Strategies for the Production of [¹¹C]LY2795050 for Clinical Use

Abstract

This report describes a comparison of four different routes for the clinical-scale radiosynthesis of the κ-opioid receptor antagonist [¹¹C]LY2795050. Palladium-mediated radiocyanation and radiocarbonylation of an aryl iodide precursor as well as copper-mediated radiocyanation of an aryl iodide and an aryl boronate ester have been investigated. Full automation of all four methods is reported, each of which provides [¹¹C]LY2795050 in sufficient radiochemical yield, molar activity, and radiochemical purity for clinical use. The advantages and disadvantages of each radiosynthesis method are compared and contrasted.

Graphical Abstract

INTRODUCTION

Opioid receptors belong to the superfamily of G-protein–coupled receptors, and the opioid subfamily is further divided into 4 receptor subtypes: δ-opioid receptor (DOR), κ-opioid receptor (KOR), μ-opioid receptor (MOR), and opioid receptor-like1 (ORL-1).^1,2 These different subtypes have been isolated, cloned, and pharmacologically characterized. In humans, the KOR is the most abundant and is widely distributed in the forebrain, midbrain, and brain stem.³ KOR has been linked to multiple brain disorders including Tourette’s syndrome, Alzheimer’s disease, and epilepsy, and it also plays a role in substance abuse.⁴ As public health officials and healthcare workers in the United States grapple with the ongoing opioid crisis, kappa opioid agonists offer potential for development as non-addictive analgesics.⁵

For all these reasons, there is enormous interest in using molecular imaging to understand the pharmacology and pharmacokinetics of the KOR, including studying its role in various neuropsychiatric disorders as well as conducting receptor occupancy studies to establish the relationship between dose and response. One such approach involves radiolabeling KOR antagonists to use in conjunction with positron emission tomography (PET) imaging for quantification of KOR. The KOR agonist [¹¹C]GR103545 and the antagonist [¹¹C]LY2795050 were the first PET radiotracers approved for imaging KORs in humans,⁴ and additional radiotracers continue to be developed.^6-8 As described previously,^9,10 [¹¹C]LY2795050 displays favorable imaging properties, including both rapid penetration of the CNS and heterogeneous regional accumulation corresponding to the known distribution of KORs (e.g. amygdala, putamen, insula and frontal cortex), along with suitable kinetics for quantification.¹⁰ Clinical trials with [¹¹C]LY2795050 are underway at the University of Michigan (Figure 1), which necessitates reliable cGMP production of this tracer on clinical scale.

Figure 1:

Representative [¹¹C]LY2795050 bolus + infusion equilibrium ratio (EQR) images in a normal control (Peak EQR = 2.8; EQR = distribution volume ratio (DVR), under equilibrium conditions). Parametric EQR images are calculated from the average of frames from 40-80 min post-injection, normalized by the value in cerebellar gray mater.

In bringing the production of [¹¹C]LY2795050 online for clinical use at our facility, we sought to compare different synthetic routes to this radiotracer (Figure 2), taking into account factors such as complexity of the reaction set up, reproducibility, precursor stability, synthesis time, and product quality reflected in the enantiomeric purity, radiochemical yield, and molar activity (ratio of [¹¹C]LY2795050 to [¹²C]LY2795050). We report here a detailed comparison of four different syntheses of [¹¹C]LY2795050. The first utilizes the Cu-mediated radiocyanation of a boronate ester precursor followed by nitrile hydrolysis (Method A).¹¹ The other three employ an aryl iodide precursor for Pd-^4,7 (Method B) or Cu-mediated¹² (Method C) radiocyanation or for Pd-mediated radiocarbonylation (Method D).¹³ Automation of Method A has been described previously by our group, while Method B, originally described by colleagues at the Yale PET Center, has not been automated to our knowledge. Methods C and D have not previously been applied to the synthesis of [¹¹C]LY2795050 prior to this work. In the present paper, the automation of each method has been optimized, and we highlight the key advantages and disadvantages of each for producing [¹¹C]LY2795050 for clinical PET imaging. These studies also provide key insights for translation of these methods to other radiotracer targets in the future.

Figure 2:

Strategies for producing [¹¹C]LY2795050

RESULTS AND DISCUSSION

We first compare our previously reported Cu-mediated radiocyanation of boronate ester precursor (Method A), with both Pd- (Method B) or Cu-mediated (Method C) radiocyanation of an aryl iodide precursor (Figure 3). Following labeling, each of these methods requires hydrolysis of the [¹¹C]-cyano intermediate to yield [¹¹C]LY2795050. In addition, we compare radiocyanation methods A-C with a Pd-mediated radiocarbonylation strategy (Method D) (Figure 3).¹³ This latter approach eliminates the need for nitrile hydrolysis following the ¹¹C radiolabeling.

