Characterization of fast-decaying PET radiotracers solely through LC-MS/MS of constituent radioactive and carrier isotopologues

Background The characterization of fast-decaying radiotracers that are labeled with carbon-11 (t1/2 = 20.38 min), including critical measurement of specific radioactivity (activity per mole at a specific time) before release for use in positron-emission tomography (PET), has relied heavily on chromatographic plus radiometric measurements, each of which may be vulnerable to significant errors. Thus, we aimed to develop a mass-specific detection method using sensitive liquid chromatography-mass spectrometry/mass spectrometry (LC-MS/MS) for identifying 11C-labeled tracers and for verifying their specific radioactivities. Methods The LC-MS/MS was tuned and set up with methods to generate and measure the product ions specific for carbon-11 species and M + 1 carrier (predominantly the carbon-13 isotopologue) in four 11C-labeled tracers. These radiotracers were synthesized and then analyzed before extensive carbon-11 decay. The peak areas of carbon-11 species and M + 1 carrier from the LC-MS/MS measurement and the calculated abundances of carbon-12 carrier and M + 1 radioactive species gave the mole fraction of carbon-11 species in each sample. This value upon multiplication with the theoretical specific radioactivity of carbon-11 gave the specific radioactivity of the radiotracer. Results LC-MS/MS of each 11C-labeled tracer generated the product ion peaks for carbon-11 species and M + 1 carrier at the expected LC retention time. The intensity of the radioactive peak diminished as time elapsed and was undetectable after six half-lives of carbon-11. Measurements of radiotracer-specific radioactivity determined solely by LC-MS/MS at timed intervals gave a half-life for carbon-11 (20.43 min) in excellent agreement with the value obtained radiometrically. Additionally, the LC-MS/MS measurement gave specific radioactivity values (83 to 505 GBq/μmol) in good agreement with those from conventional radiometric methods. Conclusions 11C-Labeled tracers were characterized at a fundamental level involving isolation and mass detection of extremely low-abundance carbon-11 species along with the M + 1 carrier counterpart. This LC-MS/MS method for characterizing fast-decaying radiotracers is valuable in both the development and production of PET radiopharmaceuticals.


Background
Positron-emission tomography (PET), as an expanding biomedical research and diagnostic imaging technique, relies on the use of radiotracers labeled with a positron emitter, often either carbon-11 (t 1/2 = 20.38 min) or fluorine-18 (t 1/2 = 109.7 min) [1][2][3]. The short physical half-lives of these radiotracers demand rapid techniques for establishing the quality of each production batch before release. A suitable technique for this purpose must conclusively identify the radiotracer and also provide specific radioactivity (SA), the ratio of the amount of radioactive compound (GBq) to the amount of compound (mol) in radioactive and non-radioactive forms (isotopologues) at a specific measurement time. Highperformance liquid chromatography (HPLC) is almost universally applied to assure radiotracer identity and to measure radiochemical purity and SA at the end of each radiotracer synthesis [4,5]. Typically, 11 C-labeled tracers are produced with only a low degree of 11 C enrichment (<0.1%) [1], representing a SA value on the order of 100 GBq/μmol. Nevertheless, an accurate determination of SA is critical in many applications especially where the radiotracer is intended to bind to low-density protein targets, such as neurotransmitter receptors, enzymes or transporters, or in specific cases where the radiotracer is pharmacologically very potent, such as some radiotracers for dopamine, nicotinic, or opiate receptors [6][7][8].
The identification of a PET radiotracer with HPLC relies on comparison of its retention time (t R ) measured with a radioactivity detector with that of the reference non-radioactive compound measured with another detector, often an absorbance detector. In conventional measurement of the SA of a radiotracer, the detector giving the mass response is calibrated for compound mass, and the radioactivity associated with a particular carrier mass peak is measured in a calibrated detector, invariably a dose calibrator (an ionization detector). Surrogate radioisotopes (e.g., 137 Cs, 57 Co) are often used to calibrate ionization detectors for measuring positron emitters. The accuracy of measurements is highly vulnerable to errors in instrument settings, and to changes in sample containers and geometries, as demonstrated in several studies with fluorine-18 [9][10][11][12][13][14]. Similar issues will pertain to shorter-lived carbon-11, although this has not been well studied [15]. In addition, the presence of unknown inadvertent impurities in the radiotracer may compromise the mass measurements and hence the accuracy of derived SA. Hitherto, no non-radiometric method for verifying conventional HPLC-radiometric measurements of 11 C-labeled tracer SA has been reported.
We recognized the potential of a highly sensitive mass spectrometric technique for developing a reliable method to characterize fast-decaying PET radiotracers. Such a method seemed practical using triple quadrupole liquid chromatography-mass spectrometry/mass spectrometry (LC-MS/MS) because its specificity, sensitivity, and detection range were envisaged to be adequate for measuring both the radioactive species and the carrier counterpart. Additionally, the MS/MS can generate, isolate, and measure ions specifically for the positron-emitting species in the presence of a thousandfold excess carrier. Here, we describe the successful use of sensitive LC-MS/MS alone to identify and measure t 1/2 and the SA of 11 C-labeled tracers. The SA values obtained by this new method are compared with those from HPLC-radiometric methods. The LC-MS/MS method was found especially useful for verifying the validity of HPLC-radiometric methods.

