Comparison of fully-automated radiosyntheses of [11C]erlotinib for preclinical and clinical use starting from in target produced [11C]CO2 or [11C]CH4

Background [11C]erlotinib has been proposed as a PET tracer to visualize the mutational status of the epidermal growth factor receptor (EGFR) in cancer patients. For clinical use, a stable, reproducible and high-yielding radiosynthesis method is a prerequisite. In this work, two production schemes for [11C]erlotinib applied in a set of preclinical and clinical studies, starting from either [11C]CH4 or [11C]CO2, are presented and compared in terms of radiochemical yields, molar activities and overall synthesis time. In addition, a time-efficient RP-HPLC method for quality control is presented, which requires not more than 1 min. Results [11C]erlotinib was reliably produced applying both methods with decay-corrected radiochemical yields of 13.4 ± 6.2% and 16.1 ± 4.9% starting from in-target produced [11C]CO2 and [11C]CH4, respectively. Irradiation time for the production of [11C]CO2 was higher in order to afford final product amounts sufficient for patient application. Overall synthesis time was comparable, mostly attributable to adaptions in the semi-preparative HPLC protocol. Molar activities were 1.8-fold higher for the method starting from [11C]CH4 (157 ± 68 versus 88 ± 57 GBq/μmol at the end of synthesis). Conclusions This study compared two synthetic protocols for the production of [11C]erlotinib with in-target produced [11C]CO2 or [11C]CH4. Both methods reliably yielded sufficiently high product amounts for preclinical and clinical use.


Background
The epidermal growth factor receptor (EGFR) belongs to the family of receptor tyrosine kinases and is one of the most frequently overexpressed proteins in malignant tumors (Ciardiello and Tortora 2008). Therefore, it has become an attractive target for cancer treatment. In this regard, EGFR specific antibodies such as cetuximab (Erbitux®) as well as small molecule tyrosine kinase inhibitors (TKIs) such as gefitinib (Iressa®) and erlotinib (Tarceva®, OSI-774) have been developed. Erlotinib reached marketing authorization in 2004 and belongs to the first-generation reversible TKIs (Cohen et al. 2005). It is used to treat non-small cell lung cancer (NSCLC) and pancreatic cancer (Singh and Jadhav 2017). One crucial parameter for prediction of treatment response is the mutational status of the EGFR. In order to identify the subset of patients who will benefit from therapy with erlotinib, positron emission tomography (PET) with [ 11 C]erlotinib has been proposed (Memon et al. 2009;Bahce et al. 2013;Petrulli et al. 2013;Slobbe et al. 2015). Owing to the short radioactive half-life of carbon-11 (t 1/2 = 20.4 min), a short overall synthesis time is desirable as it determines absolute radiochemical yields and molar activities. Hence, every adaption in the radiosynthesis protocol reducing the synthesis time significantly improves the outcome. Additionally, depending on the intended use of the radiotracer (i.e. preclinical or clinical), different approaches for formulation of the final product may be required. For example, higher doses in an overall larger volume are applied in patients, whereas in small animal studies the volume of injection is very limited and therefore, a higher concentration of the PET tracer is required.
The present study focused on comparing the radiosynthesis of [ 11 C]erlotinib starting from either [ 11 C]CO 2 or [ 11 C]CH 4 in terms of time efficiency, overall yields and molar activity. The synthesis was conducted according to Bahce et al. (2013) and was effectively implemented for preclinical and clinical studies Bauer et al. 2017). Additionally, a time-efficient HPLC method for quality control (Nics et al. 2018) was applied to minimize post-production loss of [ 11 C]erlotinib caused by a time intensive quality control and subsequent decline of molar activity.