Figure 3.

Four approaches to the radiosynthesis of Kappa opioid antagonist [¹¹C]LY2795050. (A) Cu-mediated radiocyanation of Ar-Bpin 1. (B) Pd-mediated radiocyanation of Ar–I 2. (C) Cu-mediated radiocyanation of 2. (D) Pd-mediated radiocarbonylation of 2.

Method A: Cu(OTf)₂-mediated radiocyanation of aryl-Bpin precursor 1.

In a previous report,¹¹ we disclosed automation of the Cu(OTf)₂-mediated radiocyanation of Ar–Bpin precursor 1 for the clinical production of [¹¹C]LY2795050. This procedure has subsequently been used to deliver 9 doses for brain imaging of healthy human volunteers for human radiation dosimetry measurements and initial evaluation of imaging properties at our site (see Figure 1 for representative [¹¹C]LY2795050 human PET data). The original automated procedure using this method is shown in Scheme 1. [¹¹C]HCN (~ 1 Ci) was bubbled into a solution of 0.5 M pyridine (15 equiv) in DMA (0.3 mL) in the reaction vessel of a commercially available automated radiochemistry synthesis module (GE TRACER lab FX_M). To this solution was added 0.2 M Cu(OTf)₂ (4 equiv) in DMA (0.3 mL), followed by the ArBpin 1 (1 equiv) in DMA (0.3 mL). These solutions were pre-loaded into separate vials that were attached to the synthesis module prior to delivery of [¹¹C]HCN. The addition of solutions to the reactor was then automated by the synthesis module using push gas. Following the final reagent addition, the reactor was heated at 100 °C for 5 min. Radio-HPLC analysis of the resulting mixture showed the formation of intermediate [¹¹CN]-3 in a radiochemical conversion (RCC) of 40-50%. The reactor was cooled to 5 °C, and hydrolysis of the nitrile was accomplished by adding 30% H₂O₂ (0.2 mL) and 5 M NaOH (0.2 mL) and then heating at 80 °C for 5 min. The reaction was then quenched with 5 M acetic acid (0.4 mL), and the crude product was purified by semi-preparative HPLC to yield [¹¹C]LY2795050. This reaction is typically conducted on 10 μmol scale and affords a non-decay-corrected 26 ± 11 mCi of [¹¹C]LY2795050 at end-of-synthesis (EOS) [10 ± 4% non-decay corrected radiochemical yield (RCY)] based upon starting [¹¹C]HCN, molar activity (A_m) = 914 ± 97 mCi/μmol and radiochemical purity = 94%. The overall synthesis time is 45 minutes from the end-of-bombardment (EOB).

Scheme 1.

Cu(II) mediated cyanation of ArBPin 1 to form [¹¹C]-3 followed by hydrolysis to [¹¹C]LY2795050.

Despite the success of this synthesis, we experienced some challenges during clinical production operations. First, this method requires high-quality aryl boronate ester precursor (1), as the presence of either hydrolyzed boronic acid or residual Pd (which can carry over from the synthesis of 1⁹) has a detrimental impact on radiocyanation yields, likely because they impact the interaction of Cu with BPin 1. Thus, it was critical to conduct multiple chromatographic purifications of 1 on Florisil^™ to ensure complete removal of residual Pd, while also limiting hydrolysis to the boronic acid. In addition, 1 has a relatively short shelf-life (≤3 months) as it undergoes hydrolysis to the boronic acid in the presence of ambient air/moisture. As discussed above, the presence of boronic acid results in a decline in radiochemical yield.

Second, pre-mixing of Cu(OTf)₂ and pyridine adversely impacts the yield of radiocyanation.^11,14 As such, it is necessary to add the reagents to the synthesis module in four separate vials. This is challenging to fit into the constraints of standard carbon-11 synthesis modules, which have 7 reagent vials and a small (3 mL) reaction vessel. To overcome this issue, we used smaller solution volumes (300 μL) and integrated additional external vials containing: (i) pyridine/[¹¹C]HCN solution, (ii) Cu(OTf)₂ solution, (iii) boronate ester precursor solution, and (iv) H₂O₂/NaOH solution. External vials mostly met our needs, but the addition of 200 μL of 5 M NaOH posed another challenge. The high viscosity and small volume of the solution caused significant back-pressure. This resulted in inconsistent delivery of H₂O₂/NaOH, blocked/broken lines, and frequent failure of the hydrolysis reaction.