HPLC-radiometric measurement of SA
Freshly prepared and formulated radiotracer solution (100 μL), contained in a syringe, was measured for radioactivity in a calibrated ionization chamber (Atomlab 300, Biodex, Shirley, NY, USA) and then injected onto HPLC (Beckman, Fullerton, CA, USA). The radiotracer was eluted isocratically (acetonitrile(aq) 10 to 100 mM HCO 2 NH 4 or 0.1% CF 3 CO 2 H) on a reverse-phase column (Onyx, Prodigy, or Luna, Phenomenex, Torrance, CA, USA) equipped with radioactivity and absorbance detectors. The mass of the carrier in the injectate was determined from a linear calibration curve generated from injections of known amounts of authentic non-radioactive standards. The SA was calculated from the decay-corrected activity present in the same volume of injectate, as measured in the calibrated ionization chamber.

LC-MS/MS
Methods for radiotracer analyses were developed on an API 5000 LC-MS/MS system (AB Sciex, Foster City, CA, USA), consisting of Shimadzu LC (Columbia, MD, USA) interfaced via electrospray with triple quadrupole MS/MS operated in positive ionization mode. The technique principally involves quantitative measurement of two ions that differ by two mass-to-charge ratio (m/z) units, one for the 11   The relative abundance of the M + 1 peak was measured by injecting a solution of PBR28 onto the LC-MS/MS setup to perform m/z 349, 350 [M + H] + → 121, 122 ( 12 C, M + 1) transitions and thereby to provide m/z 122 to 121 peak area ratio (%). Other compounds were similarly analyzed using respective acquisition methods edited to monitor product ions of 12 C and M + 1 species. The measured abundance (%) of M + 1 species relative to 12 C was used for converting M + 1 into 12 C peak area for radiotracer carrier. The peak area ratio (%) of M + 2 to M + 1 was measured by injecting 40 and 400 pg of PBR28 and acquiring transitions m/z 350, 351 → 122, 123.

Calculations
The SA of the radiotracer (Bq/mol) was calculated as where (1) A * M,M+1 is the sum of the measured [ 11 C, M] species peak area and [ 11 C, M + 1] species area calculated from the relative abundances [23] of 13 C, 2 H, 15 N, and 17 O in the product ion; (2) A M+1,M is the sum of the measured [M + 1] carrier peak area and [ 12 C] carrier area calculated from the measured M + 1 abundance in the nonradioactive tracer; and (3) SA Theoretical is the theoretical SA of carbon-11 (3.413 × 10 20 Bq/mol), the product of ln2/t 1/2 and Avogadro's number. The SA of the radiotracer was decay-corrected to give the SA at the end of radiosynthesis (SA 0 ) as SAe λt , where t is the time between the end of synthesis and peak elution in LC-MS/MS and λ is 0.034 min -1 (ln2/20.38). A plot of lnSA (n = 12) versus clock time of peak elution in Prism 5.02 (GraphPad, La Jolla, CA, USA) gave the value for λ and thus t 1/2 (t 1/2 = ln2/λ).

