A quantitative LC–MS/MS method for the determination of tissue brincidofovir and cidofovir diphosphate in a MuPyV- infected mouse model
1 | INTRODUCTION
Cidofovir (CDV) is a selective inhibitor of viral DNA synthesis that is currently US Food and Drug Administration (FDA) approved for the treatment of cytomegalovirus retinitis. However, CDV must be admin- istered as a 1 h i.v. infusion with probenecid and I.V. fluids to mitigate the risk of nephrotoxicity (Gilead Sciences Inc., 2010). Brincidofovir (BCV, CMX001) is a lipid conjugate of CDV. The lipid conjugation facilitates entry of BCV into target cells via endogenous lipid uptake pathways (Lanier et al., 2010; Painter & Hostetler, 2004), which is believed to improve its safety profile and may ultimately allow oral dosing (Grimley et al., 2016; US National Library of Medicine, 2019). Intracellularly, both CDV and BCV are converted to the pharmacologi- cally active moiety, cidofovir diphosphate (CDV-PP, CMX048; Painter & Hostetler, 2004). CDV-PP exerts an antiviral effect by act- ing as a potent alternate substrate inhibitor of viral DNA polymerase (Xiong, Smith, & Chen, 1997) and rapidly achieves viral clearance with adenovirus infection (Grimley et al., 2016) and improved survival in animal models of smallpox (Grossi et al., 2017). Quantification of BCV and CDV-PP in tissues is important to understand the pharmacokinet- ics of both drugs and their distribution into tissues where viral replica- tion occurs. In particular, BK virus (BKV) and JC virus (JCV) are ubiquitous polyomaviruses which can cause organ rejection and serious illness in patients with impaired immune systems (Stolt, Sasnauskas, Koskela, Lehtinen, & Dillner, 2003). Infection with BKV can result in BKV-associated nephropathy in up to 10% of kidney transplant recipients (Hirsch et al., 2002), while JCV reactivation in the brain has been associated with progressive multifocal leukoencephalopathy (PML), a fatal demyelinating disease seen in patients on potent immunosuppressive regimens used in the treat-
ment of multiple sclerosis and Crohn’s disease (Mills & Mao-Draayer, 2018; Van Assche et al., 2005). As no antivirals are approved, the cur- rent management includes reduction of immunosuppression, which can risk organ rejection or recurrence of autoimmune disease symp- toms. An active antiviral capable of reducing viral burden while maintaining immunosuppression could have a substantial impact on the survival of existing kidney grafts, as well as reducing recurrent symptoms of autoimmune diseases. BCV has potent in vitro antiviral activity against BKV and JCV with an EC50 of 0.13–0.27 and 0.006–0.1 μM, respectively (Chemaly, Hill, Voigt, & Peggs, 2019), without CDV-associated nephrotoxicity. To optimize the dose and dosing regimen of BCV in these patient populations, a murine polyoma viral infection model was used to determine the antiviral activity of BCV against murine polyoma virus (MuPyV). MuPyV is known to infect the kidney in mice and can also be implanted in the brain to create a demyelinating PML disease phenotype (Mockus et al., 2018). Measurement of drug concentrations in these tissue compartments is critical to optimizing the dosing and ultimate antiviral activity of BCV. To the best of our knowledge, this is the first published assay for the quantification of BCV and CDV-PP in mouse kidney, brain and spleen tissue homogenate. The study was conducted in mouse kidney tissue homogenate but the validated method evaluates brain and spleen tissue homogenates as well for future studies.
2 | EXPERIMENTAL
2.1 | MATERIALS
Brincidofovir (purity, 99.9%) and its stable isotopically labeled internal standard brincidofovir-d6 (BCV-d6; purity, 96.9%) were obtained from Chimerix Inc. (Durham, NC, USA). CDV-PP (as sodium salt, purity 91.8%) was also provided by Chimerix Inc. and the stable isotopically labeled internal standard 13C 15N -CDV-PP (CDV-PP-IS) was synthe- sized by Chemcyte (San Diego, CA, USA; see Figure 1). Dimethyl sulf- oxide (certified ACS), methanol and acetonitrile (both HPLC grade) were purchased from Fisher Scientific (Fair Lawn, NJ, USA). Formic acid (certified ACS, 88%), ammonium hydroxide (certified ACS) and dibasic and monobasic ammonium phosphate (reagent grade) were also purchased from Fisher Scientific. The water used was purified by Hydro Picosystem® UV Plus (Durham, NC, USA). Blank mouse tissue (kidney, brain and spleen) was purchased from BioIVT (Westbury, NY, USA). Precellys 2 ml hard tissue metal beads kit tubes (Caymen Chem- ical, Ann Arbor, MI, USA) were used to homogenize the tissues on a Precellys 24 homogenizer (Bertin Instruments, Rockville, MD, USA).
