-Advanced Biochemical Compounds [27].

[¹¹C]CO₂ fixation-reduction

[¹¹C]CO₂ is an attractive starting material for radiolabeling as it is produced directly from the cyclotron. Using [¹¹C]CO₂ as a C₁ building block for the methylation of amines in [¹¹C]PiB synthesis has been reported twice [17,19]. Despite its desirable characteristics, very few studies have used this synthetic pathway in a clinical setting [40].

The catalysts used for CO₂ fixation were originally developed for carbon recycling or capture [42]. CO₂ hydrogenation usually requires metal catalysts and high temperature and pressure, but the authors report an alternative organocatalytic reaction using silane as the reductant and nitrogen base as catalysts [42]. In another study, a series of zinc catalysts to facilitate N-methylamine synthesis using CO₂ as a C₁ building block were reported [43]. One of the reported catalysts, ZnCl₂, was used to synthesize [¹¹C]PiB directly from [¹¹C]CO₂ [19]. Figure 6 illustrates similarities with the methylation pathways, shown in Figures 3 and 5.

Figure 6.

Hypothesized reaction steps facilitating [¹¹C]CO₂ fixation-reduction of the primary amine of 6-OH-BTA-0 (adapted from Liger et al. [19]).

Another group later reported using milder reaction conditions and an organic PhSiH₃/TBAF catalyst mixture to achieve [¹¹C]CO₂ fixation-reduction for [¹¹C]PiB radiolabeling [17].

Radiolabeling strategies

There have been several reaction conditions and methodologies reported in literature for [¹¹C]PiB synthesis. Table 1 summarizes findings for [¹¹C]PiB synthesis reported including reaction conditions i.e., temperatures, solvents, and reaction times with their associated yields in chronological order of publication.

Table 1.

Summary of radiolabeling conditions of reported [¹¹C]PiB synthesis

Precursor	11C-synthon generation	Radiolabeling conditions	Radiolabeling temperature (°C)	Radiolabeling time (min)	Yield (%)	MA** (GBq/µmol)	Reference (chronological)
6-MOMO-BTA-0 (1.5 mg)	[11C]CO2→[11C]MeI (Wet method)	DMSO (0.4 mL) + 10 mg KOH	125	5 + 5 (deprotection)	28.2 (of synthon)	210	[16]
6-OH-BTA-0 (0.4 mg)	[11C]CO2→[11C]MeOTf (Wet method)	Loop reaction; MEK (0.08 mL)	20	1	11-16 (of [11C]CO2)	30-60	[28]
6-OH-BTA-0 (4-8 mg)	[11C]CO2→[11C]MeOTf (Gas method)	Acetone (0.5 mL) cooled to -20°C during bubbling	80	1	44 ± 10 (of synthon)	80-120	[46]
6-OH-BTA-0 (1 mg)	[11C]CO2→[11C]MeOTf (Wet method)	Loop reaction; Cyclohexanone (0.08 mL)	RT*	1	13-15 (of [11C]CO2)	20-60	[30]
6-OH-BTA-0 (1 mg)	[11C]CO2→[11C]MeOTf (Wet method)	MEK (200 µL), with bubbling	80	1	18 (of [11C]CO2 decay corrected)	9.25	[45]
6-OH-BTA-0 (1 mg)	[11C]CH4→[11C]MeOTf (Gas method)	Loop reaction; acetone/acetonitrile (0.1 mL 1:1 mixture)	RT*	1	32.9 (from [11C]MeI decay corrected)	143 ± 26	[24]
6-OH-BTA-0 (2 mg)	[11C]CO2→[11C]MeOTf (Gas method)	0.5 mL MEK (cooled to 13°C during bubbling)	75	2	48 (of synthon decay corrected)	183 ± 14	[21]
6-OH-BTA-0 (1 mg)	[11C]CO2→[11C]MeOTf (Gas method)	Loop reaction; 3-pentanone (0.1 mL)	RT*	5	1.6 (of [11C]CO2)	265.5	[44]
6-OH-BTA-0 (1 mg)	[11C]CO2→[11C]MeOTf (Wet method)	Loop reaction; acetone/acetonitrile (0.1 mL 1:1 mixture)	RT*	1	Not reported	25 ± 10	[31]
6-OH-BTA-0 (1.7 mg)	[11C]CO2	0.4 mL Diglyme + ZnCl2 (1.3 mg) + IPr (3.5 mg) + PhSiH3 (23 μL) cooled to 0°C during bubbling	150	20	38 (of [11C]CO2)	15	[19]
6-OH-BTA-0 (0.5-1 mg)	[11C]CO2→[11C]MeOTf (Gas method)	0.3-0.5 mL MEK (cooled to 13°C during bubbling)	75	2	35-50 (of synthon decay corrected)	44.4-107.3	[27]
6-OH-BTA-0 (0.1-0.3 mg)	[11C]CO2→[11C]MeOTf (Gas method)	C18 Solid phase supported radiosynthesis: acetone (50-150 µL)	RT*	2-3	22 (of synthon)	190	[35]
6-OH-BTA-0 (0.6-0.9 mg)	[11C]CO2→[11C]MeOTf (Gas method)	Acetone (cooled to -40°C during bubbling, 3 min)	110	2	Not reported	≥ 7.07 specification, MA not reported	[7]
6-OH-BTA-0 (1.2-4.8 mg)	[11C]CO2	Diglyme (0.5 mL) + PhSiH3 (37 µL)/TBAF 1.0 M (10 µL)	150	1	14.8 ± 12.1 (of [11C]CO2 decay corrected)	61.4 ± 1.6	[17]
6-OH-BTA-0 (1 mg)	[11C]CO2→[11C]MeOTf (Wet method)	C18 Solid phase supported radiosynthesis: acetone (500 µL)	65	1	9.8 ± 1.7 (of [11C]CO2)	57 ± 18	[36]
6-OH-BTA-0 (0.3 mg)	[11C]CO2→[11C]MeOTf (Gas method)	C18 Solid phase supported radiosynthesis: MEK (150 µL)	RT*	1.5	10.13 (calculated from synthon activity reported)	824.8 ± 177	[26]

