ARTICLE pubs.acs.org/ac
Approach To Evaluating Dried Blood Spot Sample Stability during Drying Process and Discovery of a Treated Card To Maintain Analyte Stability by Rapid On-Card pH Modification Guowen Liu,*,† Qin C. Ji,*,† Mohammed Jemal,‡ Adrienne A. Tymiak,‡ and Mark E. Arnold† †
Bioanalytical Sciences and ‡Bioanalytical and Discovery Analytical Sciences, Research & Development, Bristol-Myers Squibb Company, Route 206 and Province Line Road, Princeton, New Jersey 08543, United States ABSTRACT: Unstable drug candidates often lead to complexity for both sample collection and bioanalysis. Dried blood spot (DBS) technology is believed to be a viable solution to address this problem. However, it is currently a challenge to evaluate compound stability on DBS due to its solid format. The observed compound loss on a DBS card could be degradation and/or incomplete recovery. Therefore, a reliable bioanalytical method which can differentiate recovery loss from degradation is necessary for such stability evaluation. In this paper, the stability of an unstable drug candidate (KAI-9803) in human blood was evaluated using DBS. A reliable approach to evaluating analyte stability on DBS was developed with an appropriate timezero sample, a consistent DBS sample processing method, and a suitable positive control. Commercially available DBS cards were evaluated, and it was found that KAI-9803 degraded during the drying process. An in-house modified DBS card was developed and demonstrated to be able to stabilize KAI-9803 during the drying process by rapidly lowering the pH of the spotted blood sample. The storage stability of KAI-9803 in human blood on this new card has been established for at least 48 days at room temperature. This in-house modified DBS card could provide a generic approach for other compounds which require stabilization at a low pH.
D
ried blood spot (DBS) is an emerging microsampling technology for bioanalysis1 3 to support pharmacokinetic/toxikinetic (PK/TK) studies. It offers many advantages over conventional sample collection methods, such as small sample volume, easy sample collection, simple sample handling, and cheap sample shipment. It has also been reported that DBS can offer advantages in stabilizing unstable compounds.4 7 However, as a sample collection technique for PK/TK studies, absolute stability (zero degradation) through sample collection, transition, and storage has to be established. Unlike liquid samples, the stability of an analyte in a DBS format must be considered in two stages: the stability during the drying process and the storage stability after the sample is dry.4 It is straightforward to evaluate compound stability when all samples are completely dry. However, it is a challenge to evaluate the stability of an unstable compound during the drying process. There are several factors that could complicate the interpretation of experimental results. When extracting an analyte from a DBS disk, an internal standard (IS) is usually only included in the extraction solution, which does not track the process of solubilizing and eluting the analyte from the DBS card to the solution phase.2 Therefore, the completeness and consistency of the analyte eluting from the DBS card will affect the assay performance, as well as the stability evaluation. During a stability evaluation, the observed stability of an analyte will be a combination of the “analyte stability on the card” and the “elution r 2011 American Chemical Society
efficiency of the analyte from the card”. A reliable approach to evaluating the DBS stability during the drying process has to be established before the true DBS storage stability can be established. The approach should be able to differentiate the “recovery loss” from the “degradation loss”. To the best of our knowledge, no method has been reported to address such a critical issue. KAI-9803 is a δ-protein kinase C inhibitor that is being developed for the prevention of reperfusion injury following acute myocardial infarction.8,9 It is a compound composed of two peptides linked by a disulfide bond.9 It rapidly degrades in vivo and ex vivo. The current sample collection approach requires mixing equal volumes of chilled (2 8 °C) 7.5% trichloroacetic acid (TCA) with a fresh blood sample immediately after drawing (usually within 10 min), which is not only time sensitive, but also requires accurate pipetting of both the blood sample and the TCA solution, and handling ice, dry ice, and corrosive acid on site. Since the intended use for this drug is for critical care, the current sample collection procedures make it nearly impossible to collect appropriate PK samples in a worldwide, multisite clinical study. Therefore, a new approach to stabilize KAI-9803 with simple sample collection procedures is desired to facilitate such a study. Received: July 21, 2011 Accepted: October 13, 2011 Published: October 13, 2011 9033
dx.doi.org/10.1021/ac2023876 | Anal. Chem. 2011, 83, 9033–9038
Analytical Chemistry
ARTICLE
Figure 1. Chemical structure of KAI-9803 with stable isotope labeled positions shown for the SIL-IS. The molecular weight is 2880.3 Da for KAI-9803 and 2904.3 Da for the SIL-IS.