2.1.1 | LC–MS/MS instrumentation
A Shimadzu Prominence HPLC system including pumps (LC-20 AD), degasser (DGU-20A), and controller (CBM-20A) was supplied from Shimadzu (Columbia, MD, USA). For the BCV analysis in tissue homogenates, a Waters Xterra MS C18 (50 × 2.1 mm, 3.5um particle size) analytical column was used under reverse-phase conditions with 0.1% formic acid in water (mobile phase A) and 0.1% formic acid in acetonitrile (mobile phase B) at a flow rate of 0.600 ml/min. The column heater (CTO-20A) was set at 35◦C, and the autosampler (SIL-20 AC/HT) was maintained at 10◦C. The injection volume was 3 μl. The LC gradient was held at 45% B for 0.25 min and increased to 90% B at 1.50 min and held for 30 s. At 2.10 min the gradient returned to 45% B until a final run time of 3.50 min. The retention time for BCV and BCV-d6 was 1.65 min.
FIG U R E 1 Chemical Structure of Brincidofovir (A), Brincidofovir-d6 (B), Cidofovir diphosphate (C), and 13C315N2-Cidofovir diphosphate (D)
An API-5000 triple quadruple mass spectrometer (SCIEX, Foster City, CA, USA) operated in positive ion TurboIonspray mode was used to acquire data for this method. The source temperature was 500◦C and the ion spray voltage was 5,500 V. The declustering potential was 256 V, and the collision energy was 39 V for both the analyte and internal standard. Multiple reaction monitoring (MRM) was used to detect the analyte [precursor/product] transitions (m/z) as follows: BCV [563.5/262.0], BCV-d6 [568.5/262.0]. Multiple MRM transitions were evaluated for BCV. The transition ultimately selected for BCV represents the natural 13C isotope. This transition produced adequate sensitivity with minimal baseline.
CDV-PP analysis was performed using the same LC–MS/MS setup as described above. A Thermo BioBasic AX (50 × 2.1 mm, 5-μm particle size) column was used under anion exchange conditions with 750 mM ammonium acetate (mobile phase A) and 75:25 5 mM ammo- nium acetate–acetonitrile with 2% dimethyl sulfoxide (pH 9.95) with a flow rate of 0.250 ml/min. The autosampler and the column heater temperatures were 10 and 35◦C, respectively. Injection volume was 10 μl. The LC gradient was maintained at 25% B for 0.25 min and increased to 100% B at 1.00 min and held until 3.75 min. At 3.80 min, the gradient was set to 0% B and held until 7.50 min. The gradient returned to 25% B at 7.75 min and was held until 8.00 min, which is the total run time. The retention time for CDV-PP and CDV-PP-IS was 3.30 min. The high salt concentration in mobile phase A was required to reduce carryover in the assay. A divert valve was used to divert flow away from the mass spectrometer while the high salt concentrations were flowing through the system.
An API-5000 triple quadrupole mass spectrometer in positive ion TurboIonspray mode was used for this analysis with the following MRM transitions: CDV-PP [440.0/262.0] and CDV-PP-IS [445.0/267.0]. The source temperature was 650◦C and the ion spray voltage was 5,500 V. For the analyte and internal standard, the declustering potential was 120 V with a collision energy of 19 V.Linear regression of concentration (x) vs. peak area ratio of compound to internal standard (y) with a 1/(x2) weighting was used with Sciex Analyst software (version 1.6.2) for both analytes.
