Fully-Automated Approach for Online Dried Blood Spot Extraction and

Dec 23, 2013 - (19, 20) The online automated DBS analysis workflow consists of loading the sample cards into a DBS card carousel followed by online ex...
1 downloads 10 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Fully-Automated Approach for Online Dried Blood Spot Extraction and Bioanalysis by Two-Dimensional-Liquid Chromatography Coupled with High-Resolution Quadrupole Time-of-Flight Mass Spectrometry Regina V. Oliveira,†,§ Jack Henion,*,† and Enaksha Wickremsinhe‡ †

Quintiles Bioanalytical and ADME Laboratories, 19 Brown Rd., Ithaca, NY 14850, United States Eli Lilly and Company, Lilly Research Laboratories, Lilly Corporate Center Indianapolis, IN 46285, United States

‡

S Supporting Information *

ABSTRACT: An integrated automated approach has been developed for the direct determination of drug concentrations using a SCAP DBS system for online extraction and analysis of dried blood spots (DBS) from DBS paper cards to a multidimensional liquid chromatography system coupled to a high-resolution QTOF mass spectrometry (LC-HRMS). An accurate, precise, selective, and sensitive two-dimensional liquid chromatography-high-resolution mass spectrometry (2D LC-HRMS) assay was developed and validated using small volumes of rat blood (approximately 1.25 μL) from a 2 mm DBS punch. The methodology was validated according to internationally accepted regulated bioanalysis acceptance criteria in order to establish the validity of the combination of online DBS assay and use of HRMS for quantitative bioanalysis. The fully automated procedure exhibited acceptable linearity (r2 > 0.997) over the concentration range of 5 to 1000 ng/mL. Intra- and interday precision and accuracy runs indicated relative errors less than 20% at the LLOQ level and less than 15% at all other levels. The direct extraction and analysis of DBS samples resulted in a 5-fold improvement in assay sensitivity compared to conventional off-line extraction of punched DBS samples. In addition, the impact of blood hematocrit (Hct) on accurate quantification of the studied drugs also was evaluated, comparing Hct values of 30% and 60% against standards prepared at 45%. Hematocrit experiments show that Hct can influence the accuracy of drugs quantified by DBS and needs to be thoroughly evaluated prior to committing to validating a DBS assay. The online DBS system coupled to the LC-HRMS was then successfully applied to a pharmacokinetic study performed on male Sprague−Dawley rats after administration of a single dose of 5 and 10 mg/kg for midazolam and desipramine, respectively.

T

center of the DBS, transferring it to a tube or well(s) within a micro titer plate, and extracting the sample with solvent containing an internal standard. While the DBS procedure appears rather simple, often additional sample treatment is required such as liquid−liquid or solid-phase extraction to provide sample cleanup prior to analysis, which adds complexity and consumables cost to the work flow.17 While drug discovery and drug development groups appreciate the main advantages offered by DBS analysis, the punching step during the sample preparation is very tedious and time-consuming, especially when a large number of samples need to be prepared and impacts sample throughput. The use of automated systems for online DBS extraction and analysis represents a potential solution to overcome this barrier. With the use of the online approach, DBS samples can be

he dried blood spot (DBS) sample collection format is a well-established microsampling technique used for the screening of newborn errors of metabolism.1,2 Recently, it has gained considerable attention from the pharmaceutical industry to support pharmacokinetic and toxicokinetic studies in therapeutic drug monitoring and drug discovery as an alternative to historical and current plasma bioanalysis assays.3−9 Several reports3,10−14 have appeared, demonstrating advantages for the use of DBS over traditional plasma sampling and analysis that include a less invasive technique for blood sampling, reduced blood volume requirements, simplified way of shipping and storing dried blood samples, and reduced numbers of small animals required for pharmacokinetic (PK) and toxicokinetic (TK) studies.10−14 This approach is consistent with the goals of the 3R’s principle (reduction, refinement, and replacement). Moreover, DBS techniques offer improved stability for some unstable drugs and metabolites and lower contamination risks by pathogens.15,16 The current manual DBS work flow involves punching a disk from the © 2013 American Chemical Society

Received: November 5, 2013 Accepted: December 23, 2013 Published: December 23, 2013 1246