Results
In order to exemplify the non-radiometric LC-MS/MS method that we developed for identifying and measuring the SA of 11 C-labeled tracers, we report results on four radiotracers applied in human PET studies, namely  Figure 1A). Subsequently, the MS/MS acquisition method was edited to monitor the m/z 348 → 120 transition for [ 11 C] species in [ 11 C]PBR28. As expected, a similar LC-MS/MS analysis of non-radioactive PBR28 showed no peak for an m/z 348 → 120 transition. Furthermore, analysis of a fully decayed preparation of [ 11 C]PBR28 showed no peak for this transition. Thus, carrier PBR28 did not generate a peak due to the m/z 120 ion or any other spurious peak when performing the m/z 348 → 120 transition. Moreover, a dummy transition, entered between the transitions for [ 11   The area ratio 7.9240 × 10 −4 for [ 11 C]PBR28 implies that 1 out of 1,262 molecules was labeled with carbon-11 at the time of MS/MS detection. This ratio is consistent with the low degree of radionuclide enrichment found in 11 C-labeled preparations. Also, in Table 1, the peak areas for [ 11   each injection at the clock time of the peak elution in the ion chromatogram. Plots of the natural logarithm of SA versus measurement time gave straight lines (r 2 > 0.99) as shown, for example, for a [ 11 C]PBR28 preparation ( Figure 3). In each case, the decay constant λ for carbon-11 was obtained from the negative slope of the regression equation (y = −λx + b). The t 1/2 (t 1/2 = ln2/λ) of carbon-11 was estimated from these decay constants to be 20.43 ± 0.24 min (mean ± SD; n = 8), which is in excellent agreement with the accepted value of 20.38 min ascertained by measuring β + emission [24]. The accuracy of these t 1/2 measurements with LC-MS/MS implies that the measured peak areas of decaying [ 11 C] species and the stable [M + 1] carrier truly represent their concentrations.
Decay-corrected SA was measured with the LC-MS/MS method on three preparations of each of the four radiotracers (Table 2). Each preparation was measured six times over a period of 30 min. These measurements were remarkably reproducible (<5% RSD). Before LC-MS/MS analysis, each radiotracer was also analyzed by the calibrated radiometric method (activity using an ionization chamber dose calibrator and mass using HPLC apparatus) regularly used in our laboratory for quality control. Each analysis required at least 10 min for completion and was performed only once because of the time constraint imposed by the short half-life. The SA measurements obtained with LC-MS/MS deviated by ±7.4% overall from the mean of both radiometric and LC-MS/MS methods ( Table 2). The possible sources of error in the radiometric method include both the HPLC apparatus and the dose calibrator used for measuring carrier and radioactivity, respectively. The HPLC method used a linear calibration curve (r 2 > 0.99) generated with authentic non-radioactive compounds, such as, for example, PBR28 (>99% pure; RTI International, Research Triangle Park, NC, USA) or (R)-rolipram (>98%; Tocris, Ellisville, MO, USA). The quantification of the carrier may not always be    accurate due to lack of an internal standard and possible presence of a coeluting impurity. Measurements of the radioactivity of 11 C-labeled preparations (n = 6) in one ionization chamber were compared with measurements in another to assess their reliability. The two ionization chambers showed 10% to 12% differences in the radioactivity measurements even though both detectors had been identically calibrated. The lack of accuracy and reproducibility in measurements with ionization chambers [10][11][12][13][14]  The measurement of both species gave the SA and t 1/2 of the radiotracer while also confirming radiotracer identity. The results from multiple analyses showed that the LC-MS/MS method is fast, accurate, and reproducible enough for characterizing lowenriched, fast-decaying radiotracers. This mass-specific detection method is suitable for validating the conventional method of characterization of PET radiotracers. We verified that the excess carrier does not impact the ionization and detection of coeluting [ 11 C] species in our LC-MS/MS system. The peak for [ 11 C] species declined at an expected rate and then disappeared from the ion