2.1.2 | Validation
These analytical methods were validated to meet the acceptance criteria of the guidelines of the FDA (2015). The validation procedure consisted of three analytical runs to evaluate assay linearity, precision and accuracy, with additional runs to determine specificity, dilutions, stability, recovery and matrix effects. The tissue method was validated in mouse kidney homogenate, but additional experiments were per- formed to cross-validate in mouse brain and spleen homogenate. These additional tests consisted of stability, precision and accuracy, selectivity, matrix spikes, matrix effects and recovery.
2.1.3 | Blank tissue homogenate preparation
Blank tissue homogenate was prepared by cutting untreated mouse kidney, brain and spleen tissue (BioIVT, Durham, NC, USA) into small pieces (~10–20 mg) and placing into separate Precellys tubes (hard tissue–metal beads kit). A 1 ml aliquot of cold 70:30 acetonitrile– 1 mM ammonium phosphate (pH 7.4) was added to each tube. The tubes were homogenized on a Precellys 24 tissue homogenizer using the following setting (5,500 rpm × 60 s; three cycles; 15 s pauses between cycles). The individual tubes from each tissue type were pooled and stored for future use as blanks and as matrix for preparing daily calibration standards and QC samples. Blank homogenate was stored at −20◦C.
2.1.4 | Preparation of calibration standards and quality control samples
BCV was weighed in duplicate and prepared to 1 mg/ml in 90:10:0.1% water–methanol–ammonium hydroxide. Duplicate CDV- PP stock solutions were prepared in water with a concentration of 1 mg/ml. All stock solutions were stored at −80◦C.The BCV stocks were diluted in 70:30 methanol–water to make a set of calibration standard working solutions at 10.0, 20.0, 50.0, 200, 500, 2,000, 8,500 and 10,000 ng/ml, and a quality control (QC) working solution set with concentrations of 10.0, 30.0, 300, 8,000 and 50,000 (dilution spike) ng/ml. The working solutions were stored at −80◦C.
Calibration standards and QCs were prepared by spiking the appropriate working solution into mouse blank tissue homogenate. Mouse kidney tissue homogenate was used for calibration standards throughout the method validation and QC samples were prepared in mouse kidney, brain and spleen homogenates. The resulting concen- trations for the calibration standards were 1.00, 2.00, 5.00, 20.0, 50.0, 200, 850 and 1,000 ng/ml with QCs at 1.00, 3.00, 30.0, 800 and 5,000 (dilution QC) ng/ml.
The internal standard (BCV-d6) was prepared to 1 mg/ml in 90:10:0.1% water–methanol–ammonium hydroxide and was stored at −80◦C. This solution was then diluted to 25 ng/ml (in 70:30 methanol–water) and stored at −20◦C.For the CDV-PP tissue homogenate assay, the 1 mg/ml CDV-PP stocks were diluted in 1 mM ammonium phosphate (pH 7.4) to make a set of calibration standard working solutions at 0.500, 1.00, 2.50, 10.0, 25.0, 100, 420 and 500 ng/ml, and a set of QC working solutions were prepared at 0.500, 1.50, 20.0, 400 and 2,000 (dilution QC) ng/ml, which were stored at −80◦C.
Calibration standards and QCs were prepared in a similar fashion to BCV. Calibration standards were prepared at 0.0500, 0.100, 0.250, 1.00, 2.50, 10.0, 42.0 and 50.0 ng/ml in mouse kidney homogenate, and QC samples prepared at 0.0500, 0.150, 2.00, 40.0 and 200 (dilution QC) ng/ml in mouse kidney, brain and spleen homogenates.A 1 mg/ml CDV-PP-IS stock solution was prepared with purified water. The stock solution was diluted to 5.00 ng/ml with 50:50 water–methanol as the working solution and stored at −20◦C.