^*RT, Room temperature.

^**MA, molecular activity - converted to GBq/µmol if reported in another unit.

Reporting yields in Table 1 was a little challenging as some yields were reported decay corrected, and some as percentages of the synthon, either [¹¹C]MeI or [¹¹C]MeOTf, whereas others are reported non decay corrected from the cyclotron-produced [¹¹C]CO₂. Table 1 further includes the reported MA for each procedure; showing no strong correlation with any particular synthesis method. For example, the loop reactions have MA ranging from 20-265.5 GBq/µmol. Note for Table 1 that if the alcohol group on the precursor was protected with a methoxymethyl (6-MOMO BTA-0 + [¹¹C]MeI reaction) an additional acidic deprotection step was required [16]. Shao et al. reported a MA value of 7177 Ci/µmol; we corrected this evident misprint to 7.177 Ci/µmol as this value would have been three orders of magnitude higher than all the other reported MA values [44]. The yield of Nair et al. was calculated to report in Table 1, based on the reported values of 3547 MBq average product and 33-37 GBq as the typical output of [¹¹C]MeOTf [26]. Figure 7 displays a simplified flow diagram of all the [¹¹C]PiB synthetic pathways reported.

Figure 7.

Generalized radiolabeling pathways for [¹¹C]PiB methods as described in Table 1. The simplified synthesis pathways summarizes all different approaches used for [¹¹C]PiB synthesis that have been reported in literature.

Automation of [¹¹C]PiB synthesis

Since the original [¹¹C]PiB synthesis was performed in a V-vial heated with an oil bath (125°C) two decades ago [16], research has driven toward full automation of radiochemistry. Cheung and Ho reported a non-commercial, in-house automated apparatus for remotely operated [¹¹C]carboxylation and bubbling [¹¹C]methylation procedures, including [¹¹C]PiB synthesis [45]. The automatic synthesis module, Tracer Lab FX-C Pro (GeneralElectric) can convert [¹¹C]CO₂ to [¹¹C]MeI through the gas-phase reaction. It then produces [¹¹C]CH₃OTf and facilitates the ¹¹C-N-methylation radiolabeling reaction and SPE (Sep Pak elution) reformulation. Although two FXC-Pro studies used another supplier’s semi preparative-HPLC system, they were still controlled by the same FX-C software package [27,44]. A fully-automated [¹¹C]CO₂ fixation-reduction radiosynthesis procedure was carried out using a TRACERlab^® FX C PRO module from GE Healthcare, thus successfully circumventing the need to produce [¹¹C]MeI or [¹¹C]MeOTf before radiolabeling [17].

