Simultaneous quantification of dexamethasone and 6β-hydroxydexamethasone in rabbit plasma, aqueous and vitreous humor, and retina by UHPLC–MS/MS

Dexamethasone (DEX) is a synthetic glucocorticoid which is widely used as an anti-inflammatory and immunosuppressive agent [1–3]. DEX is metabolized in human liver microsomes to 6-hydroxy and side-chain cleaved metabolites, with 6β-hydroxydexamethasone (6OH-DEX) as the major metabolite [4–6]. A sustained-release DEX intravitreal implant, which contains a poly (D, L-lactide-co-glycolide) (PLGA) polymer matrix, is designed to continuously deliver DEX to the posterior segment of the eye over a 3-month period. By overcoming the limitations of topical and systemic administrations and the burden of repeated direct ocular injections of DEX, DEX-PLGA intravitreal implants offer promising solutions to the treatment of inflammatory conditions in a wide variety of ocular diseases such as retinal vein occlusion, posterior segment uveitis and diabetic macular edema [1,7–10]. To date, no generic DEX-PLGA intravitreal implants are available on the market largely due to the challenges associated with establishing bioequivalence of this product. The route of administration and the prolonged application duration warrant a better understanding of in vivo drug release (i.e., local absorption) and distribution (i.e., systemic exposure) to more comprehensively understand the in vivo performance of these complex drug products. A study was designed to investigate DEX local absorption and systemic exposure following intravitreal administration of DEX-PLGA implants to New Zealand white rabbits, which is the most commonly used animal model to study intravitreal pharmacokinetics [11]. Owing to the presence of DEX at low concentration levels in ocular matrices and the limited amounts of biological samples available, the analysis of DEX in different matrices of the eye requires a highly sensitive and selective bioanalytical method.

To date, numerous analytical methods have been reported for the quantitative determination of DEX in various biological matrices, such as plasma, serum, urine, saliva, ocular biofluids and different tissues of human and animals [2,5,12–21]. Immunoassays [12,22–25] have been commonly used to measure DEX in the last few decades. Nevertheless, immunoassays are susceptible to various interferences including interactions between endogenous steroids and exogenous drugs and cross-reactivity of antibodies, as well as other interferences [26–28]. In recent years, GC–MS [17,29] has reported the necessary sensitivity and selectivity for DEX but required a lengthy derivatization step. Alternately, HPLC coupled with different detection techniques, such as UV (HPLC–UV) or diode array detection (HPLC–DAD) [5,30–32], MS (LC–MS) [13,15,17] or tandem MS (LC–MS/MS) [1,2,14,16,21,33,34], has provided direct measurement of samples without a derivatization step. Of these techniques, LC–MS/MS has potential advantages due to the sensitivity and selectivity provided by the mass analyzer.

Two publications have hitherto been reported using GC–MS [6] or HPLC [5] for the identification and quantification of DEX and 6OH-DEX in urine. However, to the best of our knowledge, no LC–MS/MS method has yet been reported in the scientific literature for the simultaneous quantification of DEX and 6OH-DEX in ocular matrices. The current study aimed to develop and validate a selective and reproducible UHPLC–MS/MS method, with appropriate sensitivity for small sample volumes, for the simultaneous determination of DEX and 6OH-DEX in various biological matrices (plasma, aqueous and vitreous humor and retina) obtained from rabbits. The analytical procedure was fully validated according to the US FDA Bioanalytical Method Validation Guidance for industry [35]. The procedure features a simple sample processing procedure and a chromatography run time of 2.5 min. The method was applied to measure the concentrations of DEX and 6OH-DEX concurrently for a rabbit ocular pharmacokinetic pilot study to assist with the development of preclinical study designs for DEX intravitreal implant BE studies.

