Using Dried Blood Spot Sampling to Improve Data Quality and Reduce Animal Use in Mouse Pharmacokinetic Studies

Abstract

Traditional pharmacokinetic analysis in nonclinical studies is based on the concentration of a test compound in plasma and requires approximately 100 to 200 µL blood collected per time point. However, the total blood volume of mice limits the number of samples that can be collected from an individual animal—often to a single collection per mouse—thus necessitating dosing multiple mice to generate a pharmacokinetic profile in a sparse-sampling design. Compared with traditional methods, dried blood spot (DBS) analysis requires smaller volumes of blood (15 to 20 µL), thus supporting serial blood sampling and the generation of a complete pharmacokinetic profile from a single mouse. Here we compare plasma-derived data with DBS-derived data, explain how to adopt DBS sampling to support discovery mouse studies, and describe how to generate pharmacokinetic and pharmacodynamic data from a single mouse. Executing novel study designs that use DBS enhances the ability to identify and streamline better drug candidates during drug discovery. Implementing DBS sampling can reduce the number of mice needed in a drug discovery program. In addition, the simplicity of DBS sampling and the smaller numbers of mice needed translate to decreased study costs. Overall, DBS sampling is consistent with 3Rs principles by achieving reductions in the number of animals used, decreased restraint-associated stress, improved data quality, direct comparison of interanimal variability, and the generation of multiple endpoints from a single study.

Understanding the pharmacokinetics (pharmacokinetic) of a drug or a new molecular entity is important throughout the process of drug discovery and development. Pharmacokinetic studies require the collection of a series of blood samples in sufficient volumes from both nonclinical studies (in mice, rats, rabbits, dogs, and others) and clinical trials. Plasma from these samples are then analyzed to quantify the circulating concentrations of a drug or new molecular entity, which helps elucidate the exposure profiles (time compared with concentration relationship) and enable the computation of its pharmacokinetic properties. Collectively, this information is used to understand the disposition of the compound.

According to estimates from the Foundation for Biomedical Research and AALAS, more than 90% of all lab animals used in biomedical research are mice and rats.^1,13 Mice are used as the premier research model for understanding human diseases, predominantly due to their biologic similarities to humans, small size, short life spans, rapid reproductive rates, and relatively low maintenance costs. The increasing availability of numerous mouse models developed specifically for different disease types, ranging from diabetes to spinal cord injuries, knockout mice, humanized mice, and others, have significantly enhanced the discovery of novel treatment agents.

Historically the pharmaceutical industry has used plasma samples as the specimen for the quantification of circulating drug or metabolite concentrations for pharmacokinetic analysis. Bioanalytical assays used for the quantification of plasma samples often require volumes of plasma ranging between 50 and 100 µL, depending on the lower limit of quantification required and the sensitivity of the instrumentation. However, the collection of sufficient blood volumes from an individual mouse to generate the required volumes of plasma can be challenging, given that the circulating blood volume in a 25-g mouse is approximately 1.8 mL (compared with approximately 16 mL blood in a 250-g rat).¹⁰ Current guidelines and practices limit the maximal total volume of blood that can be collected from a 25-g mouse to approximately 10% to 15% of its circulating blood biweekly, that is, approximately 180 to 270 µL.^10,23,22

These guidelines and physiologic limitations necessitate the use of multiple rodents—especially mice—to generate a complete pharmacokinetic profile, that is, composite or sparse sampling. Therefore, the number of mice used during a pharmacokinetic study depends on the volumes of blood collected and needed for analysis. In our experience, typical pharmacokinetic or toxicokinetic studies can require an individual mouse for each collection time point and can result in the use of 25 mice or more per dose group or treatment, totaling more than 100 mice per study.