Considering these challenges, we sought to compare our standard production method with alternatives. We noted that the Yale PET Center’s original synthesis of [¹¹C]LY2795050 involved a Pd-mediated radiocyanation of aryl iodide precursor 2.⁷ Furthermore, we have recently developed a Cu-mediated radiocyanation of aryl iodides.¹² As such, these two methods were selected for the initial comparison.

Method B: Pd⁰/dppf-mediated radiocyanation of aryl-I precursor 2.

We first explored the synthesis of [¹¹C]LY2795050 from aryl iodide 2 via Pd-mediated radiocyanation (Scheme 2). A semi-automated version of this transformation was originally reported by the Yale University PET Center.⁷ Their approach used Pd₂(dba)₃ (dba = dibenzylidene acetone) and diphenylphosphinoferrocene (dppf) as mediators. The radiochemistry was conducted manually, and the product was purified by semi-preparative HPLC.¹⁰ We aimed to translate this method to a synthesis of [¹¹C]LY2795050 in which all steps (synthesis, purification, reformulation, and filtration into a sterile dose vial) were fully automated using a GE TRACER lab FX_M module. Automated syntheses are preferred in the radiochemistry community for ease of synthesis, limiting radiation exposure, facilitating cGMP compliance, and enhancing reproducibility. We began by transferring radiolabeling to the synthesis module using the same scale and stoichiometries as Yale’s manual report.¹⁵ Because there is no detrimental effect to combining the reagents in advance for this reaction, KHCO₃ (25 μmol), Pd₂(dba)₃ (1.6 μmol), dppf (2.7 μmol), 2 (4.0 μmol), and DMF (0.4 mL) were combined and loaded into the TRACERLab reactor in advance of delivering ¹¹C from the cyclotron. [¹¹C]HCN (~2 Ci¹⁶) was bubbled into the solution for 5 min, and then the reactor was sealed and heated at 80 °C for 5 min. Radio-HPLC analysis of the resulting mixture showed formation of [¹¹CN]-3 in 70 ± 7% RCC (n = 8), comparable to that reported by the Yale group.¹⁵

Scheme 2.

Pd(0)-mediated cyanation of ArI 2 to form [¹¹C]-3 followed by hydrolysis to [¹¹C]LY2795050.

We next optimized the nitrile hydrolysis step. Manual screening experiments revealed that 2 M NaOH (0.4 mL)/30% H₂O₂ (0.25 mL) could be used in place of 5 M NaOH (0.1 mL)/30% H₂O₂ (0.25 mL). This modification resolved challenges associated with the small volume and solution viscosity that hampered Method A. Furthermore, because all the synthesis reagents could be loaded directly into the reactor (vide supra), no extra vials were required to accommodate this NaOH/H₂O₂ solution.

With optimal conditions in hand, we next incorporated HPLC purification of the product (Supporting Information, Table S1). We prepared [¹¹CN]-3 as noted above and conducted the hydrolysis reaction with 2 M NaOH (0.4 mL) and 30% H₂O₂ (0.25 mL). The crude reaction mixture was purified by semi-preparative HPLC (column: Luna, C8(2), 10 μ, 100Å, 150 x 10 mm; mobile phase: 25% MeCN, 100 mM NH₄OAc in 1% acetic acid, 5 mL/min). The [¹¹C]LY2795050 product was collected between 7-9 minutes, diluted into water (60 mL) and reformulated using a C18 Sep-Pak. [¹¹C]LY2795050 was eluted from the Sep-Pak with ethanol (0.5 mL), diluted with USP saline (9.5 mL), and the formulated product was filtered through a 0.22 μm filter to deliver a sterile dose for clinical use. This initial procedure afforded [¹¹C]LY2795050 in 3.9% radiochemical yield (RCY), n=1.