Experimental section

Chemicals & reagents

Certified reference standard DEX was purchased from the United States Pharmacopeia (MD, USA). 6OH-DEX and dexamethasone-d₅ (DEX-d₅) were obtained from Toronto Research Chemicals (ON, Canada). Optima LC–MS grade methanol, acetonitrile (ACN), water and formic acid were purchased from Fisher Scientific (NJ, USA). Phosphate-buffered saline was purchased from Quality Biological, Inc (MD, USA). Bovine serum albumin was purchased from Sigma Aldrich (PA, USA). New Zealand white rabbit retinas, vitreous humor and plasma were purchased from BioIVT (NY, USA).

Preparation of stock, calibration & quality control solutions

Stock solutions of DEX, 6OH-DEX and internal standard (IS) DEX-d₅ were prepared separately in methanol to obtain a concentration of 1 mg/ml for each. Working stock solutions containing 10 μg/ml of each analyte were prepared by diluting appropriate amounts of DEX and 6OH-DEX stock solutions with methanol. Two working stock solutions were prepared from separate weighing and used for making calibration standards and quality controls, respectively. A working stock solution of DEX-d₅ (1 μg/ml) was prepared by diluting the stock IS solution with methanol. All the solutions were stored at -80°C for further use. Ten working calibration solutions were prepared by serial volume to volume dilutions from the working stock solution to obtain concentrations of 100, 50, 20, 10, 5, 2, 1, 0.5, 0.2 and 0.1 ng/ml in ACN. Working quality control (QC) solutions at five levels, 100, 80, 10, 0.2 and 0.1 ng/ml, were prepared in the same manner. A working IS solution (1 ng/ml) was prepared by diluting the working stock solution of DEX-d₅ with ACN. The working solutions were stored at -20°C until analysis.

Tissue homogenization

Retina samples were stored in centrifuge tubes and maintained at -80°C. Before homogenization, each retina sample was thawed and weighed, and then transferred to a 2 ml tube prefilled with triple-pure high impact zirconium beads from Benchmark Scientific (NJ, USA). A Benchmark Scientific D2400 BeadBlaster microtube homogenizer was used for fast and efficient tissue homogenization at 4°C. Retina samples were homogenized for about 35 s with 0.5% bovine serum albumin in phosphate-buffered saline buffer (w/v) with a volume to tissue weight ratio of 1 ml per 50 mg tissue. The homogenized retina samples were stored at -80°C for further analysis.

Sample preparation

Three different blank matrices, plasma, vitreous humor and retina of pooled-gender New Zealand white rabbits, were purchased for use. Pooled plasma was used as a blank matrix for the bioanalysis of plasma standards and samples. Pooled vitreous humor was applied to both aqueous and vitreous humor samples since they are physiologically similar. Pooled retina was likewise used for retina samples after homogenization. All blank matrices were stored at -80°C until needed.

Rabbit plasma, aqueous and vitreous humor or homogenized retina samples and their corresponding blank matrices were thawed at room temperature and vortexed. To a 1.7 ml microcentrifuge tube, a 50 μl sample aliquot was added and mixed with 100 μl of 1 ng/ml working IS solution and 350 μl of ACN for deproteinization. Calibration standards and QCs were prepared by mixing 50 μl of working calibration or QC solutions with 100 μl of 1 ng/ml working IS solution, 50 μl of blank matrix sample, and 300 μl of ACN. The tubes were capped, vortexed vigorously for 30 s, and centrifuged at 4°C at 17,850 r.p.m. for 10 min. After centrifugation, 450 μl of supernatant was transferred to a clean 12 × 75 mm glass tube and evaporated to dryness in a SAVANT SPD121P SpeedVac concentrator (Thermo Fisher Scientific, WI, USA) for about 30 min. The residue was reconstituted with 150 μl of 10% ACN/H₂O (v/v) solution containing 0.2% formic acid, vortexed for 30 s, and then transferred to an autosampler vial. A 25 μl aliquot was injected into the LC–MS/MS system.