Preparation of the synthesis module
A scheme of the synthesis module is presented in Fig. 1. All parts prior to HPLCpurification (reactor, HPLC injector, tubing) were rinsed with water and acetone and then dried with a stream of helium. Parts after semi-preparative HPLC (i.e. solid-phase extraction (SPE) tubing, dilution flask, product collection vial, product outlet tubing) were cleaned using ethanol and water. Fully-automated radiosynthesis of [ 11 C]erlotinib [ 11 C]erlotinib was synthesized by reacting 6-O-desmethyl-erlotinib with [ 11 C]CH 3 I ( Fig. 2), which was produced either from [ 11 C]CO 2 (target filled with N 2 + 1% O 2 ) or directly from [ 11 C]CH 4 (target filled with N 2 + 10% H 2 ). In method 1, [ 11 C]CO 2 was trapped on a molecular sieve (4 Å) and consequently reduced to [ 11 C]CH 4 over a Ni catalyst with hydrogen at 400°C, while in method 2 the Ni catalyst furnace was bypassed since [ 11 C]CH 4 was directly produced in the cyclotron. The [ 11 C]CH 4 was trapped in a cooled PorapakQ column and was further processed to [ 11 C]CH 3 I by standard procedure via the gas phase method (Larsen et al. 1997). Subsequently, [ 11 C]CH 3 I was released and transferred into the reactor containing the precursor solution (0.8-1.0 mg, 1.9-2.4 μmol 6-O-desmethyl-erlotinib) dissolved in 150 μL CH 3 CN and 3 μL TBAH solution (54.0-56.0% (w/v) in H 2 O) at 25°C. The reactor was sealed and heated to 75°C for 5 min. After cooling down to room temperature, the reaction mixture was quenched by addition of 0.6 mL HPLC solvent and subsequently transferred into the built-in HPLC system. Semi-preparative HPLC was performed with CH 3 CN/NH 4 OAc buffer (0.2 M, pH 5.0) (65/35, v/v) at flow rates of 8 mL/min (method 1) or 2.5 mL/min (method 2). The HPLC eluate was monitored in series for radioactivity and ultraviolet (UV) absorption at a wavelength of 254 nm. On this system, radiolabelling precursor 6-O-desmethyl-erlotinib and product [ 11 C]erlotinib eluted with retention times of 2.5-3.5 min and 4.5-5.5 min (method 1, Fig. 3a) and 5.2-6.5 min and 9-10 min (method 2, Fig. 3b), respectively. The [ 11 C]erlotinib product fraction was collected in a bulb and diluted with 100 mL of sterile water. The resulting solution was then pushed through a preconditioned C18 SPE cartridge (10 mL EtOH, 20 mL H 2 O, dried). The cartridge was washed with 10 mL of water and the purified product was eluted with 1.5 mL of ethanol into the product vial.
For preclinical use, the EtOH was then removed on a rotary evaporator and the product formulated in in 0.9% aq. saline/0.1 M HCl (100/0.1, v/v) at an approximate concentration of 370 MBq/mL for intravenous injection.
For clinical use, the SPE cartridge was washed with 5 mL 0.9% saline solution, which was collected into the product vial containing 6 mL phosphate-buffer saline (PBS) solution. The resulting solution was transferred through a 22 μm sterile filter into a sterile 25 mL vial containing further 5 mL of 0.9% saline solution. Hence, the final total volume of the product solution was 17.5 mL (containing 8.6% ethanol).

Quality control of [ 11 C]erlotinib
Chemical and radiochemical impurities were detected and quantified using analytical radio-HPLC according to the monograph in the European Pharmacopoeia (Radiopharmaceutical  (Nics et al. 2018). The chemical identity of [ 11 C]erlotinib was determined by co-injection of the unlabelled reference compound erlotinib. Molar activity was calculated by determining the mass of erlotinib in the final product solution (UV-channel). For calculation of radiochemical purity (radioactivity channel), the percentage of [ 11 C]erlotinib relative to total radioactivity was determined (threshold ≥95%). Sample chromatograms are given in Fig. 4. Residual solvents were analysed by GC. Osmolality and pH were determined to assure safe administration using standard methods. Radionuclidic purity was assessed by recording of the corresponding gamma spectrum and additional measurement of the physical half-life. Testing of sterility and concentration of bacterial endotoxins was performed using standard protocols at the Department of Infection Diseases and Tropical Medicine (Medical University of Vienna, Austria).

Results
The fully-automated synthesis and purification of [ 11 C]erlotinib was performed successfully and with high reliability within approximately 35 min after the end of bombardment (EOB) on a GE FX C Pro module starting either from [ 11 C]CO 2 (method 1) or [ 11 C]CH 4 (method 2). All values are given by arithmetic means ± standard deviation. Method 1: In 94 runs, 2.6 ± 1.3 GBq of [ 11 C]erlotinib were obtained as a sterile solution ready for application (decay-corrected radiochemical yields are 8.5 ± 4.3% and 13.4 ± 6.2% based on [ 11 C]CO 2 and trapped [ 11 C]CH 3 I, respectively). Three syntheses failed due to technical problems, especially clogging of transfer lines. Radiochemical purity exceeded 99% as determined by radio-HPLC. No precursor was found in the final product (limit of quantification 30.92 ng/mL (Nics et al. 2018)). The molar activity Fig. 3 Semi-preparative chromatogram of the reaction solution of [ 11 C]erlotinib (top: UV-channel; bottom: radioactivity-channel) for method 1 (a) and method 2 (b). Note that flow rate was 8 mL/min for method 1 and 2.5 mL/min for method 2 at the end of synthesis (EOS) was 88 ± 57 GBq/μmol (range: 12-264 GBq/μmol, erlotinib content in formulated solution: 0.11-4.04 μg/mL or 0.28-10.28 nmol/mL). A typical analytical HPLC chromatogram is shown in Fig. 4.
Typical loss during final sterile filtration was <10% of the product activity. Residual acetonitrile as determined by GC was found to be <20 ppm. Osmolality was 303 ± 21 mosmol/kg and pH was 7.5 ± 0.1. Concentration of endotoxins was found below 1.0 EU/mL and all samples passed the test for sterility. All quality parameters were in accordance with the standards for parenteral human application. The entire quality control process (except for tests for residual solvents, endotoxins and sterility) was completed within 3 min using the optimized analytical HPLC system.
A comparison of the productions starting from either [ 11 C]CO 2 or [ 11 C]CH 4 is given in Table 1.