During hot cell cleaning we noticed significant precipitate in the reactor. In a first attempt to address this issue, following hydrolysis we diluted the reaction mixture with 1 mL of HPLC buffer (25% MeCN, 100 mM NH₄OAc in 1% acetic acid). Purification and reformulation using the above conditions resulted in a slightly improved RCY of 4.8% (n = 1), but also led to detectable precipitation. In a second attempt to minimize precipitation, we lowered the amount of KHCO₃ for trapping the [¹¹C]HCN from 25 mg to 2.5 mg. Following radiocyanation, the nitrile was hydrolyzed using 30% H₂O₂ (0.25 mL) and 2M NaOH (0.4 mL), and the mixture was diluted with acetonitrile (0.6 mL). Following purification/reformulation, this sequence provided 108 ± 49 mCi at EOS (21 ± 16% non-decay corrected RCY based upon starting [¹¹C]HCN, synthesis time = 41 min, n = 6), A_m was 2075 ± 1063 mCi/μmol and >98% ee was determined by chiral HPLC.¹⁷

Overall, Method B afforded a higher overall RCY (21%) than method A (10%), while employing a precursor with dramatically higher shelf stability. Given the high radiochemical yield using this method, we completed three validation runs to confirm suitability for human use (see Supporting Information, Table S2). The major limitation is the palladium mediator, as palladium has a low permitted daily exposure (LPDE) for parenteral administration, which means that purification strategies must be highly robust.¹⁸ Using this method all doses of [¹¹C]LY2795050 had residual Pd below the limit of detection (LOD) (< 8.5 μg/10 mL batch). Nevertheless, given this potential issue, we sought to evaluate whether 2 was a suitable substrate for our newly developed Cu-mediated radiocyanation of aryl iodides,¹⁴ since copper has a LPDE of ≤340 μg/day (more than 30-fold higher than that of Pd).

Method C. Cu^I/dmeda-mediated radiocyanation of aryl-I precursor 2.

The Cu-mediated radiocyanation of aryl iodides uses a combination of CuI and 1,2-dimethylethylene diamine (DMEDA) (Scheme 3), which are both air sensitive.¹⁰ Thus, a vial in the synthesis module was pre-flushed with N₂, and a solution of CuI (1.9 mg), aryl iodide 2 (5.0 mg), and DMEDA (2.2 μL) in DMF (0.98 mL) was prepared under a N₂ sparge and added to the vial prior to delivery of ¹¹C from the cyclotron. Furthermore, all transfers into the synthesis module vials were carried out under a positive flow of N₂ to minimize exposure to ambient air. H¹¹CN (~1 Ci) was then bubbled into an aqueous solution of K₃PO₄ (2.65 mg in 0.2 mL of H₂O) in the reactor. Notably, the H¹¹CN contains traces of H₂ and NH₃,¹⁹ and it is critical to remove these impurities to avoid the formation of copper hydrides. This was achieved by sparging the aqueous K¹¹CN solution with N₂ prior to catalyst addition. Following sparging, the solution of CuI, aryl iodide 2, and DMEDA was added, and the resulting mixture was heated at 130 °C for 5 minutes. These automated conditions afforded 40-60% RCC to the cyano intermediate [¹¹C]-3 as determined by analytical HPLC (n = 10).

Scheme 3.

Cu(I) mediated cyanation of ArI 2 to form [¹¹C]-3, followed by hydrolysis to [¹¹C]LY2795050.

We next evaluated the hydrolysis step for this reaction mixture (Supporting Information, Table S3). Using the same conditions as above resulted in poor conversion of [¹¹C]-3 and low (7%) yield [¹¹C]LY2795050. Inspection of the synthesis modules following this sequence revealed substantial amounts of precipitate, indicating poor solubility of reagents in the reaction mixture. We note that the solvent system for this radiocyanation is DMA/water, which is significantly different from that of the Cu- (DMA) or Pd-mediated (DMF) reactions in Schemes 1 and 2. We made efforts to enhance solubility, with the best results obtained when we moved to H₂O:DMA (1:1) as the solvent for initial trapping of H¹¹CN and also switched from DMF to DMA for the subsequent radiocyanation reaction. We also incorporated an additional 0.2 mL of H₂O₂ (total 0.4 mL) to address potential decomposition of the H₂O₂ by Cu(I).²⁰ These changes resulted in a significant increase in yield of [¹¹C]LY2795050 (to 12%), and a decay-corrected RCY of 2.6% (n = 1). Lastly, we omitted the acetic acid quench following hydrolysis, and instead added extra DMA (0.5 mL) to better solubilize any precipitates prior to HPLC purification. Under these optimized conditions, a fully automated synthesis resulted in 21% RCC to [¹¹C]LY2795050 (determined by sampling the reactor following synthesis). Upon purification and formulation, 7 ± 5 mCi of [¹¹C]LY2795050 was obtained, corresponding to a RCY of 4 ± 1 % (non-decay corrected based upon starting [¹¹C]HCN, 40 min, n = 2). A_m was 571 ± 162 mCi/μmol and the product was obtained in high ee (>97%) as determined by chiral HPLC (see Supporting Information).¹⁷

Method C has the advantage of using the aryl iodide precursor (2), which has higher shelf stability than BPin 1. The use of a copper mediator also offers significant advantages over palladium, including both lower cost and reduced toxicity. However, since both CuI and DMEDA are air sensitive, which complicates the radiosynthesis set-up. Moreover, the RCY for the optimized process (4%) is significantly lower than that for Methods A (10%) and B (21%). As such, given the alternative approaches available, it is not the preferred approach for production of [¹¹C]LY2795050.