In this paper we describe a reliable approach to evaluating the stability of unstable compounds on a DBS card during the drying process. An unstable drug candidate (KAI-9803) was discovered to be degraded during the drying process on commercial DBS cards. An in-house modified DBS card was developed and demonstrated to be effective in stabilizing KAI-9803.
Table 1. Detailed Gradient and Switching Conditions for the Online Extraction LC MS/MS Method loading
elution
switch valve
time (min) % C % D time (min) % A % B time (min) position 0
100
0
0
100
0
0
A
0.8
100
0
0.6
70
30
0.5
B
0.9
0
100
1.0
50
50
1.2
A
Chemicals, Reagents, Materials, and Apparatus. HPLC
2.5
0
100
1.5
30
70
grade acetonitrile and methanol were purchased from J.T. Baker (Phillipsburg, NJ, USA). Formic acid (SupraPur grade) and trifluoroacetic acid (TFA) were purchased from EMD Chemicals (Gibbstown, NJ, USA). Trichloroacetic acid (TCA), citric acid (CA), and human serum albumin (97 99%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Deionized water was generated using a NANOpure Diamond ultrapure water system from Barnstead International (Dubuque, IA, USA). Human K2EDTA blood was purchased from Bioreclamation Inc. (Hicksville, NY, USA). Liquid blood was stored at 4 °C upon receipt and used within 1 week from the collection date. KAI-9803 and its stable isotope labeled internal standard (SIL-IS), [M+24]-KAI-9803, were obtained from Bristol-Myers Squibb Co. The chemical structures of KAI-9803 and [M+24]-KAI-9803 are shown in Figure 1. FTA DMPK-B and DMPK-C cards were purchased from Whatman, GE Healthcare Bio-Sciences Corp. (Piscataway, NJ, USA). The FastPrep-24 instrument and lysing matrix D were purchased from MP Biomedicals, Inc. (Solon, OH, USA). Vials used for sample preparation were 2.0 mL polypropylene vials from VWR International (Bridgewater, NJ, USA). LC MS/MS Equipment. Sample analyses were performed on a Sciex QTRAP 5500 mass spectrometer (Applied Biosystems/ MDS SCIEX, Concord, Ontario, Canada), which was directly coupled with an online extraction LC system. The LC system consisted of a Leap HTC-PAL autosampler (Carrboro, NC, USA), four Shimadzu LC-10AD VP pumps (Columbia, MD, USA), and a switch valve from Valco (Houston, TX, USA). The configuration of the online extraction LC MS/MS system was the same as shown in ref 10 reported by Huang et al. (Figure 2 in ref 10). A Waters BioSuite C18 column (PA-A, 3 μm, 2.1 50 mm) was used as the analytical column for separation. A Waters
2.6 3.0
100 100
0 0
1.6 2.5
0 0
100 100
2.6
100
0
3.0
100
0
’ EXPERIMENTAL DETAILS
Oasis HLB column (25 μm, 2.1 50 mm) was used as the loading column for online extraction. Methods and Experiments. LC MS/MS Method. An online extraction LC MS/MS system was used for the measurement of KAI-9803 and its SIL-IS. Mobile phases A and C were 0.1% formic acid (FA) in water with 10 mM ammonium formate, mobile phase B was 0.1% FA in acetonitrile, and mobile phase D was 0.1% TFA in acetonitrile. Mobile phases C and D were used for loading onto the column, and mobile phases A and B were used for the elution. The extraction column was inline with the analytical column at valve position B and decoupled from the analytical column at valve position A. The flow rate was set at 4.0 mL/min for loading and 0.5 mL/min for elution. The detailed conditions for gradient and switching are listed in Table 1. The mass spectrometer was operated in the positive ion electrospray mode. The probe temperature was set at 650 °C. The collision energy was set to 36 eV for both compounds. The quadruply positive charged parent ions of both KAI-9803 and [M+24]-KAI-9803 were selected for Q1. The selected reaction monitoring (SRM) monitored for KAI9803 was m/z 721 > 624, and that for [M+24]-KAI-9803 was m/z 727 > 630. The total run time was 3.0 min, and the retention times of the KAI-9803 and its IS were about 1.1 min. Sample Processing Methods for DBS Samples. During the initial evaluation, whole blood was spiked with KAI-9803 at two concentrations (1.0 and 10.0 μg/mL) at room temperature. 9034
dx.doi.org/10.1021/ac2023876 |Anal. Chem. 2011, 83, 9033–9038