UHPLC–MS/MS instrumentation & conditions

A Sciex ExionLC UHPLC system and QTRAP 6500⁺ tandem mass spectrometer (Sciex, MA, USA) equipped with a TurboIonSpray source were employed to perform the analysis. The UHPLC system consisted of two ExionLC AD pumps, a degasser, an autosampler and a column oven. The chromatographic separation was performed on a Waters CORTECS UPLC C18 column (2.1 × 50 mm, 1.6 μm) with a Waters CORTECS C18 VanGuard precolumn (2.1 × 5 mm, 1.6 μm). The mobile phase consisted of A: 5% ACN/H₂O (v/v) with 0.2% formic acid and B: 98% ACN/H2O (v/v) with 0.2% formic acid. The analytes and IS were eluted from the column at a flow rate of 0.8 ml/min with a 2.5 min gradient program as follows: 0–0.50 min, 8% B; 0.50–1.50 min, 8–70% B; 1.50–1.55 min, 70–90% B; 1.55–1.95 min, 90% B; 1.95–2 min, 90–8% B; and then isocratic for 0.5 min at 8% B to re-equilibrate the column. A diverter valve was used to only allow elution into MS only during the period 1–1.8 min to reduce potential contamination to the ion source. The column temperature was maintained at 40°C and the autosampler was set to 4°C during the analysis procedure.

Quantitation by multiple reaction monitoring analysis was performed in the positive ESI mode. Transitions to be monitored and compound-dependent parameters are summarized in Table 1. MS parameters were optimized for DEX, 6OH-DEX and DEX-d₅ by infusion of individual 100 ng/ml standard solutions and a mixed solution in methanol at 10 μl/min, respectively. The optimized parameters of the mass spectrometer were: curtain gas, 25 psi; ion spray voltage, 5000 V; source temperature, 500°C; nebulizer gas, 55 psi; turbo gas, 35 psi and collision activated dissociation gas, medium. Nitrogen served as curtain and collision gas. Data were acquired and processed by the Analyst 1.6.3 software package (AB Sciex).

Method validation

The bioanalytical method was successfully validated for selectivity, sensitivity, linearity, accuracy, precision, recovery, matrix effect, carryover, stability and dilution integrity in three matrices – rabbit plasma, vitreous humor or retina homogenate, respectively, in accordance with the FDA guidance [35].

The selectivity of the method was assessed by analyzing each of the three pooled blank matrix samples from six different batches, and by comparing the chromatogram of the blank sample with that of the corresponding calibration standard at the LLOQ level to investigate any matrix interference at the retention times of DEX, 6OH-DEX or the IS. Cross-reacting interference between the analytes and IS was also assessed by analyzing the high concentration standard solutions of DEX, 6OH-DEX and DEX-d₅, respectively.

Sensitivity was assessed by analyzing six replicates of LLOQ samples in three independent runs. The LLOQ was defined as the lowest concentration with a precision, expressed as %RSD, of ≤20% and accuracy of 80–120%, and demonstrated a S/N >10. Injection carryover tests were evaluated by injecting an extracted blank matrix sample immediately following the ULOQ sample.

For each matrix, three independent calibration curves were run over the analytical range of 0.1–100 ng/mL per analyte on three different days. The calibration curves were fitted by weighted linear regression (1/x²). Intra-day accuracy and precision were determined with six replicates at five QC levels – LLOQ (0.1 ng/ml), low (0.2 ng/ml), medium (10 ng/ml), high (80 ng/ml) and ULOQ (100 ng/ml). Inter-day accuracy and precision were evaluated for six replicates analyzed on three different days (n = 18).

Recovery and the matrix effect of DEX and 6OH-DEX were evaluated with six replicates at all five QC levels. Extraction recovery was calculated as the peak area ratio of an analyte in QC samples spiked before extraction to QC solutions spiked with the extracted blank samples. The matrix factor was obtained as the peak area ratio of an analyte in QC solutions spiked with the extracted blank sample compared with neat QC solutions in mobile phase. The recovery and matrix effect of DEX-d₅ was also evaluated in a similar way at a concentration of 1 ng/ml.