dx.doi.org/10.1021/ac403672u | Anal. Chem. 2014, 86, 1246−1253

Analytical Chemistry

Article

by Bioreclamation with a defined hematocrit level (30%, 45%, and 60%). All other reagents were of analytical grade. Whatman FTA DMPK-C cards and deep 2 mL 96-well plates were obtained from VWR International, LLC. Agilent Auto DBS cards were purchased from Agilent Technologies. Harris manual punches with diameters of 3.0 mm and the cutting mat were supplied by Ted Pella, Inc. The Matrix CapMats were from ThermoScientific. The Allegra X-12 centrifuge was supplied by Beckman Coulter, Inc. The Vortex-Genie 2 mixer was purchased from Scientific Industries, Inc. The repeater multipipette and the protein LoBind tubes (1.5 and 2.0 mL) were obtained from Eppendorf (Hamburg, Germany). The volumetric pipettes used for blood spotting and sample preparation were Pipet-Lite series from Rainin Instrument LLC. Automated System for DBS Extraction. Online extractions of the DBS cards were performed using a SCAP DBS system from Prolab GmbH (LEAP Technologies; Carrboro, NC).21,27 The DBS cards were robotically inserted into the LC mobile phase flow path and extracted online using the LC mobile phase as the DBS extraction solvent which passed directly through the DBS toward a trap column. A 2.0 mm diameter elution clamp was used for the DBS extraction. The stable isotope labeled internal standard (IS) (10 ng/mL for each analyte prepared in 10% of methanol in water) was injected online (20 μL) onto the clamped DBS cards via the flowing stream of elution mobile phase solvent. Before placing the DBS card into the clamp module and after the online DBS extraction, a digital picture was taken and electronically saved. This step enables card identification and also accurate location of each DBS position for sample extraction from the center of each spot. The pre- and postextraction pictures were routinely inspected as needed for correct extraction of the DBS cards. The SCAP DBS system was controlled with the SCAP DBS system software (Prolab Instruments GmbH) and the data acquisition and processing were performed with Agilent MassHunter (Agilent Technologies). Two-Dimensional Liquid Chromatography-High-Resolution Mass Spectrometry (2D LC-HRMS) Equipment and Analysis. The UHPLC system consisted of two analytical binary gradient pumps model LC 1290 Infinity equipped with degasser, a TCC 1290 Infinity column oven, and a six-port switching valve (Agilent Technologies). The trap column used was an Ascentis RP-Amide Supelguard Guard Cartridge (20 × 4 mm i.d.; 5 μm) from Supelco and purchased from Sigma Aldrich (St. Louise, MO). The analytical column was an XSelect CSH C18 (50 × 2.1 mm i.d.; 3.5 μm) from Waters. Both the trap and analytical columns were maintained at 35 °C using the column oven. Chromatographic separation was achieved using a solvent gradient program at a flow rate of 0.5 mL/min for both the trap and analytical columns. The binary mobile phase compositions consisted of 0.1% formic acid in water (mobile phase A) and methanol (mobile phase B). The gradient program used for the trap column was 0−2.00 min, 20% B; 2.01−2.20 min, 20 to 60% B; 2.21−2.50 min, 60% B; 2.51−3.70 min, 60 to 98% B; 3.71−5.00 min, 98% B; 5.01 min, 20% B. The gradient program used for the analytical column was 0−2.00 min, 20% B; 2.01− 2.20 min, 20 to 55% B; 2.21−2.50 min, 55% B; 2.51−3.70 min, 55 to 98% B; 3.71−5.00 min, 98% B; 5.01 min, 20% B. The total analysis time was 3.7 min per DBS sample. MS detection was performed using a 6540 QTOF highresolution mass spectrometer from Agilent Technologies. The

directly extracted and analyzed, eliminating the punching step. Additionally, some other advantages associated with online DBS extraction include notable environmental benefits due to reduced biohazards waste disposal,18 improved sample preparation efficiency, and higher sensitivity analyses, the latter being realized by minimizing analyte losses due to sample transfers, etc.19,20 The online automated DBS analysis workflow consists of loading the sample cards into a DBS card carousel followed by online extraction, liquid chromatography separation, and detection by mass spectrometry. Since high-resolution MS (HRMS) is a modern alternative approach for the quantification of drugs in bioanalysis due to its ability to increase the assay specificity and signal-to-noise by employing the use of accurate mass measurement and increased mass resolution, a QTOF (quadrupole time-of-flight) mass analyzer was employed in this work. Online DBS extraction and analysis employing the sample card and prep (SCAP) DBS system has been previously reported by Ganz et al.,21 who described a liquid chromatography−tandem mass spectrometry method for the online DBS extraction of bosentan and its three primary metabolites. This assay comprised of a three-column setup: two precolumns for successive online DBS sample cleanup and enrichment followed by an analytical column for chromatographic separation. DBS analysis also requires consideration of additional parameters for a successful outcome, such as spot size, blood volume, analyte stability in whole blood, on-card stability, extraction efficiency, and hematocrit effect among others. In accordance with the European Bioanalysis Forum (EBF), hematocrit is identified as the most important single parameter impacting the validity of the concentrations obtained by DBS methods.22,23 Since hematocrit (Hct) influences the viscosity of the blood, it affects the blood spot dispersion on the DBS paper card, influencing accuracy and precision of DBS analysis.13 The effect of Hct on the accurate quantitation of drugs by DBS is most apparent when a fixed diameter punch or elution area is taken, which is smaller than the area of the dried blood spot.24 Additionally, the effect of Hct level and punch location on assay bias during the measurement of compounds with varying physicochemical properties (pKa, logD7.4, and log P) has also been reported.25 In this report, we examine the effects of Hct at three levels (30%, 45%, and 60%) across two different DBS cards types (DMPK-C and Auto DBS) for two test drugs with known differences in blood cell association. The compounds studied were midazolam and desipramine with a blood-toplasma ratio of 0.5:1 and 1:1, respectively.26 Herein, we describe a novel online, fully automated DBS LCHRMS strategy that provides direct quantitative determination of two different drugs in rat DBS samples. Furthermore, we examined the effects of Hct on the quantitative determination of two different commercially available DBS cards types as well as on two test drugs with contrasting blood:plasma affinities.