Analytical Chemistry Immediately after spiking, DBS samples were prepared by applying 20 μL of blood sample to each spot on DMPK-C cards. The wet DBS samples were then placed under a nitrogen flow to dry for 30 min. Each sample was then placed in a zip locked plastic bag and stored at room temperature until extraction. After the DBS samples were stored at room temperature for 2, 20, and 140 h, the whole DBS spot was cut out and transferred into an extraction tube with 200 μL of IS working solution (5 μg/mL in water with 1% FA). A 200 μL volume of 7.5% TCA solution was added after the sample was sonicated for 5 min. The mixture was sonicated for another 5 min before centrifugation. A 20 μL volume of the supernatant was analyzed. A reference time zero (T0) sample was prepared by mixing 20 μL of liquid blood at each concentration right after preparation with 200 μL of IS working solution, and then 200 μL of 7.5% TCA solution. After centrifugation, 20 μL of the supernatant was analyzed together with the DBS samples. After the initial tests, the sample processing method was improved to address some of the concerns in the initial method. The improved method included a new time T0 to minimize its physical difference from the stability DBS samples. A freshly prepared blood sample (20 μL at 10 μg/mL) was spotted on a precut DBS disk (6 mm i.d.). The wet DBS disk was then immediately transferred to a homogenization tube preloaded with 400 μL of 7.5% TCA solution, 200 μL of IS working solution (5 μg/mL in water with 1.0% FA), and 600 μL of methanol. This sample was used as the reference T0. The DBS stability study samples were prepared the same way as T0 but were left on the bench for specific lengths of time before being transferred to a homogenization vial preloaded with the same solutions as for the T0. All the DBS samples were then homogenized together using the FastPrep-24. After homogenization, the sample vials were sonicated for 10 min and then centrifuged at 4000 rpm for 5 min. A portion (600 μL) of the supernatant was transferred to a collection vial and dried down under nitrogen flow, then reconstituted in 200 μL of reconstitution solution (10% methanol in water with 1% FA). A 50 μL volume of the final samples was injected onto an online extraction LC MS/MS system for analysis. To test this approach, positive control samples were prepared by applying 10 μL of blood sample (20 μg/mL KAI-9803) and then immediately adding 10 μL of 50% TCA to each DBS spot and drying down under nitrogen flow. A time course of 0, 10, 30, 60, and 120 min was evaluated. Experimental Details for Liquid Blood Stability, Degradation Mechanism Investigation, and DBS Card Treatments. To test the stability of KAI-9803 in human whole blood at room temperature, 10 μg/mL KAI-9803 was prepared in human whole blood (less than 1 week after collection) and tested over a time course of 0, 10, 20, 40, 60, 90, and 120 min at room temperature. Specifically, at each time point, 20 μL of blood samples was mixed with 100 μL of IS working solution (5 μg/mL [M+24]KAI-903 in water with 1.0% FA) and 100 μL of TCA solution (7.5% by weight). The mixture was then vortexed and centrifuged, and the supernatant was transferred into an injection plate. A 50 μL volume of the final solution was injected onto the online extraction LC MS system for analysis. To investigate the degradation mechanism for KAI-9803, 10 μg/mL solutions were prepared in either a 50 mg/mL human serum albumin solution or in a 1.0% formic acid solution. The stability of KAI-9803 in both solutions was evaluated following the same procedure as in human whole blood over a time course of 0, 35, 80, and 130 minutes.
ARTICLE
Figure 2. Stability of KAI-9803 in human whole blood at room temperature. The result of each time point is an average of three measurements. The calculated T1/2 is 46.5 min.
To prepare in-house treated DBS cards, DMPK-C cards were treated with 20 μL of a 7.5% TCA solution, a 50% TCA solution, or a 50% citric acid solution, and then dried down under nitrogen flow or on the bench at room temperature for at least 2 h. The dried treated cards were placed in zip locked plastic bags containing desiccant. These treated DMPK-C cards were then evaluated for their stabilization of KAI-9803.