Stability studies were performed using QC samples at five levels under different storage conditions, studying six replicates at each level. Bench top stability (4 h at room temperature), postextraction stability in the autosampler (12 h at 4°C) and freeze–thaw stability (3 cycles, -80°C to room temperature) were determined. In addition, stability of the analytes in the 100 ng/ml working solution, which had been kept at 4°C for 1 week, was evaluated.

Dilution integrity was studied to ensure the validity of the dilution integrity test which was performed over the analytical range of DEX and 6OH-DEX during the analysis of humor samples. The dilution integrity tests were carried out by measuring the diluted samples following fivefold and tenfold dilutions of the spiked samples (500 and 1000 ng/ml, respectively) with the vitreous humor blank.

Animal study

The animal study was approved by the FDA Institutional Animal Care and Use Committee (IACUC) protocol (WO 2018-63) and performed in accordance with Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) international guidelines. The rabbit is the most commonly used animal in ophthalmic research studies to model the human eye due to the anatomical similarity of the rabbit and human eye. Furthermore, rabbit eyes are of adequate size for ocular tissue dissection and harvest. Therefore, rabbits were chosen for our study in generating DEX distribution data in ocular tissues as well as in plasma after a sustained-release DEX intravitreal implantation. An adult male New Zealand White rabbit, purchased from Charles River Labs and maintained in the animal facility at the FDA White Oak campus, was used for the pilot study. The rabbit was individually housed and maintained on a 12-h light cycle. A single 0.7 mg DEX intravitreal implant was implanted into the posterior segment of the right eye of the rabbit following the manufacturer’s instruction. Blood samples were collected from alternating marginal ear veins at predose and 8 h, 1, 2, 3, 4, 7, 11, 14, 18, 22 and 25 days after treatment. Aqueous humor samples were collected predose from the left eye (control) and on days 2, 9, 16, 23 and 25 postdose from the right eye. Vitreous humor and retina samples from both eyes were collected on day 25. Blood samples of approximately 500 μl were collected in EDTA tubes and centrifuged at 4000 r.p.m. for 20 min at 4°C. Plasma samples were transferred into labeled microcentrifuge tubes and stored at -80°C. All tissue samples were collected in preweighed centrifuge tubes and stored at -80°C in the dark until analysis.

Results & discussion

Method optimization

Several validated LC–MS/MS methods have been reported in the literature for the quantitative determination of DEX in ocular tissues and biofluids. Earla et al. [14] reported a bioanalytical method, with a LLOQ of 2.7 ng/ml for the quantitative measurement of DEX in different rabbit ocular matrices using LC–MS/MS. One hundred μl samples were prepared by protein precipitation followed by a liquid–liquid extraction. Recently, Matta et al. [16] demonstrated that DEX could be quantified in various rabbit ocular matrices with increased sensitivity (a LLOQ of 0.2 ng/ml) by applying a UPLC–MS/MS system equipped with an atmospheric pressure chemical ionization probe. A 50 μl sample aliquot was processed by protein precipitation. To simultaneously determine DEX and 6OH-DEX in rabbit biological matrices after a sustained-release DEX intravitreal implantation in this study, an analytical method with a LLOQ of 0.1 ng/ml was required due to the biological matrix concentrations of DEX and 6OH-DEX distributed in various rabbit ocular matrices and plasma.