Discussion
For clinical use of [ 11 C]erlotinib in patients a stable and reproducible radiosynthesis is essential. So far, four different synthesis procedures of [ 11 C]erlotinib have been published employing different solvents (DMF or CH 3 CN), reaction temperatures (80°C or 120°C) and bases (NaH and TBAH) for the deprotonation of the alcohol functionality of the precursor ( Table 2). The highest, but also most fluctuating molar activities have been reported by Bahce et al. (2013) (Table 2). However, the [ 11 C]CH 3 I Fig. 4 Typical chromatogram of the purified and formulated [ 11 C]erlotinib using the optimized analytical HPLC (top: UV-channel; bottom: radioactivity-channel) production method is not fully specified in the work by Bahce et al. (2013), which is an important factor influencing molar activity. The two studies which used NaH as base for the synthesis of [ 11 C]erlotinib (Memon et al. 2009;Petrulli et al. 2013) did not state radiochemical yields. We employed the reaction conditions of Bahce et al. (2013) for setting-up the radiosynthesis of [ 11 C]erlotinib in our laboratories. The usage of CH 3 CN as solvent and TBAH as base led to final product amounts of 2.6 ± 1.3 GBq (method 1) or 0.76 ± 0.27 GBq (method 2), which was considered sufficient for clinical and preclinical use of the radiotracer, so that no further optimization of synthesis parameters was performed. Method 1 afforded higher final product amounts than method 2, due to the possibility to produce higher starting activities with the [ 11 C]CO 2 target. The employed starting activities for the [ 11 C]CH 4 target were 27-43 GBq [ 11 C]CH 4 for an irradiation time of 12-22 min. For a clinical use the irradiation time for [ 11 C]CH 4 production could be potentially prolonged, providing a maximum possible starting activity of approximately 70 GBq (EOB). In order to compensate the time loss due to the reduction step of [ 11 C]CO 2 in method 1, the flow rate in the semi-preparative HPLC purification was increased from 2.5 mL/min (method 2) to 8 mL/min (method 1). Precursor and product [ 11 C]erlotinib were still sufficiently separated with the increased flow rate and eluted with retention times of 2.5-3.5 min and 4.5-5.5 min, respectively (Fig. 3a). The shortening of the HPLC purification time in method 1 led to comparable total synthesis times for both methods. The [ 11 C]CH 4 method afforded higher molar activities of [ 11 C]erlotinib. This can mainly be attributed to the fact that atmospheric CO 2 is ubiquitous and can contaminate the radiosynthesis, which lowers molar activity in method 1. Andersson et al. (2009) reported 7-14-fold improvements in molar activities of four different PET tracers by using [ 11 C]CH 4 as compared with [ 11 C]CO 2 . Such high increases in molar activity were not obtained in  , the presently employed set-up (method 2) afforded molar activities at EOS up to 1 TBq/μmol. Additional optimization included the setup of an ultra-HPLC system, which reduced the time of the RP-HPLC run of the quality control from 10 to 1 min. (Nics et al. 2018) This reduction in time for the quality control would afford a 27% increase in molar activity at the time of PET tracer administration into a patient, as compared to use of the conventional HPLC system. To ensure that no radioactive impurity is missed, we compared the optimized HPLC method to the conventional assay: all peaks (including impurities), which were detected in the conventional system could also be detected in the same ratios in the improved set-up. Abourbeh et al. (2015) reported an inverse correlation between [ 11 C]erlotinib uptake in HCC827 tumor xenografts and injected carrier mass of unlabelled erlotinib, suggesting that saturation of EGFR-specific binding of [ 11 C]erlotinib occurred. Similarly, Bahce et al. (2013) reported a reduction in tumoral volume of distribution (V T ) of [ 11 C]erlotinib, when NSCLC patients underwent [ 11 C]erlotinib PET scans during erlotinib therapy as compared to PET scans when they were off therapy (Bahce et al. 2016). These data suggest that an increase in molar activity of [ 11 C]erlotinib by using the [ 11 C]CH 4 method may be beneficial for improved target-to-non target ratios in EGFR imaging, although a potential disadvantage of method 2 is the lower final product amount. However, a systematic comparison of the influence of molar activity on diagnostic performance of [ 11 C]erlotinib has not been performed yet.

Conclusion
We compared two different methods for the synthesis of [ 11 C]erlotinib, starting either from [ 11 C]CO 2 or from [ 11 C]CH 4 . Both methods reliably yielded sufficiently high product amounts for preclinical and clinical use. The [ 11 C]CH 4 method yielded 1.8-fold higher molar activities, which may be beneficial for improved target-to-non target ratios in EGFR imaging of tumors. In order to keep the time consumption to a minimum, a highly efficient RP-HPLC method was established lasting for not more than 1 min, which can be expected to lead to an increase in molar activity at the time of radiotracer injection into a patient. Availability of data and materials Please contact author for data request.