Method D. Pd⁰/Xantphos-N-mediated radiocarbonylation of aryl-I precursor 2.

Preparation of [¹¹C]LY2795050 by radiocyanation (Methods A-C) requires access to [¹¹C]HCN. Moreover, the nitrile hydrolysis step has proven challenging and required extensive optimization (vide supra). For these reasons, and also given that aryl iodides can also be employed in ¹¹C-carbonylation reactions,^21-26 we explored a carbonylative approach that enables production of [¹¹C]LY2795050 from [¹¹C]CO and eliminates the nitrile hydrolysis step. The Scott group has recently reported an in-loop [¹¹C]CO carbonylation of aryl iodides mediated by Pd⁰/Ni-Xantphos.¹⁴ We sought to synthesize [¹¹C]-4 utilizing this method with formamide as the nucleophile.²⁷ Hydrolysis under mild conditions would then form [¹¹C]LY2795050 (Scheme 4).

Scheme 4.

Pd(0) mediated carbonylation of ArI 2 to form [¹¹C]-4 followed by hydrolysis to [¹¹C]LY2795050.

We first optimized manual/semi-automated conditions for this transformation (Supporting Information, Table S4).¹⁴ Pd(dba)₂ (3.98 μmol) and Ni-Xantphos (3.98 μmol) were dissolved in THF (0.2 mL), and 2 (12.2 μmol) was added. The mixture was maintained at room temperature for 20 min, and then, 5 min prior to the end of the cyclotron production of ¹¹C, the mixture was loaded into the HPLC loop of the TRACERLab synthesis module. [¹¹C]CO₂ was produced in the cyclotron and converted to [¹¹C]CO (~2 Ci), which was captured on a silica trap immersed in liquid N₂ (see Experimental Section and reference ¹⁴ for details). [¹¹C]CO was released from the silica trap into the HPLC loop pre-loaded with the reaction mixture. The HPLC loop was then sealed for 5 min at room temperature, after which the crude reaction mixture was transferred to the TRACERLab synthesis module reactor that was charged with DMAP (25 mg) and formamide (0.1 mL). The reactor was heated at 100 °C for 5 min, quenched with 1.0 mL of 1:1 (MeCN:water), and then heated at 100 °C for an additional 5 min. Analytical HPLC analysis showed 58% RCC to acid [¹¹C]-5 and 21% RCC to [¹¹C]LY2795050. We hypothesized that conducting the reaction at 100 °C was leading to over-hydrolysis to acid [¹¹C]-5, and that lowering the temperature might minimize this undesired side reaction. Indeed, quenching the reaction under the same conditions (1.0 mL of 1:1 MeCN:water) but at room temperature generated 43% of acid [¹¹C]-5 and 44% of [¹¹C]LY2795050.

Changing the amount of water, the temperature, and the reaction time did not improve the yield of amide. However, increasing the amount of formamide from 0.1 mL to 0.2 mL and heating at 100 °C for 10 min resulted in a significant improvement, leading to 42% of acid [¹¹C]5 and 49% of [¹¹C]LY2795050. With these optimized conditions in hand, full automation was conducted, including purification and reformulation. To achieve this, [¹¹C]-4 was synthesized in the HPLC loop of the TRACERLab synthesis module as described above. The crude reaction mixture was transferred to the TRACERLab synthesis module reactor containing a solution of DMAP (25 mg) in formamide (0.2 mL). The reactor was sealed and heated at 100 °C for 7.5 min, before cooling to 25 °C and quenching with water (0.5 mL). The reaction mixture was then purified using semi-preparative HPLC (column: Luna, C18(2), 5 μ, 100 Å, 250 x 10 mm; mobile phase: 20% MeCN in 0.1%TFA; flow rate = 4 mL/min). [¹¹C]LY2795050 was collected between 12-14 min, diluted into water (50 mL) and reformulated using a C18 Sep-Pak. [¹¹C]LY2795050 was eluted from the Sep-Pak with ethanol (0.5 mL), diluted with USP saline (9.5 mL), and the formulated product was filtered through a 0.22 μm filter to deliver a sterile dose for clinical use. This process afforded [¹¹C]LY2795050 in a RCC of 43% (determined by analysis of the crude reaction mixture after synthesis), and the overall RCY was 14 ± 0.3 mCi (3 ± 1% non-decay corrected RCY based upon starting [¹¹C]CO, 48 min, n = 2). The molar activity was 1870 ± 133 mCi/μmol,²⁸ and the product was obtained in high radiochemical purity (99%) and high ee (>97%, as determined by chiral HPLC) (see Experimental Section and Supporting Information for more details).¹⁷