chromatogram as carbon-11 in the product ion decayed. Secondly, the ratio of M + 2 to M + 1 isotopologues was not altered when the radiotracer's non-labeled standard was injected at a concentration comparable to that of the carrier in the radiotracer. Also, the concentrations of [ 11 C] species and [M + 1] carrier in radiotracers were within the dynamic range of detection of the triple quadrupole MS/MS system. Before being injected into LC-MS/MS, the radiotracer preparations were diluted adequately to avoid possible saturation of the ionization and detection systems. Furthermore, no cross-talk was observed between the [M + H] + → product ion transition for [ 11  Carbon-11 decays to a stable nuclide, boron-11. This energetic decay event would instantly transform any molecule in which the carbon-11 had been incorporated. The product of carbon-11 decay would have a t R and mass different from those of the [ 11 C] species or the [M + 1] carrier . Accordingly, the LC-MS/MS acquisition detected no product or chemical impurity from the decay of any of these PET radiotracers. The measurement showed exponential loss of mass of [ 11 C] species as the radionuclide decayed thereby gave the same expected t 1/2 value for each radiotracer.
The radiometric method of SA measurement involves two detection systems, and both are prone to errors. For example, we observed a significant difference between two dose calibrators in recording the radioactivity of the same radiotracer preparation. The sources of error in the HPLC measurement of mass may include the weighing of standard, preparation of the calibration curve solutions, injection, chromatography, and UV detection. In contrast, the LC-MS/MS technique measures the mole fraction of [ 11 C] species in a radiotracer similar to the isotope ratio in a compound. Thus, SA was obtained simply by entering into a spreadsheet: (1)  The rapid LC-MS/MS method described here enables definitive characterization of radiotracers used in human PET imaging. A mass-specific detection method such as this is not susceptible to errors commonly associated with the use of the radio-HPLC method, as discussed. It may not be feasible to use triple quadrupole MS/MS regularly; however, it provides a wholly independent non-radiometric means for confirming identity and verifying SA of a radiotracer when needed. A less expensive and more commonly used single quadrupole MS instrument may offer the sensitivity to measure [ 11 C] species and [M + 1] carrier by selected ion monitoring of respective [M + H] + following ESI. However, such a less specific MS method can work only if no [M − 1] + ion is generated during ionization. Other mass analyzers such as time-of-flight and linear ion traps may also be suitable for the characterization of PET radiotracers. Finally, the LC-MS/MS method should be applicable to all 11 C-labeled tracers, except in instances where no 11 Ccontaining product ion is obtained or where interfering ions are generated. If a labeled product ion is not available, selected ion monitoring of a precursor ion ([M + H] + ) with a high-resolution MS can substitute the MS/MS method for measuring [ 11 C] species and [M + 1] carrier . Also, a highresolution MS is likely to resolve any interfering ions of isobaric species from the product ion of [ 11 C] species in 11 C-labeled tracers. In principle, LC-MS/MS may be used for characterizing PET radiotracers labeled with other short-lived positron emitters. Although an isotope separator has been used to identify the position of 18 F in labeled 1,1,1,2-tetrafluoroethane [25], no parallel mass spectrometric method has yet been reported for measuring the SA of 18 F-labeled tracers. The LC-MS/MS method should also be suitable for this purpose, as shown with the analysis of an 18 F-labeled tracer in our preliminary report [26].

Conclusions
A simultaneous mass detection of [ 11 C] species and [M + 1] carrier in 11 C-labeled PET tracers was achieved using the LC-MS/MS technique. A single rapid analysis can verify the identity including the position of label in a 11 C-labeled tracer preparation as well as provide a fully independent measure of SA. The LC-MS/MS method is a valuable adjunct to other techniques used in the quality control of PET radiopharmaceuticals. Abbreviations