To achieve this goal, a Sciex QTRAP 6500⁺ tandem mass spectrometer coupled with a TurboIonSpray source and a UHPLC system was employed to perform the analysis in positive ESI mode. The ESI source demonstrated higher sensitivity than the atmospheric pressure chemical ionization source with 0.2% formic acid included in both mobile phases. Compound-dependent and instrument-dependent parameters of the mass spectrometer were optimized for both analytes and IS as described previously. The most abundant and stable multiple reaction monitoring transitions were selected for the quantification of DEX, 6OH-DEX and DEX-d₅ (393.2→355.1 for DEX, 409.2→227 for 6OH-DEX and 398.2→360.1 for DEX-d₅). All three compounds were ionized typically at the conjugated 3-keto position to form the [M+H]⁺ ion in the positive mode in the same manner as other glucocorticoids described previously. Both DEX and DEX-d₅ shared the characteristic neutral losses of HF (20 Da) and H₂O (18 Da) as the primary fragmentation pathway. For 6OH-DEX, as previously illustrated by Matabosch et al. [36], the dominant product ion at m/z 227 plausibly corresponded to its steroid skeleton with only the A, B and C rings left after a cleavage of the D ring following the losses of hydrogen fluoride (HF) and H₂O.

Several columns were evaluated to determine the optimal chromatographic separation, appropriate retention times and acceptable peak tailing for all analytes. A Waters Cortecs C18 column was finally selected for further method optimization mainly based on better peak shape and a higher S/N which helps to improve sensitivity. The flow rate was set to 0.8 ml/min. ACN was used as the organic component of the mobile phase and the column temperature was set to 40°C to reduce the high column backpressure. The use of 0.2% formic acid in the mobile phase reduced the peak tailing while providing [H]⁺ for protonation. To eliminate carryover, the injection needle was rinsed before and after aspiration with 50% ACN/H₂O (v/v). The optimized chromatographic conditions allowed simultaneous analysis of 6OH-DEX, DEX and DEX-d_5. The observed retention times were 1.28, 1.62 and 1.62 min, respectively.

During method development multiple sample preparation techniques were compared, which include protein precipitation, supported liquid extraction and SPE. Protein precipitation was selected because of the adequate recovery obtained for both analytes and the simplicity of the procedure. However, sensitivity was compromised by using a large volume of precipitation reagent and a follow-up dilution step with aqueous solution to reduce solvent effects during LC separation. To meet the required sensitivity, a sample concentration step with a SpeedVac was included. The dried samples were reconstituted with the lower percentage organic mobile phase, which also improved the peak shape of each analyte. The signals from 6OH-DEX were significantly increased not only due to the concentrated sample, but also the enhanced solubility of 6OH-DEX in the ACN water solution used to reconstitute the sample. Moreover, the additional concentration step also allowed us to add a larger volume of precipitation reagent to further remove proteins from the matrix.

Method validation

Representative chromatograms of DEX, 6OH-DEX and DEX-d₅ in the blank and LLOQ (0.1 ng/ml) in plasma, vitreous humor and retina homogenate are shown in Figures 1–3, respectively. Figures 1–3 illustrate that no interference peaks from endogenous or other co-eluting compounds are observed at the expected retention times of DEX, 6OH-DEX and DEX-d₅. In addition, no cross-talk interference between the analytes and IS was observed, thereby verifying the specificity of the method.

The calibration curves were linear over the analytical range from 0.1 to 100 ng/ml for both DEX and 6-hydroxydexmethasone in plasma, vitreous humor and retina homogenate with correlation coefficients (r) greater than 0.99 (Table 2). The LLOQ was 0.1 ng/ml for both analytes in all three matrices.

The acceptance criteria for intra- and inter-day accuracy were 85–115% for all the QC levels, except for the LLOQ, where they were 80–120%. For precision, %RSD should be within 20% for LLOQ and 15% for the rest. Both accuracy and precision of DEX and 6OH-DEX from all three matrices were found to be acceptable at all QC concentrations (Table 3). Accuracy ranged between 94.1 and 105.7% for both analytes. The results for both intra- and inter-day %RSD were below 5.6, 6.1 and 10% for DEX at all QC levels in plasma, humor and retina homogenate, respectively. Similarly, for 6OH-DEX, the results in terms of %RSD were below 9.6, 7.3 and 8.7%, correspondingly.