Current bioanalytical methods used for plasma analysis predominantly are based on liquid chromatography–tandem mass spectrometry (LC–MS-MS). The sensitivity of LC–MS-MS instrumentation has improved over the years, enabling the quantification of lower concentrations.³⁶ In addition, advances have been made in collecting and analyzing small volumes of plasma. Recent developments include the introduction of dried blood spot (DBS) sampling and DBS analyses to support drug discovery and development, thus enabling the quantification of the concentrations of a drug or new molecular entity from a single 10- to 20-µL volume of blood.^{7,26,31,33,35}

Although the application of DBS sampling to support drug discovery was demonstrated in the early 2000s,⁵ it was not until later in that decade that the technique gained broader implementation.^20,31 A majority of the applications have been for clinical use and supporting clinical trials,³⁷ most likely due to the significant advantages resulting from DBS sampling, including the ease of implementation (globally, across multiple sites, remote sites that may lack proper instrumentation and infrastructure to support plasma sampling and storage), use in neonatal and special populations,^32,38 and the significant savings in shipping costs.²

Different variations of DBS sampling,^19,39 microsampling of blood stored as a liquid,^4,16 and microsampling of plasma^6,17,25 have been reported. However, DBS sampling seems to be the most practical and widely implemented to date.^9,14

The feasibility of serial sampling in mice by using DBS sampling has been reported.^7,18,26,35 The data presented here demonstrate the adoption and advantages of DBS sampling in drug discovery for routine mouse studies, its effects on study design, and the resulting savings in animal use aligning with the 3Rs (reduction, refinement, replacement).^3,24,27,29 In addition, the ability to generate additional data, such as biomarkers or pharmacodynamics markers, from the same mouse in addition to the pharmacokinetic data is highlighted. Additional information regarding the bioanalytical applications and challenges of DBS sampling can be found in recent reviews.^14,20

Materials and Methods

CD1 mice were purchased from Harlan Laboratories (Indianapolis, IN), and test compounds were obtained from Lilly Research Laboratories (Indianapolis, IN). All procedures were in compliance with the Guide for the Care and Use of Laboratory Animals.¹⁵ All animal studies were reviewed and approved by the IACUC. Animal studies were conducted in an AAALAC-accredited program, and veterinary care was available to ensure appropriate animal care.

Mice were housed in groups of 3 per cage in standard open-topped polycarbonate shoebox-style cages with corncob bedding that were maintained on a 12:12-h light:dark cycle at a room temperature that ranged from 68 to 79 °F (20.0 to 26.1 °C). Enrichment was provided in the form of small cotton pads, which the mice tear apart and use to sleep on. A commercial diet (Harlan Teklad 2010, Harlan Labs, Indianapolis, IN) was provided as food. Cages were changed every 7 d.

Mouse plasma pharmacokinetic study.

Test compounds were administered as a single 10-mg/kg oral dose formulated in 1% hydroxyethlycellulose (Dow Corning, Midland, MI) (w/v), 0.25% polysorbate 80 (Sigma-Aldrich, St. Louis, MO) (v/v), and 0.05% antifoam (v/v; Dow Corning, Midland, MI) in purified water. Blood samples were taken at 8 time points from before dosing until 24-h after dosing by using 2 survival bleeds followed by a terminal cardiac bleed (3 samples per mouse) in a sparse-sampling design as depicted in Figure 1. Blood was collected into an EDTA-treated vacuum phlebotomy tube and centrifuged, and the plasma collected and stored (and shipped) frozen. For analysis, 50-µL aliquots of plasma from each time point were transferred to a 96-well plate, protein precipitated with 200 µL of 1:1 methanol:acetonitrile (containing internal standard), the supernatant was diluted with 1:1 methanol:water and analyzed by LC–MS-MS.

Figure 1.

Typical sparse-sampling mouse plasma study design (n = 3 per time point).

Mouse DBS study.

Test compounds were administered as a single 10-mg/kg oral dose formulated in 1% hydroxyethlycellulose (w/v), 0.25% polysorbate 80 (v/v), and 0.05% antifoam (v/v; (Dow Corning) in purified water. Serial blood samples were taken from each mouse until 24 h after dosing to yield a complete profile from each mouse (n = 3 mice). Blood was collected via a tail snip directly into a 20-µL EDTA-coated capillary and immediately spotted onto a Whatman DMPK-C DBS card (GE Healthcare Bio-Sciences, Piscataway, NJ). Tail snips were performed by removing approximately 1 mm of the tail by using a scalpel. Blood flow was initiated by gentle squeezing of the tail. Generally a fresh cut was not needed for subsequent bleeds within the same day, and blood collections were performed via the removal of the clot. A fresh cut (approximately 1 mm) was performed for the 24-h sample. Plastic tube restrainers were used during collections. No analgesia or anesthesia was used during blood collections. A single sample and spot was collected per time point. The DBS cards were allowed to dry for approximately 2 h at room temperature, after which the cards were placed in a zip-top bag, stored, and shipped at ambient temperature. For analysis, a single 3-mm disc was punched from each time point and extracted with 100 µL 1:1 methanol:acetonitrile (containing internal standard), and the supernatant was analyzed by LC–MS-MS (after dilution with 1:1 methanol:water).