2.1.5 | Sample extraction
Mouse tissue samples (kidney, brain, or spleen) were weighed (<1 mg) in Precellys tubes (hard tissue–metal beads kit) quickly, on dry ice, to keep the tissues frozen. Following weighing, 1 ml of cold 70:30 acetonitrile–1 mM ammonium phosphate (pH 7.4) was added to each tube and samples were homogenized. For BCV analysis, the tissue homogenate (100 μl) was mixed with 50 μl of 70:30 methanol–water containing stable, isotopically labeled BCV-d6, the internal standard. The samples were vortexed, centrifuged and transferred into a 96-well plate for LC–MS/MS analysis. For CDV-PP analysis, 200 μl of tissue homogenate was mixed with 50 μl of 50:50 methanol–water containing the internal standard (CDV-PP-IS). Samples were vortexed and centrifuged. The superna- tant was transferred to a clean tube and evaporated to dryness under a stream of nitrogen at 50◦C for 30 min using a Biotage TurboVap LV (Charlotte, NC, USA). Samples were reconstituted with 75 μl of 1 mM ammonium phosphate (pH 7.4) prior to LC–MS/MS analysis. 2.1.6 | Pharmacokinetic study design The design of this trial has been previously presented (Naderer, 2018 Kidney Week conference San Diego, CA). In brief, 27 C57BL/6 mice (8–12 weeks old) were assigned to either three placebo groups (three mice per group) or one of three treatment groups (six mice per group). Mice were inoculated in hind footpads with 1.0 × 106 PFU of MuPyV, a mouse-specific polyomavirus, on day 0, 24 h prior to the first dose of BCV. In the treatment groups, BCV was administered via intraperi- toneal injection at 80 (day 1), 40 (days 1 and 4) or 13 (days 1 through 6) mg/kg over a 1 week period. Phosphate-buffered saline was used as a control. BCV and CDV-PP concentrations in kidney were mea- sured at day 7, see Table 1. 2.2 | RESULTS AND DISCUSSION 2.2.1 | Validation results Selectivity for BCV and CDV-PP was evaluated using homogenates of six separate lots of kidney, brain and spleen tissue. No interfering peaks were detected at the retention times of the analytes or internal standards in any of the tissue lots evaluated. Representative chromatograms for matrix blanks, blanks with internal standard and a lower limit of quantitation (LLOQ) standard are shown in Figure 2 (for BCV) and Figure 3 (for CDV-PP). Linearity Three accuracy and precision runs were performed during the method validations for BCV and CDV-PP. All calibration curves employed linear regression with 1/x2 weighting and correlation coefficients exceeded 0.99. Calibration standards were included at the beginning and the end of each analytical run to bracket all injected QCs. Peak area ratios of the calibration standard to internal standard were used to construct the calibration curves. All runs passed according to our predefined acceptance criteria for calibration standards of accuracy ≤±15% (≤±20% at the LLOQ). Accuracy and precision Accuracy and precision were evaluated by analysis of six replicates of mouse kidney tissue homogenate QC samples of BCV prepared at the LLOQ (1.00 ng/ml) and at three additional concentrations (3.00, 30.0 and 800 ng/ml) over three runs. Similarly, six replicates of mouse kidney tissue homogenate QC samples of CDV-PP were prepared at the following concentrations: 0.0500 (LLOQ), 0.150, 2.00 and 40.0 ng/ml. Intra-assay statistics were determined from QCs (n = 6) within the same analytical run. Inter-assay statistics were determined from replicate analysis of QC samples (n = 6) in each of the three separate analytical runs (n = 18). Precision was calculated as the coefficient of variation (CV) while accuracy was calculated as percentage bias from the nominal concentration. All results met the predefined acceptance criteria of ±15% (±20% at the LLOQ). Accuracy and precision results are shown in Table 2. FIG U R E 2 Example chromatograms of BCV and BCV-IS from a blank (A), blank with internal standard (B), and a LLOQ standard (C) FIG U R E 3 Example chromatograms of CDV-PP and CDV-PP-IS from a blank (A), blank with internal standard (B), and a LLOQ standard (C) Carryover/crosstalk No carryover was observed for any analyte or internal standard fol- lowing the injection of a blank sample immediately following the highest calibration standard. Crosstalk was also evaluated by extracting a high standard without internal standard being added. No peak was seen in the internal standard channel showing no positive bias in the internal standard response at higher analyte concentrations. Recovery, matrix effects Three