Clemente et al. 2012 reported an automated [¹¹C]PiB radiosynthesis combining three Bioscan systems, [¹¹C]CO₂ conversion to [¹¹C]MeI used the wet method and subsequently, [¹¹C]MeOTf synthesis. They then used the loop method for radiolabeling and semi preparative HPLC purification [31].

A recent trend in the automation of PET radiochemistry uses cassette-based kits for production of various radiotracers on the same module with few technical modifications between syntheses [35]. Further, these single-use GMP-grade cassettes adhere to regulatory guidelines for clinical use and obviate cleaning and drying of the module between consecutive batches [35]. Boudjemeline et al. 2017 reported the first cassette-based [¹¹C]PiB production and excluded preparative-HPLC; the same Sep Pak C18 cartridge performed purification and formulation to improve automation and routine application of [¹¹C]PiB PET. Another dual method was the use of Synthra auto-synthesis module to convert [¹¹C]CO₂ to [¹¹C]CH₃OTf. It was then then transferred to the cassette-based Scintomics GRP automated module for [¹¹C]PiB radiolabeling and purification [35].

Our group reported a fully-automated, single module, cassette-based [¹¹C]PiB synthesis procedure for both [¹¹C]MeOTf preparation and [¹¹C]PiB radiolabeling on the Trasis AllinOne with MA acceptable for clinical application [36].

Recently a GE FastLab kit method was reported that could handle two consecutive [¹¹C]PiB productions on the same cassette, thus increasing the number of productions, with no burden on the infrastructure. This method required separate production(s) of [¹¹C]MeOTf on GE TracerLab FXc Pro, which was transferred to the kit for further production steps. From Table 1, this method also had the highest reported MA [26].

Steps to improve [¹¹C]PiB radiolabeling yields and specific activity

Significantly higher radiochemical yields (44 ± 10%) were reported when [¹¹C]MeOTf, rather than [¹¹C]MeI, was used as ¹¹C-methylation agent [46]. Anhydrous Na₂SO₄ or 4 Å molecular sieves can be applied as drying agents to improve yields of the water sensitive [¹¹C]MeOTf radiolabeling reaction [47]. MgSO₄ has also been reported as a desiccant for the cyclohexanone solvent used during the methylation step [30].

Preparation and dilution of the reagents inside an inert atmosphere glove bag (or box) ensures high specific activity of the final [¹¹C]PiB product, specifically when LiAlH₄ is used in the synthesis [45].

To further enhance anhydrous conditions, the reactor can be rinsed with acetone, flushed with helium and dried under vacuum at 100°C for 30 min. The reactor was then cooled to 25°C and kept under a positive pressure and ultra-pure helium, before delivering radioactivity to the auto synthesis module, to perform the radiochemistry [17]. Blow drying the auto synthesis unit’s lines with nitrogen gas (1 bar) for 1 hour prior to radiolabeling has also been reported to increase yields [30].

Bubbling ¹¹C-methylation reaction

The first methodology used in [¹¹C]PiB radiolabeling was bubbling ¹¹C-methylation [16] and it is still being used [35]. Some reported examples of the reaction mixture being cooled (13°C MEK and -20°C acetone) prior to transfer and improve the capture of [¹¹C]MeOTf during the bubbling process [21,46].

Loop ¹¹C-methylation reaction

The loop reaction, or captive solvent method has been reported for both [¹¹C]MeOTf and [¹¹C]MeI radiotracer methylation reactions [47] including PiB radiolabeling [28]. Briefly, the normethyl precursor (0.4 mg), dissolved in 80 µL dimethyl formamide (DMF) or DMSO or MEK with base added (for [¹¹C]MeI) and the crude reaction mixture was loaded onto a semi-preparative HPLC injection loop (2 mL) set to the load position. The [¹¹C]MeOTf or [¹¹C]MeI was then passed through the coated HPLC loop with a stream of N₂ gas at room temperature. A proximal γ detector monitors radioactivity captured, after a peak was observed, N₂ flow ceases; the loop position was switched to inject position, and the contents are transferred onto the HPLC purification column [28,44,47,48].