The recoveries of analytes from all three matrices were evaluated at five QC levels in six replicates. The overall mean recoveries across QCs were 102.2, 96.5 and 101% for DEX and 100.7, 100.1 and 94.2% for 6OH-DEX in plasma, vitreous humor and retina homogenate, respectively. No meaningful matrix effect (≤8.6%) was observed, except for 6OH-DEX in vitreous humor, where there was 18.9% suppression. The results are presented in Table 4. The recovery for DEX-d₅ was 102.3, 96.8 and 93% in plasma, vitreous humor and retina homogenate, respectively. The matrix factor for DEX-d₅ was 97.4, 89.1 and 95.8%, respectively. The IS-normalized matrix effect for 6OH-DEX in vitreous humor was considerably improved by dividing the matrix factor of 6OH-DEX by the IS matrix factor. The use of deuterium labeled IS was able to minimize matrix related interferences, reduce ionization variability and improve method precision.

Both DEX and 6OH-DEX from all three matrices demonstrated acceptable benchtop (4 h at room temperature), autosampler (12 h at 4°C) and freeze–thaw (3 cycles) stability. The mean % nominal values of the analytes ranged from 93 to 113% for DEX and from 90 to 109% for 6OH-DEX at all five QC levels under all the tested conditions. The working solution of both analytes was found to be stable for up to 7 days at a storage temperature of 4°C. Blank samples injected immediately after the ULOQ samples did not show any significant carryover for both analytes and IS. Dilution integrity tests were performed to validate the fivefold and tenfold dilutions in vitreous humor. Accuracy of the measurements ranged between 93.3 and 109% and %RSD ≤5.74% for both analytes diluted from the spiked 500 and 1000 ng/ml humor samples.

Pilot study

The validated protein precipitation extraction-based UHPLC–MS/MS method was successfully implemented to simultaneously quantify DEX and 6OH-DEX in rabbit plasma, aqueous and vitreous humor or retina homogenate to evaluate the drug release from DEX intravitreal implant (0.7 mg) over a period of 25 days. Plasma concentrations from the pilot study samples were ≤0.24 ng/ml for DEX and below the LLOQ (0.1 ng/ml) for 6OH-DEX. The concentrations of DEX were between 21.6 and 318 ng/ml and those of 6OH-DEX were between 0.25 and 0.61 ng/ml in aqueous and vitreous humor. The plasma concentration of DEX was 0.1 ng/ml on day 11 postdose and continued to increase up to 0.24 ng/ml on day 25. The plasma concentration of 6OH-DEX was below the LLOQ during the 25-day period. However, results from aqueous humor samples showed a higher DEX concentration of 21.6 ng/ml on day 2 and continued increasing until day 23 (177 ng/ml). The one-time collection of vitreous humor on day 25 showed a higher concentration at 318 ng/ml. The concentrations of DEX and 6OH-DEX in retina collected on day 25 were 2120 and 2.26 ng/g, respectively. The results are presented in Table 5.

Conclusion

The present study reports the development and validation of a sensitive and specific UHPLC-MS/MS method for the simultaneous determination of DEX and 6OH-DEX in rabbit biological matrices (plasma and ocular tissues). Advantages include a chromatography run time of 2.5 min and the use of a small volume of sample (50 μl) with improved analytical sensitivity (0.1 ng/ml). The method was designed according to the current FDA bioanalytical guidance requirements as fit for purpose.

Future perspective

Considering the route of administration and duration of drug release, designing ocular pharmacokinetic studies for sustained-release dexamethasone intravitreal implant is very challenging. Appropriate bioanalytical methods developed for simultaneously quantifying DEX and 6OH-DEX with necessary performance characteristics will be helpful to further understand in vivo release, local and systemic distribution and metabolism of dexamethasone from DEX intravitreal implants.