experiments were performed to evaluate recovery and matrix effects. Peak area responses of the analytes and the internal standards extracted from mouse kidney, brain and spleen tissue homogenate samples at the low, mid and high QC levels (pre-extracted) were com- pared with blank extracts spiked with analyte and internal standard corresponding to 100% recovered levels (post-extracted) and neat solution samples (un-extracted) corresponding to 100% recovered levels. True recovery was calculated as the ratio of the mean peak area response of the compound in pre-extracted QCs to the mean peak area response in post-extracted QCs. Matrix effects (the pres- ence of ion signal suppression or enhancement) were calculated as the ratio of the mean peak area response of the compound in post- extracted QCs to mean peak area response of the compound in un- extracted solvent QCs. Therefore, a matrix effect value of 100% would indicate the lack of ion suppression or enhancement. The range of recoveries for BCV was 99.1–104.7% (kidney), 98.6–103.3% (brain) and 94.3–104.4% (spleen) for the three QC levels evaluated. Recovery of BCV-d6 was 94.5–97.0% (kidney), 94.7–100.4% (brain) and 92.0–98.1% (spleen). The matrix effects for BCV were in the ranges 98.7–106.4, 98.8–107.1 and 97.2–108.7%, for kidney, brain and spleen homogenates, respectively. Matrix effects for the BCV-d6 were 102.6–109.8, 99.2–109.6 and 96.2–107.9% for kidney, brain and spleen homogenates, respectively, across all three QC levels. These data indicated good recovery for both analytes as well as low matrix effects. Recovery ranges for CDV-PP were 92.8–104.1% (kidney), 100.8–103.4% (brain) and 91.5–98.6% (spleen). For CDV-PP-IS, the ranges of recovery were 94.5–102.1% (kidney), 99.7–100.2% (brain) and 94.1–97.4% (spleen). The matrix effects for CDV were 81.2–83.9% (kidney), 82.3–87.7% (brain) and 65.3–65.7% (spleen). Similarly, for the CDV-PP-IS, the matrix effects were 78.8–82.7%, 81.4–88.3% and 65.3–65.7% for kidney, brain and spleen, respectively. Overall, good recovery and consistent matrix effects for the analyte and internal standard were observed in the various matrices. An additional evaluation of matrix effects was performed using the experiments previously described by Matuszewski (2006). Six dif- ferent lots of blank tissue homogenate (kidney, brain and spleen) were each spiked with BCV or CDV-PP at the low, mid and high QC levels and extracted in triplicate with the average peak area ratios from the analyses being plotted against nominal QC concentrations for each of the six lots evaluated. The slope values from these six curves were compared for both analytes. The values of CV from the six slope values for BCV were <3.2 and <1.8% for CDV-PP in all three matrices. As indicated by Matuszewski, these results indicate the lack of matrix effect related to different lots of tissue homogenate used in the extraction owing to the CV value being below that required (<3–4%) to infer that the matrix effect are minimal. Blank kidney, brain and spleen tissue homogenates from the six selectivity lots were spiked with BCV or CDV-PP at the mid QC level (n = 3) and quantified against the kidney homogenate calibration curve (kidney, brain and spleen) to determine if assay performance was consistent in different tissue types. Precision (CV) and accuracy (bias) were within the ±15% acceptance criteria, showing that the use of different types of tissue homogenate does not impact accurate quantitation of BCV or CDV-PP. These data are presented in Table 3. Dilutions Dilution QCs for BCV (5,000 ng/ml) and CDV-PP (200 ng/ml) were evaluated as part of the method validation. The dilution QC was diluted with blank kidney homogenate 10x prior to extraction. Both analytes met 15% precision and accuracy requirements, demonstrat- ing their ability to dilute samples with concentrations up to 5,000 ng/ml and 200 ng/ml for BCV and CDV-PP respectively, into the validated calibration range. Stability Stability of the analytes cannot be assessed directly from tissue owing to it not being possible to generate a QC at a known concentration in intact tissue. Tissue samples for BCV and CDV-PP analysis are frozen in liquid nitrogen immediately following collection. The samples are weighed quickly, while stored on dry ice, followed by an immediate addition of cold extraction solvent and homogenization. This process ensures that the tissues are never exposed to room temperature or freeze–thaw conditions prior to homogenization. As