■

EXPERIMENTAL SECTION Chemicals, Reagents, Materials, and Apparatus. Methanol, water, 0.1% formic acid in water were of HPLC grade and were obtained from Honeywell Burdick and Jackson. Midazolam, desipramine hydrochloride, [2H4]-midazolam maleate, and [2H3]-desipramine hydrochloride were supplied by Cerilliant. The structures of target compounds are shown in Figure S-1 of the Supporting Information. Control Sprague− Dawley rat whole blood (K3EDTA anticoagulant) was supplied 1247

dx.doi.org/10.1021/ac403672u | Anal. Chem. 2014, 86, 1246−1253

Analytical Chemistry

Article

level of 45%) was conducted according to the Good Laboratory Practices (GLP) guidelines of the Food and Drug Administration (FDA). 28 The parameters evaluated included selectivity, linearity, and sensitivity, intra- and interassay precision and accuracy, reproducibility of sample re-extraction from the same DBS spot, extraction efficiency, carryover, and stability assays. For a detailed account of the experimental procedure for the preparation of intermediate and working solutions, preparation of calibration standard and QC samples, and for the method validation, please see the accompanying Supporting Information. In Vivo Studies. Surgically modified (arterial catheter) male Sprague−Dawley rats (approximately 230−350 g at 8−12 weeks of age, from Harlan Laboratories, IN) were given a single dose at 5 (intravenous) and 10 (intraperitoneal) mg/kg for midazolam and desipramine, respectively, where n = 4 rats per dose using a standard suspension dosing vehicle. Midazolam and desipramine were separately dosed to individual animals. For the desipramine assay, blood samples were taken at 0.25, 0.5, 1, 2, 4, 8, and 24 h post dose and for midazolam assay, blood samples were taken at 0.08, 0.25, 0.5, 1, 2, 4, and 8 h postdose. At each time point, approximately 50−75 μL of blood were collected into EDTA tubes via the arterial catheter using a needle and syringe. Then, 20 μL aliquots were spotted onto DBS cards (DMPK-C and Auto DBS). The DBS cards were dried for at least 2 h at room temperature and placed in plastic bags containing desiccants. PK Calculations. The AUC (AUC 0‑tlast ), maximum concentration (Cmax), and the time corresponding to the Cmax concentration (Tmax) were calculated for each animal’s DBS profiles using Watson bioanalytical LIMS (v.7.4) from ThermoScientific.

6540 QTOF was equipped with a Dual Jet Stream ESI ion source, and the instrument was operated in the positive ion polarity mode using 4 GHz ADC (analog-to-digital) acquisition under the following conditions: gas temperature, 200 °C; drying gas, 12 L/min; nebulizer gas, 20 psi; capillary voltage, 2.5 kV; and fragmentor and skimmer were set at 150 and 45 V, respectively. The sheath gas temperature was maintained at 300 °C and sheath gas flow was held at 12 L/min. The nozzle voltage was set at 150 V. Nitrogen was used as drying, nebulizer, and sheath gas. For sample acquisition, the MS system was operated in the mass range from m/z 100 to m/z 400. The reference mass was a constant background ion observed at m/z 185.1145. The acquisition rate was set at 1.0 spectrum/sec, and resolution was 32000 (full-width at halfmaximum, fwhm) at m/z 622. The quadrupole was operated in the RF-only mode. For LC and MS instrument control, Agilent MassHunter (version B.05.00, build 5.0.50420) installed on a single PC was employed. For qualitative and quantitative data processing, Agilent MassHunter Qualitative Analysis (version B.05.00, build 5.0.519.13, service pack 1) and Agilent MassHunter Quantitative for TOF (version B.05.00, build 5.0.291.0) were used, respectively. Prior to the actual measurements, optimal ion source parameters were determined by continuous infusion of individual standard solution of midazolam or desipramine (500 ng/mL prepared in 50% methanol in water containing 0.1% formic acid) at a flow rate of 10 μL/min. For the first 2.6 min and from 4.0 to 5.0 min of the analytical run, the LC flow was diverted to waste via an automatic postcolumn switching valve. Sample concentrations were determined from the peak area ratios of respective exact mass for the analyte relative to the IS using a 10 ppm window for extraction, as follows: m/z of 326.0855 for midazolam; m/z 330.1106 for [2H4]-midazolam; m/z 267.1856 for desipramine, and m/z 270.2044 for [2H3]desipramine. DBS Off-Line Sample Punching and Extraction. For a detailed account of the experimental procedure for the DBS offline sample preparation please see the accompanying Supporting Information. Online DBS Extraction Procedure. The fully automated process for DBS extraction and LC-HRMS analysis made use of a multidimensional system composed of two coupled columns which included a trap column and an analytical reversed-phase column. This approach enabled fully automated online DBS extraction, sample cleanup, and chromatographic separation coupled with HRMS. After spotting 20 μL of fortified blood onto DBS cards (DMPK-C and Auto DBS) and allowing them to dry for at least 2 h at room temperature, a 2 mm diameter disk (approximately 1.25 μL of blood) was directly extracted by passing the elution solvent directly through the DBS and onto the trap column. The SCAP DBS system used a 10-port switching valve for IS injection into the flowing elution solvent stream and extraction of the DBS. A 6-port switching valve from the LC system was used to couple the trap column to the analytical column. With the use of the 10-port valve, the extraction of the DBS and subsequent loading of analytes and IS onto the trap column (Ascentis RP-Amide Supelguard Guard Cartridge; 20 × 4 mm i.d.; 5 μm) was carried out by flushing mobile phase through the loop and the DBS at a flow rate of 0.5 mL/min (Figure S-2 of the Supporting Information). Validation Procedures. A full validation procedure in Sprague−Dawley rat whole blood (K3EDTA anticoagulant, Hct