Solid phase supported ¹¹C-methylation reaction

Sorbent-supported radiolabeling is well-established and has been used to produce several ¹¹C-tracers [49] including [¹¹C]PiB [26,35,36]. The process involves passing ~100 µL of 6-OH-BTA-0 (~2 mg/mLin acetone) over a disposable Sep-Pak C18 cartridge under a stream of N₂ or Ar, effectively drying the acetone and leaving only the precursor. The radiolabeling method involves passing gaseous [¹¹C]MeOTf through the preloaded Sep-Pak C18 cartridge. Once a proximal radiodetector determines that the maximal amount of [¹¹C]MeOTf was trapped on the Sep-Pak cartridge, it was allowed to react with the precursor for 2 minutes at room temperature or heating to 65°C [35,36]. The final [¹¹C]PiB was then eluted for further purification and/or formulation.

Preparative HPLC and purification methods for [¹¹C]PiB

Considering pKa values of PiB are essential to ensure sharp peak shapes in HPLC and trapping on SPE cartridges, the phenol had a pKa value of 9.3, and methylamine (protonated) group was 3.0 [16]. Table 2 summarizes all reported prep-HPLC or other purification technique parameters.

Table 2.

Preparative HPLC or other purification technique conditions

Mobile phase	pH	Phase (Column)	Flow rate (ml/min)	Reported retention time (min)	Reference (chronological)
35% Acetonitrile/65% triethylammonium phosphate buffer	7.2	C18 (Phenomenex Prodigy 10 µm ODS-prep 10 × 250 mm)	15	11.5	[16]
40% Acetonitrile/60% aqueous 0.1 N NH4HCO2	Not reported	C18 (Phenomenex Luna 10 µm, 10 × 250 mm)	9	9	[28]
75% Acetonitrile/25% aqueous 0.2% triethylammonium acetate	7.3	C18 (Supelco ABZ+ 5 µm, 250 × 10 mm)	8	4.5	[46]
50% Acetonitrile/50% water	Not reported (± 7)	C18 (Waters, 7 µm, 7.8 × 300 mm)	3	7	[30]
40% Acetonitrile/60% aqueous 0.1 N ammonium formate	Not reported	C18 (Alltech Alltima 5 µm; 250 mm × 10 mm)	7	10-11	[45]
60% Acetonitrile/40% aqueous 0.01 M sodium acetate	9	C18 (Macherey-Nagel Nucleosil 100-7 C18, 250 × 16 mm)	8	8-9	[24]
60% Acetonitrile/40% aqueous 0.1 M ammonium acetate/0.2% 2 M triethylammonium acetate	Not reported	C18 (Supelcosil LC-ABZ+ 5 µm, 250 mm × 10 mm)	8	4.9	[21]
40% Acetonitrile/60% aqueous 50 mM ammonium acetate	Not reported	C18 (Phenomonex Luna C18, 150 × 10 mm)	5	~8	[44]
40% Acetonitrile/60% aqueous 0.1 N ammonium formate	Not reported	C18 (Phenomenex Luna 5 µm, 10 × 250 mm)	9	7-8	[31]
40% Acetonitrile/60% water	Not reported (± 7)	C18 (Waters Symmetry-Prep 7 μm, 7.8 × 300 mm)	4	11	[19]
40% Ethanol/60% 0.01 M trisodiumcitrate; 0.1% ascorbic acid	Not reported (acidic)	C18 (Phenomenex Luna 10 µm, 10 × 250 mm)	6	13-16	[27] Method A
40% Acetonitrile/60% 0.01 M-trisodiumcitrate; 0.1% ascorbic acid	3.5	C18 (Agilent Eclipse XDB 5 µm ODS-prep 9.4 × 250 mm)	4.5	9-11	[27] Method B
2.5 mL of 50% EtOH/acetate buffer	3.7	C18 SPE purification was achieved through selective washing and elution			[35]
50% Acetonitrile/50% water	Not reported (± 7)	C18 (Phenomenex Luna 5 μm 100 Å 250 × 10 mm)	4	8.5-9.5	[17]
60% Acetonitrile/40% water	Not reported (± 7)	C18 (Waters XBridge BEH Shield RP18 OBD, 130 Å, 5 µm, 10 × 250 mm)	5	4.5	[36]
2.5 mL of 50% EtOH/acetate buffer	3.7	C18 SPE purification was achieved through selective washing and elution			[26]