’ RESULTS AND DISCUSSION Initial Evaluation of KAI-9803 Stability Using DBS. The instability of KAI-9803 in human blood was confirmed in an in vitro test. As shown in Figure 2, KAI-9803 degraded in human blood at room temperature with a half-life (T1/2) of 46.5 min, which makes it necessary to stabilize it quickly once the blood sample is drawn to enable accurate measurement of its in vivo concentrations. As mentioned earlier, to avoid the complicated sample collection procedures in a worldwide multisite clinical study under the critical care clinical settings, DBS technology was evaluated as the alternative sample collection method. The same as any other sample collection technique, absolute stability of KAI-9803 on a DBS card has to be demonstrated before it can be used. The initial test indicated significant compound loss during the first two hours. This experiment was repeated several times with variable results, although the trend was similar. These results suggested that significant variations existed in the method, which has to be improved for evaluating the true stability of KAI-9803 on DBS. The assay variation could be attributed to the analyte recovery variation during the DBS sample extraction step, which was not compensated by using an SIL-IS. Furthermore, the physical differences between the DBS stability samples (solid) and the reference T0 sample (liquid) could lead to significant recovery difference for KAI-9803, which could contribute the observed significant compound loss during the first two hours. Of course, degradation during the drying process (the first two hours) could be another cause for the observed compound loss. Taken all together, the observed loss of KAI-9803 during the first two hours could be degradation loss, recovery difference between T0 and stability DBS sample, or a combination of both. Reliable Approach To Evaluating Drug Stability during Drying Process. Based on the results from the above experiments, a method to differentiate inconsistent recovery from degradation is needed. Our strategy to address this issue was 9035
dx.doi.org/10.1021/ac2023876 |Anal. Chem. 2011, 83, 9033–9038
Analytical Chemistry
Figure 3. Measured relative stability of KAI-9803 on DBS under different extraction conditions (A) with a homogenization step and (B) without a homogenization step. The DBS samples were prepared by treating the wet blood spot with a TCA solution immediately after the blood sample was spotted on the DMPK-C card. The average result from the two measurements at 0 min with homogenization was used as 100%.
Figure 4. Stability of KAI-9803 on different types of DBS cards during the drying process. The result of each time point is an average of three measurements. (A) DMPK-B; (B) DMPK-C.
minimizing the physical differences between different samples, therefore eliminating inconsistent recovery. First, instead of using liquid blood sample as the reference T0, freshly prepared DBS sample was used as T0. Since the T1/2 of KAI-9803 in liquid human blood at room temperature is about 46.5 min, minimal degradation should occur within 1 min. Second, all DBS samples (wet, half dry, completely dry) were homogenized before extraction. Since all DBS samples were homogenized, they could be treated as physically equivalent. This approach should minimize the recovery differences across different samples. Another key element to evaluating this method is to have a good positive control. Since it was known that a TCA solution can be used to stabilize KAI-9803 in whole blood, blood spots were treated with a TCA solution right after preparation (when the DBS sample was still wet) and then dried. These TCA treated DBS samples were used as positive controls. A comparison of the extraction with or without homogenization of the DBS samples was carried out using the positive control samples described above. As shown in Figure 3, with a homogenization step, the data were consistent across all time
ARTICLE
Figure 5. Stability of KAI-9803 in human serum albumin solution (A) and in 1.0% FA aqueous solution (B).
points. Without the homogenization step, however, a large variation was observed. This confirmed the large variations observed in the initial tests came from the method in which no homogenization was done on the DBS disks before extraction. These results demonstrated that the homogenization step minimized the recovery differences between different types of samples as expected. Data from the positive control samples has also demonstrated that this new approach is eligible for evaluating analyte true stability on a DBS card during the drying process. This new approach was then used to reevaluate the stability of KAI-9803 on both DMPK-B (chemically treated) and DMPK-C cards (untreated). As shown in Figure 4, instability was observed for KAI-9803 on both types of DBS cards. Three measurements were made at each time point, and the data were consistent across all time points. These data clearly suggested that degradation occurred during the drying process. A solution is needed to stabilize KAI-9803 during the drying process before DBS can be used to support the coming worldwide clinical study. Discovery of a Citric Acid Treated DBS Card. The cause for the degradation of KAI-9803 during the drying process was briefly investigated. As shown in Figure 5, incubation with human serum albumin solution (containing free thiol) led to rapid degradation of KAI-9803, but not in the low pH aqueous solution (1% formic acid in water). As shown in Figure 1, there is one disulfide bond in KAI-9803. It is recognized11 that disulfide bonds can undergo thiol disulfide exchange at high pH when a free thiol group is available. It was also known that at low pH (e.g., p H