■

RESULTS AND DISCUSSION Online DBS Method Development. The SCAP DBS system used for DBS analysis provides a fully automated system for online DBS extraction, eliminating the need for manual punching and off-line extractions, which greatly simplifies the sample preparation aspects of DBS bioanalysis. The direct extraction of the target compounds from whole blood spots is accomplished by using the LC mobile phase for DBS flowthrough-extraction of the DBS samples and direct elution to the liquid chromatography system. Because the LC mobile phase is used for both DBS extraction and LC analysis, an optimal mobile phase is required to provide satisfactory DBS extraction efficiency, chromatographic peak shape, and retention time. Preliminary investigations for selection of the optimal analytical column demonstrated that midazolam and desipramine (Figure S-1 of the Supporting Information, pKa = 6.57 and 10.2, respectively) could be satisfactorily analyzed on reversed-phase stationary phases. A XSelect CSH C18 column was selected due to its charged surface hybrid (CSH) technology, which helps to overcome problems related to peak shape asymmetry and poor loading for basic compounds when using acid mobile phases.29 For the duration of the method development, the mobile phase was composed of methanol and water, containing 0.1% formic acid in gradient elution mode and was evaluated for the online DBS extraction and further LC analysis. However, a significant sensitivity loss was observed in the course of less than 10 successive DBS extractions resulting from the coextraction of matrix components from the DBS samples. Typically, as higher percentages of organic solvent are used for DBS extraction, one obtains more selectivity for extracting 1248

dx.doi.org/10.1021/ac403672u | Anal. Chem. 2014, 86, 1246−1253

Analytical Chemistry

Article

position 2 (position 2-2), redirecting the effluent mobile phase to the analytical column. Midazolam and desipramine were transferred from the amide to the XSelect CSH C18 column during the 2.7 to 3.2 min time window using a gradient elution of water containing 0.1% formic acid and MeOH at a flow rate of 0.5 mL/min (details in the Experimental Section). Next the 6-port valve was switched back to position 1 for reconditioning of the amide column (position 2-1). The clamp mode was maintained with the DBS card for 4.9 min (position 2-1), enabling the washing of the extraction clamps and connecting tubes via the trap column to minimize carryover associated with the DBS analysis of midazolam and desipramine under these chromatographic conditions. Meanwhile, the target compounds were analyzed on the C18 column, which was interfaced to the JetStream ion source of the Agilent 6540 QTOF mass spectrometer. The total analysis time for the 2D-LC-HRMS method was 3.7 min with no additional time for sample preparation and with a total cycle time of 5.0 min. The prolonged cycle time of 5.0 min was required to promote flushing of the extraction clamps and connecting tubes via the trap column with a higher composition of methanol (98% in water containing 0.1% formic acid). The online DBS extraction followed by 2D-LC-HRMS analysis provided a selective method based on a column switching technique. As a consequence of the method optimization, approximately 2000 DBS samples could be analyzed without analytical column backpressure increase or decrease in its chromatographic performance. However, the trap column was replaced after approximately 200 DBS extractions or when a decrease in its chromatographic performance was observed to preserve the second column (analytical). High-Resolution QTOF Mass Spectrometry. Over the past few years, the use of high-resolution TOF and Orbitrap mass spectrometers has been reported for quantitative assays along with the results regarding sensitivity, precision, and accuracy compared to those obtained with the gold standard triple quadrupole (TQ) mass spectrometers.33−36 Indeed, recently produced QTOF instruments offer adequate resolving power (>30000), accurate mass measurement, high sensitivity, wide dynamic range, and fast data acquisition rate, which provides compelling evidence for its application in assays where quantitative and qualitative data can be obtained within a single run.33,34,36 Moreover, high-resolution MS (HRMS) does not rely on collision-induced fragmentation and the data acquisition produce full MS data on all analytes over a selected mass range. Selectivity is obtained from a combination of accurate mass extracted ion chromatogram (XIC) with a narrow mass extraction window (MEW) and mass resolution. In this paper, a 6540 QTOF from Agilent Technologies was employed for the DBS analysis of midazolam and desipramine. Accurate mass extracted ion current profiles for the theoretical exact m/z of each analyte (m/z 326.0855 for midazolam; m/z 330.1106 for [2H4]-midazolam; m/z 267.1856 for desipramine, and m/z 270.2044 for [2H3]-desipramine) with a mass extraction window of ±10 ppm was used for quantification. The determination of the MEW is an important step for precise and accurate quantification of compounds using HRMS. The MEW should not be too narrow nor too large in order to avoid false-positive or false-negative as a result of mass accuracy shift.35 Preliminary experiments to identify the MEW were performed using an extraction window of ±5, 10, and 20 ppm around m/z theoretical of the studied compounds. Whereas the