Reformulation

According to USP 823, facilities, equipment, and procedures must ensure a sterile PET product, although its sterility can only be tested through incubated growth after use. The chemical synthesis of a parenterally administered radiotracer may take place in an open or closed system, so some steps can be performed on a bench top. However, all activities downstream of the 0.22 µm membrane filter should be conducted in a closed, sterile system to minimize contamination risk. Radiopharmaceuticals must be dispensed in a clean room, hot cell, biosafety cabinet, or other environment with an ISO 5 or lower rating [50].

Reformulation steps for [¹¹C]PiB were very standardized, with little-to-no variation reported in the literature. The fraction collected from the preparative HPLC was diluted with water (20-100 mL), trapped on a C18 SPE cartridge, optionally washed with water (0-15 mL), eluted with ethanol (1 mL), and then diluted with physiologic solution or saline to reduce the ethanol concentration to ≤ 10%. The final product was then filtered through a 0.22 µm membrane filter into the final product vial [21,24,46]. Some groups add 1 mL of phosphate buffer (125 mM) [21] or another fraction of saline [36] to the final product. In an exception to the standardized reformulation method, radiolabeling was carried out and the reaction mixture purified and formulated on the same C18 SPE cartridge, using a 50% ethanol solution as eluent [26,35].

Total synthesis time comparison

For routine preparation and high resolution [¹¹C]PiB PET images, the total synthesis time is important; the longer the radio decay period, the lower the MA expressed as GBq/µg [25]. For saturable processes (e.g., receptors on proteins) competition between the [¹¹C]PiB and indistinguishable cold PiB impairs the quality of the PET scan data. Therefore, MA has been identified as a key factor in both preclinical and clinical studies dealing with tracer-binding densities [23].

Total synthesis time for recent automated procedures ranges from 20 to 35 minutes as measured after the end of bombardment. For examples of total synthesis time was ~25 min [36]; 32 min [17]; 20 min [35]; 30 min [27]; 25-30 min [31]; 35 min [26]. The fastest reported synthesis method of 20 min, did not include preparative HPLC, indicating the efficiency of optimized SPE purification and formulation methods [35]. The slowest method had reported the highest MA [26].

We believe that most of the radiolabeling strategies (Table 1) can be used with any of the purification methods (Table 2), which might lower the total synthesis time significantly. A laboratory considering implementing a new synthesis strategy for [¹¹C]PiB or optimizing their existing production method, can identify a radiolabeling and purification method that best suits their facility needs.

Quality control (QC) considerations

Every batch of [¹¹C]PiB intended for human use must undergo around 9 QC tests before it can be administered. They include: 1) visual inspection and assess, 2) pH, 3) radiochemical purity, 4) radionuclide identity, 5) PiB identity and concentration, 6) molar activity, 7) residual solvent analysis, 8) chemical purity, and 9) stabilizer concentration (if any) [50]. Every production facility must meet parameters and document specifications for each [¹¹C]PiB synthesis procedure prior to administering a batch [18].

Optimizing the HPLC-UV method can significantly improve MA and activity concentrations at the time of injection [23]. The HPLC-UV chromatogram should show clear separation of the solvent front, precursor and PiB peaks at baseline level [22]. Several strategies can reduce HPLC-UV QC run time; for example, by increasing the flow rate and temperature, while reducing the particle size and column volume using a thinner or shorter column. A compliant [¹¹C]PiB HPLC-UV can be run in as little as two minutes [23]. The literature reports an upper limit of 13.4 µg of PiB per injection and with a lower limit of 1.34 µg of 6-OH-BTA-0 per injection [7,26]. We have recently published a review on the QC design process for ¹¹C-tracers which highlights all the aspects of each of nine QC tests [51].