LC–MS-MS analysis.

An HPLC system consisting of LC-10ADVP pumps and a SCL-10A pump controller (Shimadzu, Columbia, MD) was used in combination with an autosampler (model 215, Gilson, Middleton, WI). A Betasil C18 5-µm 20 × 2.1 mm Javelin HPLC column (Thermo Electron, West Palm Beach, FL) was used. Mobile-phase solvent A consisted of HPLC-grade bottled water:1 M NH₄HCO₃ (2000:10, v/v), and solvent B consisted of methanol:1 M NH₄HCO₃ (2000:10, v/v). Analytes were separated by using a linear gradient starting at 60% solvent B and increasing to 90% by 0.2 min, holding until 0.35 min, ramping to 98% at 0.36 min, and holding until 0.76 min. The HPLC column was held at ambient temperature, and a flow rate of 1.5 mL/min was used with an injection volume of 10 µL. Mass spectrometric data were generated by using an API 4000 triple-quadrupole mass spectrometer and acquired by using Analyst software, version 1.4 (Applied BioSystems, Foster City, CA). Selected reaction monitoring transitions were optimized for each compound via direct infusion and used for quantification. Acquisitions were performed at unit resolution by using positive-ion atmospheric pressure ionization at a source temperature of approximately 700 °C and an ionspray voltage of approximately 1500 V.

Pharmacokinetic calculations.

The AUC from time 0 until the last time point (AUC_0-tlast), maximal concentration (C_max), and the time corresponding to the C_max concentration (T_max) were calculated for each animal's plasma and DBS profiles using Watson bioanalytical LIMS (version 7.4) from Thermo Scientific (Billerica, MA).

Results

The number of mice required to generate a complete pharmacokinetic profile is determined by the volume of blood collected at a given time point, as shown in Figure 2. In the current study, the volume of plasma needed for analysis (as much as 50 μL) and the bioanalytical assay sensitivity needed to quantify the circulating concentrations limited pharmacokinetic sampling to 3 time points per mouse. The plasma and DBS pharmacokinetic profiles after the administration of a single oral dose of a test compound are shown in Figure 3, and the corresponding pharmacokinetic properties are summarized in Table 1. The plasma data are based on sampling of 9 mice and composite pharmacokinetic analysis, whereas the DBS data represent serial sampling of 3 mice. The blood:plasma ratio for this study, calculated according to the DBS AUC and plasma AUC, is approximately 1.2, indicating that the compound is associated with the cellular components of the blood.³⁴

Figure 2.

Number of mice needed to generate a pharmacokinetic profile representing 8 time points (based on a 25-g mouse).

Figure 3.

Time compared with concentration profiles after a single 10-mg/kg oral dose administered to mice. The solid circles and corresponding profile are the average from triplicate measurements. (A) Plasma concentration profile via sparse sampling (composite profile) using a total of 9 mice. (B) Dried blood spot (DBS) profile via serial sampling using 3 mice.

Table 1.

Comparison of plasma (sparse sampling) and DBS (serial sampling) pharmacokinetic parameters after a single 10-mg/kg oral dose to mice

Parameter	Units	Plasma	DBS
No. of mice		9	3
AUC(0-t)	ng × h / mL	4710	5700
AUC(Extrap)	ng × h / mL	4760	5943
Cmax	ng / mL	930	1080
T1/2	h	3.5	2.4
AUC/Dose	ng × h / mL / mg / kg	471	570

AUC_(0–t), AUC from time 0 to last quantifiable time point; AUC_(extrap), AUC extrapolated to infinity from time 0; C_max, maximal plasma concentra­tion; T_1/2, terminal elimination half-life.

In an attempt to gain efficiency, we standardized the discovery DBS studies such that each pharmacokinetic study consisted of 6 individual test compounds that were individually dosed to separate groups of mice (3 mice per compound, 18 mice per study). The DBS concentration data corresponding to the 6 test compounds were used to generate exposure profiles as shown in Figure 4. Pharmacokinetic analysis of the exposure data and the physical-chemical properties of the 6 compounds are summarized in Table 2. From a study design perspective, this novel approach translates to a 66% reduction in the number of mice per study compared with a plasma study design such as that depicted in Figure 2 (18 mice compared with 54 mice).