targeted compounds from a DBS versus endogenous blood components resulting in reduced matrix effects.30 However, when higher percentages of methanol (up to 100%) were tested with or without acid or base as additives for the DBS extraction, the results demonstrated reduced extraction selectivity, indicating that this higher percentage of organic solvent was extracting additional matrix components from the DBS sample, causing a higher matrix suppression effect and, consequently, reduced MS response for the target analytes. In order to circumvent this issue, different chromatographic columns composed of C18, C8, and HILIC stationary phases were evaluated with different composition of mobile phases. However, satisfactory results were not obtained for online DBS extraction and or LC analysis. Therefore, for the determination of midazolam and desipramine using direct DBS extraction and data acquisition by HRMS, the data showed that it was not feasible to achieve a one-dimensional (1D) LC-HRMS method that could adequately control matrix effects, and further method development was required. The matrix effects observed are in agreement with the results of Heinig et al.31 who have demonstrated that during a DBS extraction of a hydrophilic compound using organic solvent, it caused break-through on the trap column used for analysis and significantly extracted more lipids than when water was used, causing a higher matrix suppression effect. To manage the poor selectivity for the 1D LC-HRMS method, a multidimensional 2D LC-HRMS system was employed. Multidimensional chromatography promotes higher chromatographic separation efficiency for a mixture of compounds present in complex matrices: sample enrichment for compounds present in low concentrations and more efficient online sample cleanup by reducing sample matrix and exogenous compounds.32 During the multidimensional method development, it is necessary to adjust parameters such as the mobile phase used in the first and second dimension, flow rates, and transfer time. Ideally, the analyte of interest is transferred from the trap column to the analytical column in the narrowest heart-cut window, avoiding the transfer of interfering endogenous compounds. In this study, an Ascentis RP-Amide Supelguard (20 × 4 mm i.d.; 5 μm) trapping column was introduced in the first dimension and operated using a heart-cutting approach where a narrow time slice of the first dimension is directed onto an analytical column in the second dimension. Amide columns provide improved peak shape and retention for bases and could also promote a different selectivity by using an alternate stationary phase for the multidimensional chromatography separation. The employed column-switching system used is illustrated in Figure S-2 (Supporting Information). Using a 10-port valve from the SCAP DBS system and a binary chromatographic system comprising of two pumps and one 6-port switching valve, the DBS card was first transferred by a card gripper to the clamp module and the loop filled with 20 μL of internal standard ([2H4]-midazolam and [2H3]-desipramine 10 ng/mL each) (valves position 1-1). After automatic switching of the 10-port valve, the extraction of the DBS and loading of analytes and internal standards onto the amide column (position 2-1) was carried out for 2.0 min using water containing 0.1% formic acid and MeOH (80:20 v/v) at 0.5 mL/min. After that, the gradient program started (details in the Experimental Section). The 6-port valve remained in position 1 for 2.7 min, directing the flow rate to waste. Then, the 6-port valve was switched to 1249

dx.doi.org/10.1021/ac403672u | Anal. Chem. 2014, 86, 1246−1253

Analytical Chemistry

Article

acquisition was performed in centroid and profile mode, m/z determination and XIC for quantification were done with centroid data. When the MEW was ±5 ppm it impacted the accuracy and precision obtained for the calibration curves of midazolam and desipramine due to a mass accuracy shift. Therefore, the distribution of the centroid data of the analyte was outside of the selected mass extraction window and the XIC showed sudden drops to zero of the LC peak (absence of signal) (Figure S-3 upper panel, Supporting Information). Whereas, when ±10 or 20 ppm were selected as MEW, the m/z distribution based on centroid data was extracted almost entirely inside the MEW resulting in precise and accurate results. Therefore, mass accuracy demonstrated to be less than 10 ppm and greater than 5 ppm. When ±6, 7, 8, and 9 ppm were checked as MEW, fewer chromatograms showed loss of data and overall better precision and accuracy was obtained when compared to those obtained for the MEW of 5 ppm. The MEW of ±10 ppm was selected for all the data processing (Figure S-3 lower panel, Supporting Information). Figure S-4 of the Supporting Information shows the measured mass accuracy of midazolam and desipramine from a single analytical batch, containing 90 samples. The mass shift observed for midazolam and desipramine was within ±0.4 mDa from the theoretical value of 326.0855 and 267.1856, respectively, and represented a mass accuracy of approximately 1.2−1.5 ppm, when a MEW of ±10 ppm was selected for the data processing. These data demonstrate that the mass accuracy remained consistent throughout the entire sample analysis and it is fully compatible with a MEW of ±10 ppm, utilizing a mass resolution of 32000. These results are in agreement with those reported by Zhang et al.36 when evaluating the quantitation of small molecules using high-resolution accurate mass spectrometry. Validation. For the method validation the 2D LC-QTOFHRMS assay was evaluated using two different DBS cards: DMPK-C and Auto DBS. Calibration plots of analyte/internal standard peak area ratio versus the nominal concentration of midazolam or desipramine in whole rat blood were constructed and a weighted 1/x quadratic regression applied to the data. This provided the most accurate and precise responses for the quantification of midazolam and desipramine over the concentration range of 5 to 1000 ng/mL in rat blood with mean r2 values higher than 0.997. For the calibration curves, precision and accuracy were obtained within limits of 15%, except for the lower limit of quantification (LLOQ) level. The LLOQ of midazolam and desipramine was 5 ng/mL for each analyte, and it was defined by the lowest concentration that gave a signal-to-noise ratio equal to or greater than 10 while exhibiting an accuracy and precision ≤ 20%. The LLOQ reported here is not as low compared to most quantitative assays performed with a tandem TQ instrument.37 However, it should be noted that the actual volume of sample being extracted when performing direct extraction of DBS is a small fraction (approximately 1.25 μL) of the actual volume of blood spotted onto the DBS cards (20 μL). Selectivity. Selectivity was evaluated by monitoring for possible interfering peaks at the same chromatographic retention times as the analytes and IS. Control blanks and zero blanks from six individual lots of rat blood extracted from DBS cards were evaluated and showed no interferences greater than 20% of that observed for the LLOQ, at the retention time of the analytes of interest nor the IS. Representative chromatograms of LLOQ (5 ng/mL) and IS (10 ng/mL, 20 μL loop injection) for desipramine are shown in Figure 1.