Releasing a [¹¹C]PiB batch for a clinical PET scan

After the appropriate QC tests are completed and the data and documentation have been reviewed [52], a designated qualified expert authorizes the use of the [¹¹C]PiB final batch by dated signature. Certain QC tests, such as sterility tests can be performed as post-release controls, provided the synthesis procedure has been well validated [7,50]. A conditional release or reprocessing may be possible if one of the QC tests could not be completed or obtained a questionable result but only under certain conditions and with written procedures in place [50,52].

Overview of [¹¹C]PiB PET imaging applications

[¹¹C]PiB is the most commonly used Aβ plaque imaging radiotracer due to its fast uptake and low nonspecific binding in the human brain [17,27]. Many clinical studies support its effectiveness in correlating of Aβ plaques with Alzheimer’s disease (AD) progression and mild cognitive impairment (MCI) [53]. Zhang et al., reported a systematic review on the use of [¹¹C]PiB PET for the early diagnosis of AD and an assessment of the effectiveness and accuracy of early diagnosis. The same group also concluded that a standardized threshold for a positive result was required before [¹¹C]PiB PET could be widely applied to clinical practice [54]. A few studies reported a comparison between [¹¹C]PiB PET retention longitudinal assessment of Aβ deposition of healthy older adults with those of AD patients [55,56]. [¹¹C]PiB PET imaging can distinguish AD from frontotemporal dementia [57] and frontotemporal lobar degeneration [58]. Aging effect of Aβ deposition in non-demented adults with down syndrome has been studied using [¹¹C]PiB PET, indicating an age-related deposition Aβ plaques in the down Syndrome population [59].

A proposed amyloid quantification index (AQI) can reportedly differentiate MCI from AD in otherwise clinically indistinguishable AD cases. A 30-min [¹¹C]PiB PET scan can estimated AQI based on clearance rate and mid-phase PiB retention in featured brain regions [60].

Comparison to other radiotracers

Studies have compared the accuracy, specificity, and selectivity of [¹⁸F]FDG versus [¹¹C]PiB for diagnosing or predicting AD. A review and statistical analysis found that PiB was more sensitive in predicting MCI progression to AD, but might have less specificity [53]. Later studies suggest a benefit in combining and interpreting the imaging data from [¹⁸F]FDG and [¹¹C]PiB PET scans [61]. [¹¹C]PiB PET’s ability to identify AD and to distinguishing it from MCI has also been compared to [¹⁸F]FDDNP. It was reported that for the identification of AD and differences in non-displaceable binding potential between patients with AD, patients with MCI, and controls were more pronounced using [¹¹C]PiB PET. Difference in regional binding were moderately correlated suggesting that the two radiotracers adsorb to related, but different characteristic brain regions of AD [62].

The FDA approved [¹⁸F]flutemetamol (FMM) radiotracer is structurally related to [¹¹C]PiB (refer to Figure 8). A strong correlation in standardized uptake value ratio (SUVR) was reported comparing [¹¹C]PiB with previously scanned [¹⁸F]FMM subjects, however the SUVR for [¹⁸F]FMM PET in the AD group was lower [63]. Another similar study reported when a standardized visual read technique was applied to classify scans as Aβ-positive or Aβ-negative [¹⁸F]FFM was reported to have higher white-matter retention, and care must be taken not to interpret it as a false-positive result [64]. White matter is highly lipidic with the tissue mainly composed of myelin [65] and because [¹⁸F]FMM is more lipophilic than [¹¹C]PiB, it has been suggested non-specific uptake might be a key factor for the increased uptake, rather than actual/specific target protein uptake [66]. This was also studied for [¹¹C]PiB both in vitro and in vivo [67] conditions. Studying uptake of [¹¹C]PiB in white matter has been proven to be useful in identifying lesions and studying multiple sclerosis [68] non-specific for AD. Targeted binding of [¹⁸F]FMM to Aβ protein was also characterized as a slow kinetic component [66].

Figure 8.

Aβ-plaque PET imaging agents approved by the US Food and Drug Administration. The chemical structure of all the FDA approved Aβ-plaque PET imaging agents, for visual comparison with the [¹¹C]PiB structure.