Figure 4.

Time compared with DBS concentration profiles after a single 10-mg/kg oral dose of 6 test compounds administered as a single study. Each compound was dosed individually to 3 mice (serial sampling), and the entire study used 18 mice. DBS concentrations for all 6 compounds were generated in a single LC–MS/MS analysis.

Table 2.

Pharmacokinetic data generated from DBS sampling and physical–chemical data corresponding to 6 test compounds dosed individually (n = 3 mice per compound)

Compound	Molecular weight	cLogP	Aqueous solubility (µM) at pH 7.4	AUC(Extrap) (ng × h / mL)	Cmax (ng/mL)	Tmax (h)	T 1/2
1	545	3.02	29	15025	1478	3	5.4
2	550	2.05	21	17925	2109	1.5	4.8
3	405	2.88	>100	702	827	0.4	0.4
4	481	4.44	11	8218	775	1	12.9
5	449	2.92	>100	126	144	0.4	0.6
6	458	4.22	>100	293	27	0.75	9.1

AUC_(extrap), AUC extrapolated to infinity from time 0; C_max, maximal plasma concentration; T_1/2, terminal elimination half-life; T_max, time to maximal plasma concentration.

In the current study, DBS sampling was also shown to allow measurement of both pharmacokinetics and pharmacodynamics data from the same mouse, resulting in less variability and more accurate depiction of the pharmacokinetic–pharmacodynamic relationship (Figure 5). In this case, DBS sampling allowed the collection of sufficient blood for both a complete pharmacokinetic profile (that is, time points at 0.5, 1, 2, 4, 8, 24 h) of the compound and an additional 5 μL of blood for the measurement of blood glucose concentration, a pharmacodymanic marker for this compound, at each time point.

Figure 5.

Time compared with DBS concentration profile after a single oral dose of a test compound administered to 4 mice via oral gavage (black circles). Mean blood glucose concentrations measured as a pharmacodynamic marker from the same 4 mice at the same time points are plotted against the right axis (open circles).

Discussion

The analytical sensitivity (lower limit of quantification) for a given analyte is a function of the volume of plasma available for analysis and the limitations of the analytical instrumentation. Over the past decade, advances in pharmaceutical research have resulted in the generation of novel compounds that are very potent and often tested at lower doses, resulting in smaller circulating concentrations. In parallel, major improvements in LC–MS-MS instrumentation have enabled the ability to quantify the desired concentration ranges by using smaller volumes of blood (approximately 100 µL blood drawn to get at least 25-µL plasma for analysis), resulting in the ability to obtain 3 or 4 blood samples from a single mouse. Historically, large blood volumes (greater than 200 µL) often were required, thus limiting the acceptable number of collections from a mouse to a single terminal bleed or a single survival bleed followed by a terminal bleed (depending on the age and weight of the animal and the corresponding total volume of blood). In contrast, DBS sampling uses approximately 20 µL of blood per time point and avoids the need to generate plasma, thereby enabling the collection of an entire pharmacokinetic profile from a single mouse.

The data and examples we present here demonstrate the utility and advantages of DBS sampling in mouse studies. DBS sampling has been and continues to be used to support numerous analyses, ranging from newborn screening to therapeutic drug monitoring and HIV detection and screening.^8,11,21,30 However, its implementation across the pharmaceutical industry has been limited, likely due to the high degree of regulation, conservatism within the industry, and the lack of bioanalytical methods or advantages for changing from plasma to blood.

Plasma has become the ‘gold standard’ for pharmacokinetic analysis, driven by the fact that the analytical techniques available previously (predominantly LC–UV) required a ‘clean’ extract for analysis, such that plasma was selected as the matrix (given that whole blood was considered complex or ‘dirty,’ difficult to aliquot or pipette, and difficult to extract). Plasma data and blood or DBS data are equally valid from a PK perspective, but blood or DBS data should not be considered as equivalent to plasma data since that relationship will be dependent on the blood:plasma partitioning ratio. Therefore, care should be taken when comparing DBS data with plasma data. Some pharmacokinetic parameters (that is, Cmax, AUC) are expected to differ between the plasma and DBS data, whereas other parameters (that is, Tmax, T_1/2) are expected to be similar, depending on extent of partitioning and association of the individual compound with blood cells.