Figure 1. Representative chromatograms from DBS analysis of desipramine using DMPK-C cards. (A) LLOQ (5 ng/mL) and (B) [2H3]-desipramine (10 ng/mL; 20 μL loop injection).

Moreover, representative chromatograms of blank rat DBS samples and LLOQ (5 ng/mL) for midazolam using DMPK-C and Auto DBS card are shown in Figure S-5 of the Supporting Information. The results obtained for the intra- and interday accuracy and precision of the assay were within acceptance criteria for assay validations28 and were within the predefined 15% limits required. Moreover, no difference in the results for the quantitative analysis of desipramine and midazolam was observed between DMPK-C card and Auto DBS card where precise and accurate results were obtained (Table S1−S4 of the Supporting Information). Additionally, the matrix interlot accuracy and precision were also obtained within 15% variation for both concentration investigated. Reproducibility of Sample Re-Extraction from the Same DBS Spot. When incurred samples are collected and spotted onto DBS card, not all samples may form circular blood spots; also, the location of the dried blood spot may not be centered. Taking this into consideration, the reproducibility of sample re-extraction from the same DBS spot was evaluated by extracting/analyzing QC samples prepared at 15.0, 150, and 900 ng/mL (n = 1). Each concentration was extracted five times (center and off-center areas), using an extraction area of 2 mm, as shown in Figure 2. Comparison results of QC samples were obtained for midazolam and desipramine using the two studied DBS cards (DMPK-C and Auto DBS). The DMPK-C cards showed slightly better results than Auto DBS cards, which may be due to different dispersion of the blood on the paper card. For the analysis of midazolam, the difference between the center spots and peripheral spots ranged from −10.1% to 11.7%, when using DMPK-C cards and from −7.5% to 21.6% when using Auto DBS cards. For desipramine, the deviations 1250

dx.doi.org/10.1021/ac403672u | Anal. Chem. 2014, 86, 1246−1253

Analytical Chemistry

Article

temperature before spotting them onto DBS cards. For the long-term stability, the results indicated that midazolam and desipramine were stable in the DBS format for 50 days at room temperature when compared with those samples of the same concentration prepared and analyzed at the same day. Stock solutions of midazolam, desipramine, [2H4]-midazolam, and [2H3]-desipramine were stable at least for 3 months stored at −20 °C. Direct Extraction of DBS versus Manual Extraction− Assay Sensitivity. The sensitivity of the online analysis of DBS samples by direct extraction was compared with that of manually extracted samples. The 2D-LC and MS conditions were identical for both approaches. For the automated direct DBS extraction, a 2 mm spot area was used while for the manual extraction the area punched was of 3 mm. A single precision and accuracy determination (calibration curve and QC samples) was performed by using both DBS cards, where the automated method for DBS extraction was compared to a traditional method using punching for DBS extractions. The results provided a 5-fold improvement in the LLOQ (5 ng/mL) obtained from the direct DBS extraction compared to manual punching of the DBS cards (LLOQ = 25 ng/mL) for the selected compounds. Hematocrit Studies. Hematocrit has been shown to influence drug concentration measurements in DBS analyses, and recently the European Bioanalysis Forum (EBF) has reported that hematocrit is identified as the most important single parameter capable of invalidating the results obtained by DBS methods.22,23 In this work, the effect of Hct was examined at two extreme levels of 30% and 60%, and the results compared to those obtained with Hct level of 45% across two different DBS cards types (DMPK-C and Auto DBS). As previously demonstrated, accuracy and precision using the Hct of 45% were within 15% for both drugs and on both DBS cards studied. Precision and accuracy of the QCs prepared using Hct levels of 30% and 60% and analyzed over three analytical runs showed an influence of Hct on drug concentration and was more noticeable at Hct of 60% compared to Hct of 30%, where accuracy was not within 15%. When comparing the two different DBS cards, the accuracy and precision values obtained for QCs prepared using Hct of 30%, showed a higher number of values outside the acceptable range when using Auto DBS cards compared to DMPK-C cards. When Hct of 60% was used, both Auto DBS and DMPK-C cards showed comparable number of samples with values outside the acceptable range. Overall, the results indicate that Hct plays an important role in the quantification of drugs by DBS and needs to be evaluated prior to validating a quantitative assay (Tables S1−S4 of the Supporting Information). It is also extremely important to understand the range of Hct that may be observed during the actual preclinical or clinical studies and to tailor the Hct evaluation to cover the appropriate range. Typically, Hct is not an issue for preclinical studies (i.e., mice, rats, dogs, monkeys, etc.) since these animals are carefully selected and bred for laboratory use and are not expected to have significant differences in the Hct value between animals. In Vivo Studies. PK profiles obtained for midazolam and desipramine using DBS samples taken from male Sprague− Dawley rats after intravenous or intraperitoneal injection of a single dose at 5 and 10 mg/kg for midazolam and desipramine, respectively, are shown in Figure 3. The time versus concentration profiles show that Auto DBS cards and the Whatman DMPK-C cards were almost identical for both drugs.