Although [¹⁸F]Florbetaben is not structurally related to [¹¹C]PiB (Figure 8), it exhibits a high affinity for Aβ-plaques and can distinguish between AD positive and negative patients. However, [¹⁸F]Florbetaben has been reported to have a lower degree of retention, for a narrower dynamic range of SUVR values than [¹¹C]PiB. This was demonstrated through total SUVR for [¹¹C]PiB in AD patients that was 75% greater than in healthy subjects while only 56% greater for [¹⁸F]Florbetaben [69].

Further, [¹⁸F]florbetapir or [¹⁸F]AV-45, was found to have similar binding characteristics compared to [¹¹C]PiB, with [¹⁸F]AV-45 PET able to discriminate healthy controls from mild AD. However, the SUVR contrast between the lowest and highest uptake regions was higher for [¹¹C]PiB PET scans. It was theorized this may be due to higher non-specific white matter uptake of [¹⁸F]AV-45 [70]. A more recent study, by Chen et al., 2023, into anti-amyloid-β monoclonal antibodies effects were found statistically significant across both [¹¹C]PiB and [¹⁸F]AV-45 radiotracers. For one of the treatments, gantenerumab, [¹¹C]PiB SUVRs decreased more rapidly than that of [¹⁸F]AV-45. Type 1 error is reduced in clinical trials that only used one Aβ-tracer to evaluate Aβ treatment effects, and overall power was lower in trails using [¹⁸F]AV-45 compared to [¹¹C]PiB [71]. [¹⁸F]AV-45 has been described as lipophilic [72], thus in a similar manner to [¹⁸F]FMM, has been theorized to have non-specific uptake into white matter [73]. There is no observable difference between [¹¹C]PiB and [¹⁸F]AV-45 white matter uptake [67,74]. However, studying [¹⁸F]AV-45 uptake in white matter reveals a slight difference in AD patients and healthy individuals. The difference is too subtle for using the SUVR approach and MANOVA and discriminant analyses is required to observe it. Nevertheless, this information could reveal white matter modifications [65].

Limitations of [¹¹C]PiB

Some limitations have been identified in [¹¹C]PiB PET-based diagnosis of AD. Aβ plaques have diverse structures; many include massed, fibrillar polymers, but some lack the defining features of amyloids. In particular, [¹¹C]PiB affinity for cotton wool plaques, common in familial AD but rare in sporadic AD was problematic [75,76]. A mutation of the PSEN1 gene associated with familial AD produces a high cortical fibrillar Aβ load and lower than expected cortical [¹¹C]PiB retention in the brain [75].

Further, Aβ deposits have been reported in approximately a third of asymptomatic older adults. In vivo Aβ imaging comparing functional magnetic resonance imaging (fMRI) activity with [¹¹C]PiB PET imaging of the cortical regions, collectively known as the default network, demonstrates that high levels of Aβ deposition can be correlated with the observed fMRI activity in asymptomatic, dementia-free older individuals. Whether they will develop clinical dementia or AD [77] remains unknown.

Some evidence indicates that Aβ plaque deposits are necessary but not sufficient to cause the cognitive dysfunction associated with AD [78]. Fibrillar Aβ detected by [¹¹C]PiB PET can increase over time in older adults without causing dementia; further research is required to identify factors that explain this resilience to Aβ deposition [55,56]. A longitudinal assessment of Aβ deposition increase and cognition decline was found to be weakly correlated, suggesting that downstream factors affect the progression of AD and MCI symptoms [56]. Although β-amyloid deposition precedes AD and MCI, β-amyloid levels may not strongly correlate with cognitive impairment of AD patients [79] and MCI appears to be brought about by other factors or a combination of factors [39]. A recent study suggests a correlation of vascular risk factors and AD processes leading to MCI [80].

Similar to [¹¹C]PiB, [¹⁸F]AV-45 has high affinity for Aβ-plaques; however, because of longer half-life of fluorine-18 (¹⁸F), uptake time can be extended, and a stable plateau accumulates within 50 minutes allowing robust imaging and easier application of more than one dose per batch [81].