A key concern with regard to DBS data has been the ‘hematocrit effect,’ which refers to a bias in the concentrations that arises due to a difference between the hematocrit value of the sample and the standard curve used for quantification. This phenomenon could be significant in clinical trials, which might show marked inter-individual variability in the study population due to different disease states, age, and even different ethnic backgrounds. This concern is minimized in nonclinical studies using purpose-bred animals, for which inter-animal variability is expected to be low relative to clinical populations.

Implementing serial bleeding to generate a complete pharmacokinetic profile from a single mouse generates significant savings in animal use and is consistent with 3Rs principles in achieving a direct reduction in number of animals used (reduce); less restraint-associated stress due to the collection of smaller volumes of blood, improved data quality, and less variability (refine); and the collection of multiple endpoints from a single study such as pharmacokinetic and pharmacodynamic data from same mouse (replace). In addition, serial blood sampling enables direct comparison of interanimal variability, which is not possible with composite sampling. The need to dose 3 mice compared with 8 or 24 mice (depending on the number of blood samples collected from each mouse, see Table 2) also results in a significant savings in the amount of compound needed for dosing, especially during early discovery, when availability can be very limited. Furthermore using markedly fewer mice per study translates to the use of fewer animal cages and less husbandry overall.

Because of the low sample volumes needed, DBS study animals may provide blood in excess of that needed for pharmacokinetics analyses; this excess sample can be used for the analysis of metabolites or biomarkers. A shortcoming of mouse DBS sampling, as we have used and described it here, is the fact that only a single 20-µL spot is collected per time point; this practice may limit any subsequent reanalysis for incurred sample reanalysis (required for Good Laboratory Practices studies), or evaluation of metabolites, and so forth. Compared with mice for plasma sampling (which typically are euthanized during blood sampling), those used in DBS studies usually are not euthanized and can be reused, if desired, after an appropriate washout period. The low blood volumes needed for DBS samples allow intravenous–oral crossover studies to be executed in a single mouse (rather than involving multiple animals; data not shown). Care should be taken to use alternative sites for DBS sampling (saphenous vein, mandibular vein) when intravenous doses are administered through the tail vein. An alternative study design for intravenous–oral crossover studies is to use cannulated mice (typically using jugular vein cannulas) for intravenous dosing followed by tail-snip DBS sampling. This intravenous phase can then be followed with oral dosing and tail snip sampling—after an appropriate washout period—thus generating pharmacokinetic parameters for both routes and oral bioavailability information from the same mouse.

The standardization of mouse discovery studies to evaluate a series of compounds (6 compounds in this study) within a single study (individually dosed) enables rapid data generation because all of the samples can be analyzed within a single bioanalytical run (using a cassette standard curve) followed by the issue of a single final report. Overall, implementing DBS sampling to support mouse pharmacokinetic studies can not only save several thousand mice but also, when combined with the ease and simplicity of DBS sampling (given the lack of a need for centrifugation and separation of plasma), reduce the number of technical and husbandry personnel needed per study, thus reducing study costs. Typically, teams engaged in the early discovery phases of drug development would be screening hundreds of molecules spanning diverse chemical structures, or ‘scaffolds.’ The ability to routinely execute similar study designs by using DBS and generate pharmacokinetic data for discovery compounds enhances the ability to correlate pharmacokinetic data (AUC, C_max, T_max, T_1/2, and so forth) with physical–chemical data, helps to identify and streamline drug candidates, and provides critical scaffold-related information for lead generation and target optimization.

In addition, large numbers of mouse studies are conducted to assess in vivo pharmacology where pharmacokinetic data are also needed. In many of these instances, special mouse models are used, including knockout mice, humanized mouse models, and athymic mice for oncology studies. These data indicate that prudent planning and well-designed studies can enhance the study outcomes, as demonstrated in Figure 5.

Additional considerations that should be weighed prior to using DBS sampling for a compound entering drug development include an understanding of its unbound fraction in plasma and blood, its blood cell affinity, and the anticipated hematocrit range.^12,28 Finally, several bioanalytical limitations and challenges related to DBS analysis, which are beyond the scope of the current article, need to be considered preemptively.¹⁴