Figure 2. Performance of the SCAB DBS automated online extraction system for spot recognition. (A) Multiple extractions from a single DBS spot and (B) precise spot recognition and optimal positioning of the elution area of an noncentered incurred DBS sample.

from the center spot to peripheral spots ranged from −13.1 to 12.1%, when using DMPK-C cards and from −14.0% to 8.0% when using Auto DBS cards. Extraction Efficiency. The average extraction efficiency of midazolam and desipramine from three concentration levels was found to be 80.6% and 80.9%, respectively, when using the DMPK-C card, and 66.2% and 65.9%, respectively, when using the Auto DBS cards. The extraction efficiency observed indicates analyte stability under the extraction conditions used and favorable extraction efficiencies indicating that there is no major loss of analyte during the online DBS extraction. Carryover. Carryover was observed after the DBS extraction of the concentration level of 900 and 1000 ng/mL, and to a lesser extent, at lower levels. To effectively reduce the carryover of these particularly sticky compounds to less than 20% of the LLOQ, three successive extractions of blank DBS cards were performed after each extraction at those concentrations. For the incurred sample analyses, five blank DBS cards were extracted after the extraction of four replicate DBS spots (representing n = 4 animals for a given time point) to ensure satisfactory reduction in carryover. The additional extractions of blank cards to minimize the carryover significantly impacted the throughput of the assay and extended the total time required to complete the analysis for a sample batch. Additional experiments to evaluate carryover were performed, and the chromatographic conditions were modified to include methanol and water containing 0.5% of formic acid and flow rate of 1.0 mL/min. Additionally, after 1 min of extraction, the clamp was opened and reclamped in a blank region of the DBS card (offset position) for flushing. It is relevant to note that this step is not feasible with the Auto DBS card due to its physical layout. This new approach eliminated the carryover problem, but the chromatography peak of midazolam was impacted by peak tailing. The new chromatographic conditions were not evaluated for extraction efficiency of the DBS nor validated; however, it shows that carryover reduction can be achieved by modification of the extraction solvent. On the other hand, any modification in the extraction solvent will directly impact the chromatography performance, since the mobile phase is used for extraction and analysis simultaneously. Stability Assays. The stability of midazolam and desipramine in rat blood was determined under conditions expected during typical DBS sample storage and preparation. Midazolam and desipramine were stable in wet blood for 4 h at room 1251

dx.doi.org/10.1021/ac403672u | Anal. Chem. 2014, 86, 1246−1253

Analytical Chemistry

Article

Overall, Whatman DMPK-C and the Auto DBS cards showed similar performance during method development, validation, and analysis of the incurred samples. The automated online methodology described here was successfully applied for the analysis of in vivo samples and could be readily adopted to support both preclinical and clinical studies. There is considerable merit for automation of DBS card analyses, especially when large numbers of DBS cards need to be analyzed, when supporting multiple drug discovery programs and/or large clinical trials.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address §

Chemistry Department, Federal University of São Carlos, São Carlos, SP, Brazil 13565-905. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors wish to thank Agilent Technologies and LEAP Technologies for generous consignment of their corresponding instrumentation used in this DBS applied research study. R.V.O. thanks the Federal University of São Carlos for the sabbatical leave granted. J.H. and R.V.O. thank the Eli Lilly and Company for financial support of this work.

Figure 3. Pharmacokinetics profile from incurred samples after midazolam and desipramine administration using DMPK-C and Auto DBS cards.

■

REFERENCES

(1) Guthrie, R.; Susi, A. Pediatrics 1963, 32, 338−343. (2) Mei, J. V.; Alexander, J. R.; Adam, B. W.; Hannon, W. H. J. Nutr. 2001, 131, 1631S−1636S. (3) Pandya, H. C.; Spooner, N.; Mulla, H. Bioanalysis 2011, 3, 779− 786. (4) Turpin, P. E.; Burnett, J. E. C.; Goodwin, L.; Foster, A.; Barfield, M. Bioanalysis 2010, 2, 1489−1499. (5) Wickremsinhe, E. R.; Abdul, B. G.; Huang, N. H.; Richard, J. W.; Hanes, J. L.; Ruterbories, K. J.; Perkins, E. J.; Chaudhary, A. K. Bioanalysis 2011, 3, 1635−1646. (6) Wickremsinhe, E. R.; Huang, N. H.; Abdul, B. G.; Knotts, K.; Ruterbories, K. J.; Manro, J. R. Bioanalysis 2013, 5, 159−170. (7) Beaudette, P.; Bateman, K. P. J. Chromatogr., B. 2004, 809, 153− 158. (8) Barfield, M.; Wheller, R. Anal. Chem. 2011, 83, 118−124. (9) Spooner, N.; Lad, R.; Barfield, M. Anal. Chem. 2009, 81, 1557− 1563. (10) Burnett, J. E. C. Bioanalysis 2011, 3, 1099−1107. (11) Demirev, P. A. Anal. Chem. 2013, 85, 779−789. (12) Henion, J.; Oliveira, R. V.; Chace, D. H. Bioanalysis 2013, 5, 1− 19. (13) Li, W. K.; Tse, F. L. S. Biomed. Chromatogr. 2010, 24, 49−65. (14) Spooner, N.; Ramakrishnan, Y.; Barfield, M.; Dewit, O.; Miller, S. Bioanalysis 2010, 2, 1515−1522. (15) Adam, B. W.; Chafin, D. L.; de Jesus, V. R. Clin. Biochem. 2013, 46, 1089−1092. (16) Liu, G. W.; Ji, Q. C.; Jemal, M.; Tymiak, A. A.; Arnold, M. E. Anal. Chem. 2011, 83, 9033−9038. (17) Liu, G.; Patrone, L.; Snapp, H. M.; Batog, A.; Valentine, J.; Cosma, G.; Tymiak, A.; Ji, Q. C.; Arnold, M. E. Bioanalysis 2010, 2, 1405−1414.