Views of authoritative and legislative bodies on PET radiopharmaceuticals

The National Institute on Aging-Alzheimer’s Association (NIA-AA) criteria for AD neuropathological diagnosis was revised a decade ago to include all Aβ-immunoreactive plaques, assessed according by phases of its deposition [82,83]. In 2014, The U.S. Food and Drug Administration (FDA) adjudged that licensed amyloid ligand tests do not establish a positive diagnosis of AD nor predict progression from MCI to AD [54]. The current view of the field seems to hold that although AD pathogenesis remains unclear, there are several mechanisms to be considered including Aβ protein deposition, tau protein hyperphosphorylation, inflammation, oxidative stress, mitochondrial, and cerebrovascular changes [84]. [¹¹C]PiB PET data must be combined with other neuropsychological testing to better define AD progression [85].

Although [¹¹C]PiB PET is now the gold standard for imaging Aβ plaques in the brain [17,21], [¹⁸F]AV-45 or [¹⁸F]Florbetapir, registered under the trade name of Amyvid, was the first Aβ-plaque PET imaging agent approved by the US FDA in 2012 [2,78]. Two other FDA-approved radiopharmaceuticals, Neuraceq ([¹⁸F]Florbetaben) and Vizamyl ([¹⁸F]Flutemetamol) were also used to estimate Aβ plaque density in the brain (Figure 8) (https://www.accessdata.fda.gov/scripts/cder/daf/).

Conclusion

Aβ-plaque imaging is likely to play a critical role in the development of treatments for AD by monitoring the response to anti-amyloidal therapies or other treatments and improving early treatment possibilities.

Comparing certain aspects of radiochemistry methods can be challenging. For example, yields can be reported as a percentage of a [¹¹C]synthon or the [¹¹C]CO₂ produced as corrected or uncorrected yields. A few publications do not report a % yield at all but just the total activity or specific activity obtained at the end of synthesis. Standardizing how radiochemical yield is reported would facilitate direct comparison of synthesis methods but for now, the task is daunting. MA was reported consistently for the most part in GBq/µmol. However, it varies so much even between very similar synthesis methods, we have to wonder whether the radiosynthesis method affects MA at all.

Although [¹¹C]PiB cannot be commercially shipped as the structurally similar [¹⁸F]FMM or other Aβ-plaque tracers such as [¹⁸F]AV-45 or [¹⁸F]Florbetaben, there are still advantages of using [¹¹C]PiB in facilities that have ¹¹C radiochemistry capabilities. These advantages include lower white-matter uptake in Aβ-negative subjects, a higher dynamic rage and higher SUVR for AD positive subjects [63,64,69,70].

In terms of future developments for [¹¹C]PiB synthesis, no one synthetic method combining all three optimizations reported in the literature has been reported (Figure 1). Hypothetically, the ideal method would use [¹¹C]CO₂ fixation [19] and have no deprotection [28] or prep-HPLC steps [35], thus minimizing potential failure points and maximizing speed. In addition, no cassette-based [¹¹C]CO₂ fixation for [¹¹C]PiB synthesis has been reported and might be worth exploring to further simplify cassette-kits.

Disclosure of conflict of interest

None.

Abbreviations

Aβ

amyloid-β

AD

Alzheimer’s disease

AQI

Quantification index

BET

Bacterial Endotoxin Test

DMF

Dimethylformamide

DMSO

Dimethylsulfoxide

FDA

U.S. Food and Drug Administration

FMM

Flutemetamol

fMRI

functional Magnetic Resonance Imaging

GC

Gas Chromatography

HPLC

High-performance liquid chromatography

MCI

Mild cognitive impairment

MeI

Methyl Iodide - CH₃I

MEK

Methyl ethyl ketone/Butan-2-one

MeOTf

Methyl trifoliate/trifluoromethanesulfonate - CH₃OSO₂CF₃

PET

Positron Emission Tomography

PiB

Pittsburgh compound B/[¹¹C]6-OH-BTA-1/2-(4-N-methylaminophenyl)-6-hydroxybenzothiazole

QC

Quality Control

RT

Room temperature

SPE

Solid phase extraction

SUVR

Standardized uptake value ratio

UV

Ultra Violate