SUMMARY AND CONCLUSIONS

In summary, four methods for producing clinical doses of [¹¹C]LY2795050 have been evaluated. Three involve transition metal-mediated radiocyanation of either ArBPin or ArI precursors, while the final one involves palladium-mediated ¹¹C-carbonylation of an ArI precursor. Cu-mediated radiocyanation of ArBPin 1 (Method A) has been used for routine production of [¹¹C]LY2795050 in our facility to support clinical studies for over two years. Using less toxic copper instead of a traditional palladium catalyst made it straightforward to validate the method and obtain approval for human use. However, it became apparent that the BPin precursor 1 has stability issues and requires experienced chemists for routine use. Moreover, because the precursor needs to be kept separate from the Cu mediator until after preparation of [¹¹C]HCN, the radiosynthesis setup requires reconfiguration of the synthesis module with additional vials.

These issues can be addressed through radiocyanation of a more stable aryl iodide precursor using either a palladium (Method B) or copper (Method C) mediator. In both cases precipitation is a major issue that impedes the nitrile hydrolysis step but can be addressed by changing the solvent and reaction conditions. Method B provided higher radiochemical yields and has also been validated for clinical production at our site. However, it does use toxic palladium, which requires robust control over purification and QC procedures. This can be circumvented using a Cu-mediator (Method C) but with the caveat that both CuI and DMEDA are air sensitive and challenging for less experienced chemists to handle for routine production purposes.

Methods A-C require on-site access to [¹¹C]HCN. However, ArI 2 is also amenable to ¹¹C-carbonylation (Method D). This approach uses our in-loop ¹¹C-CO insertion methodology, and also enables production of [¹¹C]LY2795050 by laboratories that have access to [¹¹C]CO.

Overall, this study demonstrates the power of modern metal-mediated radiolabeling methods for accessing clinically relevant radiopharmaceuticals like [¹¹C]LY2795050. All four methods are suitable for production of [¹¹C]LY2795050, offering PET Centers multiple approaches to generate this product for kappa PET imaging studies. The methods offer different advantages and disadvantages (Table 1), and selection of a specific method can depend upon the skill level of local chemistry staff, the availability of the appropriate precursor (ArBPin 1 versus ArI 2), and the accessibility of the respective ¹¹C synthon ([¹¹C]HCN versus [¹¹C]CO). We have translated both Method A and Method B for clinical production, and have had good success with both. Based upon our experiences with routine production, Method B (Pd-mediated radiocyanation of ArI 2) is a good starting point for most labs wishing to translate [¹¹C]LY2795050 for clinical use given precursor stability, radiochemical yields and ease of radiosynthesis.

Table 1:

Summary of production methods for [¹¹C]LY2795050^a

Method	A	B	C	D
Ease of precursor handling	Moderate (ArBPin 1)	Easy (ArI 2)	Easy (ArI 2)	Easy (ArI 2)
RCY	High	High	Moderate	Moderate
Am	High	High	High	Highest
RCP	High	High	High	High
Residual metals	Cu	Pd	Cu	Pd

^aColored text indicates acceptable and preferred features of Methods A – D

EXPERIMENTAL SECTION

Chemistry

See Supporting Information for full details of the synthesis of unlabeled reference standards and labeling precursors.

Radiochemistry – General Considerations

Unless otherwise stated, reagents and solvents were commercially available and used without further purification: sodium chloride, 0.9% USP, and sterile water for injection, USP, were purchased from Hospira; ethanol was purchased from American Regent; HPLC grade acetonitrile was purchased from Fisher Scientific. Other synthesis components were obtained as follows: sterile filters were obtained from Millipore; sterile product vials were purchased from Hollister-Stier; C18 Sep-Paks were purchased from Waters Corporation. C18 Sep-Pak was flushed with 10 mL of ethanol followed by 10 mL of Milli-Q water prior to use.