The resulting pharmacokinetic analysis also showed comparable values between the two different DBS cards as shown in Table S-5 of the Supporting Information.

■

CONCLUSIONS A fully automated 2D-LC-HRMS method for the quantitation of midazolam and desipramine in 20 μL of rat blood collected on two different commercially available DBS cards (approximately 1.25 μL from a 2 mm extraction area) was developed and validated. The results show the methodology to be accurate, precise, reproducible, and selective. The automated extraction of DBS cards minimized the loss of the target compounds during extraction, where the same solvent system was used for DBS extraction as well as LC separation. The online 2D-LC-HRMS methodology described here provided a lower LLOQ compared to manual punching and off-line processing of the DBS cards. The use of high-resolution QTOF mass spectrometry showed good selectivity, linearity, precision, and accuracy for quantitative analysis of desipramine and midazolam in DBS samples. Hematocrit experiments show that Hct can influence the accuracy of drugs quantified by DBS and needs to be thoroughly evaluated prior to committing to validating a DBS assay. One of the potential solutions is to use novel DBS card substrates that produce the same spot diameter irrespective of the Hct. An alternative approach may be elution of the whole DBS spot; however, this requires the volumetric dispensing of a fixed volume of blood on to the DBS cards. 1252

dx.doi.org/10.1021/ac403672u | Anal. Chem. 2014, 86, 1246−1253

Analytical Chemistry

Article

(18) Li, F. M.; Ploch, S. Bioanalysis 2012, 4, 1259−1261. (19) Abu-Rabie, P.; Spooner, N. Anal. Chem. 2009, 81, 10275− 10284. (20) Deglon, J.; Thomas, A.; Mangin, P.; Staub, C. Anal. Bioanal. Chem. 2012, 402, 2485−2498. (21) Ganz, N.; Singrasa, M.; Nicolas, L.; Gutierrez, M.; Dingemanse, J.; Dobelin, W.; Glinski, M. J. Chromatogr., B 2012, 885, 50−60. (22) Vries de, R.; Barfield, M.; van de Merbel, N.; Schmid, B.; Siethoff, C.; Ortiz, J.; Verheij, E.; van Baar, B.; Cobb, Z.; White, S.; Timmerman, P. Bioanalysis 2013, 5, 2147−2160. (23) Timmerman, P.; White, S.; Globig, S.; Luedtke, S.; Brunet, L.; Smeraglia, J. Bioanalysis 2011, 3, 1567−1575. (24) Denniff, P.; Spooner, N. Bioanalysis 2010, 2, 1385−1395. (25) O’Mara, M.; Hudson-Curtis, B.; Olson, K.; Yueh, Y.; Dunn, J.; Spooner, N. Bioanalysis 2011, 3, 2335−2347. (26) Obach, R. S. Drug Metab. Dispos. 1999, 27, 1350−1359. (27) Thompson, J. W.; Zhang, H. Y.; Smith, P.; Hillman, S.; Moseley, M. A.; Millington, D. S. Rapid Commun. Mass Spectrom. 2012, 26, 2548−2554. (28) FDA. Guidance For Industry-Bioanalytical Method Validation; UCM368107, 2013, pp 1−34. (29) Iraneta, P. C.; Wyndham, K. D.; McCabe, D. R.; Walter, T. H. Charged Surface Hybrid (CSH) Technology and Its Use in Liquid Chromatography. Waters White Paper; 720003929EN, 2011. (30) Deglon, J.; Thomas, A.; Cataldo, A.; Mangin, P.; Staub, C. J. Pharm. Biomed. Anal. 2009, 49, 1034−1039. (31) Heinig, K.; Wirz, T.; Bucheli, F.; Gajate-Perez, A. Bioanalysis 2011, 3, 421−437. (32) Dixon, S. P.; Pitfield, I. D.; Perrett, D. Biomed. Chromatogr. 2006, 20, 508−529. (33) Fung, E. N.; Xia, Y. Q.; Aubry, A. F.; Zeng, J. N.; Olah, T.; Jemal, M. J. Chromatogr., B 2011, 879, 2919−2927. (34) Hopfgartner, G.; Tonoli, D.; Varesio, E. Anal. Bioanal. Chem. 2012, 402, 2587−2596. (35) Rochat, B.; Kottelat, E.; McMullen, J. Bioanalysis 2012, 4, 2939− 2958. (36) Zhang, N. R.; Yu, S.; Tiller, P.; Yeh, S.; Mahan, E.; Emary, W. B. Rapid Commun. Mass Spectrom. 2009, 23, 1085−1094. (37) van Dongen, W. D.; Niessen, W. M. A. Bioanalysis 2012, 4, 2391−2399.

1253

dx.doi.org/10.1021/ac403672u | Anal. Chem. 2014, 86, 1246−1253