a result, stability evaluations are performed on the tissue homogenate for mouse kid- ney, brain and spleen. Two sets of QCs representing the low and high ends of the cali- bration range (3.00 and 800 ng/ml) for BCV were assayed in triplicate after being left out at room temperature for 23 h (kidney, spleen) and 18 h (brain) and another set going through three freeze–thaw cycles. The accuracy (bias) results of these QCs (low, high) from nominal were as follows: room temperature, 12.6, 8.4% (kidney), 10.1, 10.6% (brain) and 12.0, 7.8% (spleen); and freeze–thaw, 7.7, 10.0% (kidney), 7.8, 5.0% (brain) and 12.8, 7.5% (spleen). Tissue homogenate extracts from QC samples at the low (3.00 ng/ml) and high (800 ng/ml) concentra- tions were analyzed in triplicate following storage for 3 days in the autosampler at 15◦C. Comparisons of the resulting concentrations with the nominal levels for the low and high QCs respectively, bias values were 12.9 and −10.5% (kidney), 8.2 and 6.3% (brain) and 12.2 and 8.5% (spleen), meeting the accuracy acceptance criteria of ±15% bias. The stability of CDV-PP in mouse kidney, brain and spleen homogenate after approximately 21 h at room temperature was evaluated. Two sets of QC samples representing the low (0.150 ng/ml) and high ends (40.0 ng/ml) of the calibration range were assayed in triplicate after standing at room temperature for 21 h. Also, stability was evaluated after three freeze–thaw cycles between room temperature and −80◦C. Two sets of QC samples representing the low (0.150 ng/ml) and high ends (40.0 ng/ml) of the calibration range were assayed in triplicate each after completing three freeze–thaw cycles. The accuracy (bias) QC results (low, high) of treated vs. nominal for these evaluations were as follows: room temperature, 0.5, 5.8% (kidney), 2.5, 7.8% (brain) and 1.5, 9.4% (spleen); and three freeze–thaw cycles, 1.5, 8.9% (kidney), 2.1, 13.5% (brain) and 0.9, 6.4% (spleen). The back-calculated concentrations of the QCs for both evaluations were all within the ±15% acceptance criterion. Long-term stability of stock solutions was evaluated by compar- ing previous prepared solutions that were stored at −80◦C with solu- tions that were freshly prepared. The percentage differences from the stored stock were 0.6% (BCV) and −5.1% (CVD-PP), showing that BCV is stable for at least 164 days in 90:10:0.1% water–methanol–ammonium hydroxide and for at least 355 days for CDV-PP in water when stored at −80◦C. In addition, we evaluated the stability of the spiking solutions of BCV prepared in 70:30 methanol–water and spiking solutions of CDV-PP prepared in 1 mM ammonium phosphate. Fresh spiking solutions prepared at the low and high QC levels were compared with solutions that were stored at −80◦C for 48 days for BCV and 22 days for CDV-PP. The percentage differences between these solutions at the low and high QC levels were −9.4 and −9.75% for BCV and −4.7 and −5.8% for CDV-PP, respectively, demonstrating stability in their respective solvents when stored at −80◦C. 2.2.2 | Pharmacokinetic study results Mouse kidneys for all groups included in the study were collected at day 7. Six days after a single dose of 80 mg/kg, kidney concentration of BCV was undetected while CDV-PP concentration was 30 ng/g, approximately. BCV and CDV-PP levels were detectable for the regime of twice weekly dose at 40 mg/kg and daily dose for 6 days at 13 mg/kg. Results are shown in Figures 4 (BCV) and 5 (CDV-PP). FIG UR E 4 BCV concentrations in mouse kidney after a one-week treatment using different dosing regimen. Kidneys were collected at day 7 FIG U R E 5 CDV-PP concentrations in mouse kidney after a one- week treatment using different dosing regimen. Kidneys were collected at day 7 3 | CONCLUSIONS Two LC–MS/MS assays have been developed for the analysis of BCV and CDV-PP in mouse kidney, brain and spleen tissue. These methods were fully validated according to the FDA guidance for industry standards in mouse kidney homogenate with additional cross-validation experiments performed for mouse brain and spleen for BCV and CDV-PP. Both assays demonstrated linearity and sensitivity over a calibration range of 1.00–1,000 ng/ml and 0.050–50.0 ng/ml for BCV and CDV-PP, respectively, along with acceptable accuracy and precision. These methods have a short extraction time and can be used to evaluate BCV and CDV-PP in tissue exposures in animals to characterize pre-clinical pharmacology that will ultimately inform advancement into clinical studies.