Production of ¹¹C

Production of [¹¹C]HCN (Methods A-C):

[¹¹C]CO₂ was produced with a GE PETtrace cyclotron. The ¹⁴N(p,α)¹¹C nuclear reaction was performed by proton bombardment of a pressurized gas target containing high-purity nitrogen and 0.5% oxygen to generate [¹¹C]CO₂, which was delivered to a GEMS Process Cabinet via stainless steel lines. [¹¹C]CO₂ was trapped on molecular sieves at room temperature, and then released at 350 °C with He carrier gas and mixed with H₂ gas over a preheated nickel oven at 400 °C for conversion to [¹¹C]CH₄. The [¹¹C]CH₄ was then passed over Ascarite and Sicapent columns to remove unreacted [¹¹C]CO₂ and water, respectively. The gas mixture was then combined with anhydrous ammonia gas and passed through a platinum oven at 950 °C to form [¹¹C]HCN. The [¹¹C]HCN (1 – 2 Ci) was transferred to a GE TRACERlab FX_M synthesis module and trapped by passing through a solution of base.

Production of [¹¹C]CO (Method D):

[¹¹C]CO₂ was produced with a GE PETtrace cyclotron. The ¹⁴N(p,α)¹¹C nuclear reaction was performed by proton bombardment of a pressurized gas target containing high-purity nitrogen and 0.5% oxygen to generate [¹¹C]CO₂, which was delivered to a GEMS Process Cabinet via stainless steel lines (see reference ¹⁴ for more details). [¹¹C]CO₂ was trapped on a molecular sieve column (Grace (p/n 5622) or Ohio Valley Specific Company (p/n 5326), 4Å, 60/80 mesh, 0.4–0.5 g) at room temperature. The accumulated [¹¹C]CO₂ was then released into an online reduction column by heating the molecular sieve trap to 360°C. [¹¹C]CO₂ was passed through a heated molybdenum column (Alfa Aesar (p/n 3089) or Goodfellow (p/n 802-109-88), 100 mesh powder, 10 g, ~6–7 cm in the middle of the 20 cm quartz column) at 850°C, in a stream of helium at 20 mL/min. The gas was purified through an ascarite column and delivered out of the GEMS Process Cabinet to a sealed hot cell. The produced [¹¹C]CO was trapped on a silica trap (100 mg silica gel in a 1/8 inch tube) immersed in liquid nitrogen and then, after delivery, the [¹¹C]CO was released and pushed by helium at 5 mL/min into the HPLC loop for reaction.

Production of [¹¹C]LY2795050

Method A:

~1Ci of [¹¹C]HCN was bubbled into a solution of 0.5 M pyridine (15 equiv) in DMA (0.3 mL) in the reactor vessel of a TRACER lab FX_M automated synthesis module. To this solution was added 0.2 M Cu(OTf)₂ (4 equiv.) in DMA (0.3 mL) followed by Ar–Bpin precursor 1 (1 equiv.) in DMA (0.3 mL). The reaction mixture was heated at 100 °C for 5 min. The reactor was cooled to 5 °C, and hydrolysis of the nitrile was accomplished by adding 30% H₂O₂ (0.2 mL) and 5 M NaOH (0.2 mL) and heating at 100 °C for 5 min. The reaction mixture was quenched (0.4 mL acetic acid), and the crude material was purified by semi-preparative HPLC as outlined below.

The final formulation was passed through a 0.22 μm sterile filter into a sterile dose vial. 26 ± 11 mCi of [¹¹C]LY2795050 was obtained, corresponding to 10 ± 4% radiochemical yield. Molar activity = 914 ± 97 mCi/μmol and RCP = 94%. The overall synthesis time was 45 minutes from the end of the cyclotron beam.

Method B:

~1Ci of [¹¹C]HCN was bubbled into the reactor vessel of a TRACER lab FX_M automated synthesis module containing iodo precursor 2 (2 mg) Pd₂(dba)₃ (1.5 mg), dppf (1.5 mg), and KHCO₃ (2.5 mg) dissolved in DMF (400 μL). The reaction mixture was heated at 80 °C for 5 min. The reactor was cooled to 5 °C, and hydrolysis of the nitrile was accomplished by adding 30% H₂O₂ (0.25 mL) and 2 M NaOH (0.4 mL) to the reactor and heating at 80 °C for 5 min. The reaction mixture was diluted with MeCN (0.6 mL), and the crude material was purified by semi-preparative HPLC as outlined below.

The final [¹¹C]LY2795050 product was obtained in 21 ± 16% radiochemical yield (108 ± 49 mCi at EOS, n = 6), 2075 ± 1063 mCi/μmol molar activity and 92% radiochemical purity. The enantiomeric purity was >98% ee as determined by chiral HPLC. The overall synthesis time was 41 minutes from the end of the cyclotron beam.

Method C:

[¹¹C]HCN was bubbled into the reactor containing K₃PO₄ (2.65 mg) in H₂O/DMA (1:1) (0.2 mL) to trap H¹¹CN. A solution of CuI (1.9 mg), aryl iodide 2 (5.0 mg) and DMEDA (2.2 μL) in DMA (0.98 mL) was added into the reactor. The reaction mixture was heated at 130 °C for 5 min. The reactor was cooled to 5 °C, and hydrolysis of the nitrile was accomplished by adding 30% H₂O₂ (0.4 mL) and 2 M NaOH (0.2 mL) and then heating at 80 °C for 5 min. The reaction mixture was then quenched with DMA (0.5 mL), and the crude material was purified by semi-preparative HPLC as outlined below.

The final product was obtained in 4 ± 1% radiochemical yield (7 ± 5 mCi at EOS, n = 2), 571 ± 162 mCi/μmol molar activity and 98% radiochemical purity. The enantiomeric purity was >97% ee as determined by chiral HPLC. The overall synthesis time was 40 minutes from the end of the cyclotron beam.

Method D:

Pd(dba)₂ (3.98 μmol) and NiXantphos (3.98 μmol) were dissolved in THF (0.2 mL), and aryl iodide precursor 2 (12.2 μmol) was added. The mixture was maintained at room temperature for 20 min, and then, 5 min prior to the end of the cyclotron production of ¹¹C, the mixture was loaded into the HPLC loop of the TRACERLab synthesis module. [¹¹C]CO₂ was produced in the cyclotron and converted to [¹¹C]CO which was captured on a silica trap immersed in liquid N₂ (see Experimental Section).¹⁴ [¹¹C]CO (~2 Ci) was released from the silica trap into the HPLC loop pre-loaded with the reaction mixture. The HPLC loop was then sealed for 5 min at room temperature after which the crude reaction mixture was transferred to the reactor that was pre-charged with DMAP (25 mg) and formamide (0.2 mL). The reactor was sealed and heated at 100 °C for 7.5 min, before cooling to 25 °C and quenching with water (0.5 mL). The crude material was purified by semi-preparative HPLC as outlined below.

This process afforded 14 ± 0.3 mCi of [¹¹C]LY2795050, in RCY of 3 ± 1% (non-decay corrected, 48 min, n = 2) and RCP = 99%. Molar activity = 1870 ± 133 mCi/μmol, and the product was obtained in high ee (>97%), as determined by chiral HPLC (see Experimental Section and Supporting Information for more details).

Semi-preparative HPLC Purification and reformulation of [¹¹C]LY2795050

Semi-preparative HPLC Purification and reformulation for Methods A-C:

The crude material was purified by semi-preparative HPLC (column: Luna 10u, C8(2), 10 μm, 100A, 150 x10 mm; mobile phase: 100 mM NH₄OAc in 25% MeCN with 1% AcOH; flow rate: 5 mL/min). The [¹¹C]LY2795050 product peak (t_R ∼6-8 min) was collected for maximum 2 min and diluted into a round-bottom flask containing milli-Q water (60 mL). This solution was passed through a C18 Sep-Pak to trap [¹¹C]LY2795050 on the cartridge. The C18 cartridge was washed with 5 mL of sterile water, and the product was eluted with ethanol (0.5 mL) and diluted with USP saline (9.5 mL).

Semi-preparative HPLC Purification and reformulation for Method D:

The reaction mixture was purified by semi-preparative HPLC (column: Luna, C18(2), 5 μ, 100 Å, 250 x 10 mm; mobile phase: 20% MeCN in 0.1%TFA, 4 mL/min). The [¹¹C]LY2795050 product peak (t_R ~12-14 min) was collected and diluted into a round-bottom flask containing milli-Q water (50 mL). This solution was passed through a C18 Sep-Pak to trap [¹¹C]LY2795050 on the cartridge. The C18 cartridge was washed with 5 mL of sterile water, and the product was eluted with ethanol (0.5 mL) and diluted with USP saline (9.5 mL).